Supervisor:

Dr Rosa Hoekstra

Department of Psychiatry, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Kimberley Goldsmith

Dr Merge Feyasa

Dr Nadine Seward

Project Outline:

Ethiopian and Kenyan children with developmental disabilities (including intellectual disability and autism) and their caregivers experience severe challenges; most families receive no formal support. The NIHR-funded SPARK project aims to improve support for these children and their caregivers. SPARK includes a large randomised controlled trial of the World Health Organization‘s Caregiver Skills Training programme. Projected to comprise data from around 540 caregiver-child dyads followed over a time span of 10 months, the SPARK trial will be one of the largest trials conducted in developmental disability research. 

Our previous work showed that quality of life of Ethiopian and Kenyan families with a child with a developmental disability is severely compromised (Borissov et al., 2022), and impacted by financial difficulties, stigma, and social exclusion (Tilahun et al., 2016; Tekola et al., 2020; Gona et al., 2016). For the first time, the longitudinal nature of the trial data will allow us to consider temporal causal mechanisms behind the association between poverty, stigma, lack of social support, and quality of life.

This PhD will focus on disentangling these causal mechanisms using cutting edge methods such as interventional effects applying a robust causal inference framework and machine learning. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Jonna Kuntsi

Social, Genetic & Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Professor Richard Dobson

Project Outline:

Background- Late adolescence and the transition to adulthood is a highly challenging and potentially critical period for young people with ADHD that can lay the foundations for diverging adulthood trajectories. Many of the conditions that frequently co-occur with ADHD emerge in adolescence and major life transitions lead to multiple new demands and changes in available support networks. This vulnerable phase coincides with the clinical transition from child and adolescent mental health care to adult services, which itself is a focus of major current clinical concern: most youth with ADHD do not successfully transfer to adult services, despite significant needs for ongoing treatment. Opportunities for intervention are currently not fully realised due to both the young people’s disengagement from clinical services and our limited understanding of real-world targets for more holistic interventions.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Sabine Landau

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Giouliana Kadra-Scalzo

Professor Robert Stewart 

Project Outline:

Background-  Routinely collected clinical data provide a powerful resource for evaluating interventions for real-world patient populations because they are large, naturalistic resources; however, internal validity is frequently impacted by confounding. Hernan and colleagues have proposed a trial emulation framework which applies causal inference concepts to mimic a target trial using observational data. In this context propensity scoring approaches are often used to handle multiple measured confounders.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Daniel Leightley

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisors:

Professor Josip Car

Dr Lei Lu

Project Outline:

Vision- This PhD project aims to create Remote Assessment and Symptom Tracking (RAST), a collaborative and co-designed platform specifically focused on the early detection of trauma-related mental health conditions. RAST will leverage real-time modelling of multimodal data streams to pinpoint the onset of symptoms. This innovative approach is aimed not only to identify early warning signs but also to prompt crisis and clinical care teams to the developing risks. Such timely action is crucial for enabling swift interventions, thereby significantly improving patient outcomes.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Daniel Leightley

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisors:

Dr Lei Lu

Professor Josip Car

Project Outline:

Vision- This PhD project aims to improve chronic disease management by harnessing AI algorithms and large language models to develop personalised treatment plans. The focus will be on the integration and analytics of patient data, encompassing patient profiles, lifestyle patterns, and environmental influences. By leveraging this data, the PhD seeks to develop predictive models that can accurately tailor treatment strategies to individual patient needs. The personalised treatment plans will be delivered to patients through a dedicated smartphone application, which will also serve as a platform to monitor and encourage patient adherence to the prescribed regimen. This project represents a step towards more adaptive, responsive, and effective management of chronic diseases in digital therapeutics. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Daniel Leightley

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisor:

Professor Josip Car

Project Outline:

Vision- To develop a framework for the assessment of digital health technologies used in a health care setting, addressing the current limitations of the Digital Technology Assessment Criteria (DTAC) and similar frameworks. This project aims to establish a more comprehensive, agile, and holistic set of standards that not only address the rapid advancements in digital health but also anticipate future challenges and opportunities (notably around AI). The project will use diverse stakeholder perspectives, focus groups and surveys to create a framework that is robust, adaptable, and applicable to advance digital nations. But importantly understandable to those who commission the products, and end-users. The main ambition is to develop and validate a new benchmark in digital health technology assessment that ensures safety, efficiency, and innovation, ultimately leading to improved health outcomes and a stronger, more resilient healthcare system. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Janaka Karalliedde

British Hearth Foundation Centre, School of Cardiovascular and Metabolic Medicine and Sciences

Co-Supervisor:

Dr Salma Ayis

Project Outline:

Background - There is an urgent need to see people with diabetes who are likely to deteriorate whilst awaiting an appointment. 

However this requires strategies to increase capacity within the setting of restricted resources.  

Our aim is to develop and validate an algorithm that can identify and prioritise people with diabetes attending hospital clinics using electronic health records of more than 10,000 people with >15 years of longitudinal data. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Zunera Khan

Department of Old Age Psychiatry, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Dag Aarsland

Mr Miguel Vasconcelos Da Silva 

Project Outline:

Background - The UK is a multi-ethnic society, and the diversity of the population continues to increase. Around 3% of people living with dementia are from Black and Asian Minority Ethnic (BAME) communities, amounting to approximately 25,000 people. This number is likely to double by 2026 with the steepest increase expected in South Asian communities. Although major causes of ill-health are largely the same across all ethnic groups, there are some important differences such as altered risks for diabetes and heart disease and subtle differences in genetics that could affect brain health. BAME communities are less likely to be involved in research and consultations than white communities. Yet it is critical that researchers, health professionals and policy-makers understand their needs so they can address them. Digital research offers a way to reach out and engage people in research in new ways. The CARE (Community Ageing Research across Ethnicities) digital cohort could be used to increase involvement of individuals from ethnic communities in research and provide a valuable new means of improving diversity in brain health research. This study will build the reach and impact of CARE network by facilitating the development and evaluating the feasibility of the CARE App to improve wellbeing of older adults from diverse ethnic back grounds.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Jorge Cardoso

Department of Biomedical Engineering, School of Biomedical Engineering & Imaging Sciences

Co-Supervisors:

Dr Marc Modat

Professor James Teo

Project Outline:

Background - In computer vision, the rapid progress of deep learning methods was underpinned by very large datasets with millions of examples. Current medical imaging datasets pale in comparison, with the largest ever available dataset being only 40 thousand samples big. This limited data, combined with the fact that medical images are 3D means that downstream models cannot capture the full anatomical, pathological and signal variability. This limits the models that can be trained, their accuracy, and make them biased and unfair. The creation of an AI model that can generate synthetic images from any organ, modality and key pathologies would transform the field.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Yee Mah

Department of Biomedical Engineering, School of Biomedical Engineering & Imaging Sciences

Co-Supervisor:

Dr Jorge Cardoso

Project Outline:

Background - Characterising the motor behaviour of patients and staff is an important part of both clinical assessment and associated investigation and treatment. Advances in edge computing and computer vision, have enabled us to create a small edge device (MoCat) capable of reducing human motion down to their constituent major joints and installed a network of them into the Stroke Unit at King’s College Hospital.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Lei Lu

Department of Biomedical Engineering, School of Biomedical Engineering & Imaging Sciences

Co-Supervisors:

Dr. Daniel Leightley

Professor Josip Car

Project Outline:

Background - Diabetes stands as one of the most prevalent chronic diseases, impacting over 6% of the adult population in Western society, with its prevalence is dramatically increasing on a global scale. Managing diabetes requires ongoing monitoring and timely intervention to control its progression and alleviate associated complications. Improper management of diabetes can lead to severe health issues, such as heart disease, chronic kidney disease, nerve damage, and challenges related to vision, oral health, and mental well-being. Conventional diabetes management involves the manual measurement of blood glucose levels, insulin sensitivity, and other physiological parameters. However, these methods face limitations in providing personalised, real-time, and proactive care. Recent technological advancements, especially in deep learning, present new opportunities for enhancing diabetes management through the identification of digital biomarkers.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Lei Lu

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisors:

Dr. Daniel Leightley

Professor Josip Car

Project Outline:

Background - Cardiovascular diseases (CVDs) stand as the predominant global cause of mortality, accounting for approximately 32% of all deaths annually. Electrocardiogram (ECG) serves as an important diagnostic tool for CVDs, demanding meticulous interpretation for effective risk stratification and diagnosis. However, the interpretation of ECGs is impeded by time constraints and inconsistencies among physicians, resulting in misdiagnosis and potent health risks. Recent advances in deep learning, particularly in large language models (LLMs), offer a compelling avenue for reshaping healthcare applications, yet its full potential in enhancing cardiac diagnosis and CVD management remains largely unexplored.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Lei Lu

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisors:

Dr. Daniel Leightley

Professor Josip Car

Project Outline:

Background - Pregnancy is a transformative and delicate period in a woman’s life, marked by physical, emotional, and psychological changes. It is estimated approximately 10% of pregnant women and 13% of postpartum women experience depression, with higher rates in developing countries; If managed unproperly, these mental health challenges could have enduring effects on the well-being of both the mother and the infant, impacting birth outcomes and the long-term health of the family. Conventional methods for mental health management rely heavily on subjective assessments, facing the challenges of delayed intervention and providing tailored support. Recent breakthroughs in deep learning offer a paradigm shift in medical diagnosis and healthcare management. Harnessing the power of deep learning in the context of maternal mental health presents an unprecedented opportunity to revolutionise the way that we approach prenatal care.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Nicola Bailey

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Elena Seminati

Project Outline:

Background - A critical part of a prosthesis for people who have an amputation, is the socket. This is the interface between the prosthesis and the users residual limb. It is important that the socket is comfortable to wear and ensures that the prosthesis will remain securely attached to the user. If the socket has a poor fit, this can lead to discomfort and/or lack of usability resulting in user abandonment across all types of prostheses.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Wei Liu

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Professor Kate Tchanturia

Project Outline:

Background - Empirical research addressing cognitive processing in eating disorders has noted an overlap with autism. People with autism typically get distressed by situations that overwhelm them. This may be due to an overload of sensory stimulation: noise, visual stimuli overloaded environment.  Predictability, rules and routines help autistic people regulate overwhelming feelings. Emotional problems are common in people with eating disorders too. When they feel distressed, they often try to manage these feelings by over-exercising, restricting or bingeing on food. Research shows that almost one-third of people with eating disorders have autism as well. Recent work about adaptations of eating disorders treatments identified a gap in helping adults with eating disorders and autism regulate emotions. However, there are limited tools available to help them regulate their emotions and support their sensory difficulties.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Jie M. Zhang

Co-Supervisor:

Dr Martin Chapman

Project Outline:

Background - The emerging field of AI, particularly deep learning, has become a cornerstone in advancing clinical practices and healthcare operations. This research proposal introduces a groundbreaking AI task known as Multi-Modal Deep Learning in Obesity Healthcare. Targeting the pervasive health issue of obesity, prevalent in many developing countries, the project is designed to monitor, manage, and provide actionable suggestions for obesity care. The approach integrates critical body metrics, nutritional intake, and exercise habits into a cohesive model.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Angus Roberts

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Tao Wang

Professor Fiona Gaughran

Project Outline:

Background - Identifying eligible participants for a clinical trial requires matching a person’s detailed medical information with the complex criteria of the trial. Although electronic health records (EHR) have facilitated access to patient data, most information for accessing trial eligibility (e.g. family history) is recorded as free text and cannot easily be retrieved through a database query. Clinicians/researchers often need to manually examine clinical text to identify potential participants, which is time-consuming and expensive. 

Natural language processing (NLP) techniques have been used to identify a trial cohort from EHR text. However, existing systems often rely on document-based information extraction, without summarized patient-level information across multiple documents. Thus, complex trial criteria (e.g., “young females who are not trying to become pregnant”) cannot be queried seamlessly, which limits their usability. Also, without integrating multimodal data (both structured and unstructured information) to form a holistic and unified representation space for patients, these systems cannot provide insights into the quality of a cohort and the impacts of various trial criteria. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Letizia Gionfrida

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Angela Kedgley

Project Outline:

Background - Human movement dynamics play a crucial role in predicting outcomes such as injury risk and musculoskeletal disease progression. While significant strides have been made in lower-limb kinematics using markerless techniques, upper limb assessments remain challenging, particularly outside laboratory environments. This project aims to leverage and extend the capabilities of OpenCap, an open-source platform, for accurate and accessible markerless motion capture of upper limb kinematics and dynamics.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Ernest Kamavuako

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - Cardiovascular diseases (CVDs) are the leading cause of death worldwide, which accounts for approximately 30% of the mortality. Swallowing issues such as aspiration (swallowing down "the wrong way") affect 1 in 25 adults annually, whilst dehydration affects one in seven patients at an annual NHS cost of �1 billion, with 10% 30-day mortality higher than patients without dehydration. To address these problems, my laboratory, the BSSlab at KCL, is developing a wearable multimodal sensor capable of measuring multiple physiological parameters at the same time. These parameters are vital for the early detection and monitoring of cardiovascular and swallowing disorders. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Ernest Kamavuako

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Professor Rym M'Halla Ep Aounallah

Project Outline:

Background - Almost 3/4 of all deaths each year are due to non-communicable diseases, with 86% of these deaths primarily occurring in low- and middle-income countries. One of the reasons why there is a disproportionate effect of non-communicable diseases in lower-income countries, as found by Bollyky et al. (2017), is that primary care in most of these countries is focused on episodic care and is poorly situated to deliver access to affordable prevention, diagnosis, and treatment services that many non-communicable diseases require(2). Brain drain further complicates this issue, with Misau et al. (2010) noting that 11% of the global population lives in Sub-Saharan Africa. Yet, 25% of the global disease burden exists in this region, which only has 3% of the global health workforce. One way this issue is being tackled is with telemedicine, defined as the use of electronic information and communications technologies to provide and support healthcare when distance separates the participants.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Ahmad Al Khleifat

Department of Basic and Clinical Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Ammar Al-Chalabi

Dr Alfredo Iacoangeli

Project Outline:

Background - Current classification of neurodegenerative diseases based on clinical phenotypes does not take into account either underlying disease heterogeneity or overlapping disease mechanisms, thus hindering therapy development. Segregation and re-classification of neurodegeneration phenotypes based on genotype is one way forward. In this project we will reclassify existing phenotypic groupings using genetic analysis within and across complex neurodegenerations including Alzheimer’s Disease, Amyotrophic Lateral Sclerosis (ALS), and ALS-Frontotemporal Dementia (ALS-FTD).

There is a strong shared genetic component in these three neurodegenerations. For example VCP, ERRB4 and C9orf72 genes are known to harbour variants associated with these neurodegenerative diseases, yet not all those at genetic risk go on to become affected. The genetic aetiology of Alzheimer’s disease, ALS, and ALS-FTD is multifactorial, and a significant proportion of genetic variance remains unexplained. This hidden heritability may be harboured in structural genomic variation as well as rare variants that may be unique to an affected individual or family. Structural variants comprise different forms of genomic imbalance including copy number variants, insertions, deletions, inversions, duplications and inter-chromosomal translocations, as well as repeat sequences and repeat expansions.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Yijing Li

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Alexandru Dregan

Project Outline:

Background - This study received the data support from London Ambulance Services, and is designed to conduct a comprehensive analysis of healthcare service inequality from both macro and micro perspectives. Its primary objectives are to identify methods for enhancing service equity and efficiency and to provide practical reform recommendations.At the macro level, we will utilize open-source healthcare resource datasets to investigate spatial patterns of healthcare service inequality among different hospital trusts within England. Additionally, we will analyze demographic and socioeconomic factors to gain a comprehensive understanding of the underlying causes of healthcare inequality. Furthermore, this analysis will help identify specific groups or regions facing challenges, thereby laying the foundation for the development of targeted and effective strategies.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Sara Simblett

Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Professor Laura Goldstein

Project Outline:

Background - Project Overview:

The EpiCalm project will address the prevalent issue of anxiety in individuals with epilepsy developing and testing an innovative digital mental health intervention. This aligns seamlessly with DRIVE-Health's focus, particularly in: ‘Co-designing Impactful Patient-Centric Healthcare Solutions’, 'Next Generation Clinical Interface' and 'Multimodal Patient Data Streams.'

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Richard Dobson

Biostatistics and Health Informatics Department, Institute of Psychiatry, Pschology and Neuroscience

Co-Supervisors:

Professor Robert Stewart

Dr Tao Wang

Project Outline:

Background - People with severe mental illness (SMI) have a 10-20 year reduction in life expectancy than the general population, where a majority of these premature deaths are attributed to multimorbidity, i.e., co-existence of multiple chronic conditions (Hayes et al. 2017; Woodhead et al. 2014). To reduce this mortality gap, a vital step is to understand how combinations or clusters of illness affect this population. 

Existing evidence on multimorbidity mainly comes from cross-sectional studies that focus on examining the prevalence of a disease and groups of diseases (Firth et al. 2019; Reilly et al. 2015). However, little is known about how the current care system can be changed to improve health outcomes for people with SMI. This is largely because there is no large-scale longitudinal population study to better understand how multimorbidities and their treatments interact over time and how the interactions affect the quality of life among people with SMI. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Heba Sailem

Institute of Pharmaceutical Science, School of Cancer & Pharmaceutical Sciences

Project Outline:

Background - Breast cancer poses a significant challenge in the UK, especially for patients diagnosed at an advanced stage, leading to uncertainties in the best treatment strategies. The project aims to address this unmet need by developing personalized treatment strategies. Histopathology, the standard for cancer patient diagnosis, involves evaluating cellular and tissue changes. While artificial intelligence (AI) and machine learning have shown promise in analyzing large-scale cancer images, translating these models into clinical practice remains challenging. This project focuses on leveraging AI and deep neural networks to predict patient response to treatment by systematically scoring tumor composition and topology. The goal is to stratify patients for tailored treatments and explore alternative pathways, including participation in clinical trials.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Alan Hodgkinson

Department of Medical & Molecular Genetics, School of Basic & Medical Biosciences

Co-Supervisor:

Dr Alfredo Iacoangeli

Project Outline:

Background - Human mitochondria operate system-wide to regulate key biological processes. Consequently, mitochondria have been implicated in a wide range of common diseases, particularly those that occur in high energy tissues such as the brain. However, it is often not known whether altered mitochondrial function occurs because of changing cellular environments that are driven by disease states, or whether mitochondrial dysfunction forms part of the causal pathway of the disease itself. In this project we will identify whether tissue and cell type-specific mitochondrial transcriptional processes play a causal role in neurological disorders such as Parkinson's Disease and ALS. To do this, well will utilise large quantities of gene expression data from interconnected regions and cell types of human brains to computationally model the genetic mechanisms that contribute to variation in mitochondrial transcriptional processes using complex machine learning approaches, before integrating protein and metabolite data to validate downstream impacts on mitochondrial function. Validated models will then be applied to large-scale independent datasets (such as UK Biobank) to identify which mitochondrial processes are casually associated with disease. This is important to understand, as it will allow a better focus on the biological pathways that could be targeted therapeutically to reduce disease risk.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Yulan He

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Petr Slovak

Project Outline:

Background - Cognitive reappraisal (CR) is a technique in the psychological and emotional regulation area that involves changing one's interpretation of a situation in order to modify the emotional response it triggers. It is a commonly used technique in therapies like cognitive-behavioural therapy (CBT) to help individuals manage their emotions and reduce negative emotions such as stress, anxiety, anger or sadness, thereby improving overall mental well-being. Developing CR skills often requires regular access to therapy sessions, which poses issues about affordability and limited therapist availability. As such, digital technologies offer a viable alternative to support individuals in developing their CR skills. However, existing approaches exhibit limitations in terms of flexibility and adaptability when confronted with unseen situations. They also struggle to offer tailored responses to individuals. The recent development of Large Language Models (LLMs) presents an opportunity to build interactive agents designed to facilitate the learning of CR skills. Although LLMs can produce coherent and human-like languages, making them suitable for helping individuals develop CR skills faces the following challenges:

(1) Comprehensive understanding of the situations individuals present and the accurate identification of their initial appraisals and feelings requires the incorporation of commonsense knowledge and situational awareness. But LLMs were not trained for contextual understanding and situational awareness.

(2) The generation of tailored responses, such as proposing alternative thoughts, requires the selection of appropriate therapeutic strategies, which poses a challenge for LLMs.

(3) The evaluation of the outcomes of CR skill scaffolding also poses challenges since it requires continuous monitoring of emotional changes following the application of the most plausible alternative thoughts.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Kai Xu

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - INGESTIBLE electronics that can allow in-vivo monitoring of genomic and proteomic information within the gastrointestinal (GI) tract can shed light into personalized health through enabling new insights on the impact of complex bio-molecular interactions and their effects on macro-level accurate diagnosis. The project will involve multidisciplinary biomedical and engineering research and design practices under a very constrained capsule volume at millimetre-scale, namely, the miniaturized RF-antenna design for both power and data transfer, the ambient energy-harvesting (EH)-powered ultra-low-power (ULP) CMOS clock-free wireless telemetry System-on-Chip (SoC) design, and the system integration with multiplex biosensors. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Cathryn Lewis

Social, Genetic & Developmental Psychiatry Centre, Institute of Psychiatry, Pscyhology and Neurosceince

Co-Supervisors:

Dr Angus Roberts

Dr Tao Wang

Project Outline:

Background - The increasing burden of mental health issues on society and the NHS has led to a surge in antidepressant prescriptions. However, a significant challenge persists: a lack of understanding about the mechanisms behind antidepressant efficacy and why treatments fail. This PhD project aims to bridge this gap by leveraging electronic health records to assess treatment outcomes, paving the way for personalized antidepressant prescribing.

Antidepressant prescriptions in England increased threefold between 1998 and 2018 and account for 6% of all drugs dispensed. Only one third of patients respond to the first antidepressant prescribed, and many patients try multiple antidepressants before finding an effective therapy. Longterm antidepressant use is common, with the average duration of antidepressant prescription for depression reported to be 4.8 years. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Mariana Pinto da Costa

Department of Psychological Medicine, Institute of Psychiatry, Psychology and Neursoscience

Co-Supervisors:

Professor Fiona Gaughran

Professor Robert Stewart

Project Outline:

Background - The number of people with mental disorders has increased (McManus S, 2016), but access to treatment is still scarce, with studies suggesting that about one third of the people with mental health disorders receive appropriate treatment (Evans Lacko S et al, 2018). Remote consultations, using telephone and internet-based resources, are promising approaches to reduce the mental health treatment gap, especially in rural regions or in settings where mental health care is limited (Hoeft TJ et al, 2018). 

Tele-health is a core component of the UK mental health care in the NHS 5-year plan. A recent umbrella review of systematic reviews on the evidence-based guidance on tele mental health reported that, whilst the findings suggest that video-based communication could be as effective and acceptable as face-to-face contacts, the extent of digital exclusion and how it can be overcome was found to be lacking (Barnett P et al 2021). In England, a study of patients’ experiences with remote mental health services reported mixed results (Liberati E, et al, 2021). On one hand, patients valued the possibility of maintaining contact with their clinicians instead of scheduling appointments with new staff. On the other hand, they highlighted limitations in the therapeutic relationship due to the lack of non-verbal cues and technical training, and the exclusion of certain services that were not implementing remote care. Bonding is a key element of the therapeutic relationship, particularly important for people with mental health disorders. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Peter Hodkinson

Centre for Human & Applied Physiological Sciences, School of Basic & Medical Biosciences

Co-Supervisor:

Professor Yang Gao

Project Outline:

Background - Spaceflight provides a unique set of challenges for the human body and associated health care systems. Humans have had permanent presence in Low Earth Orbit on the International Space Station for over 20 years and in the next 10 years humans will return to the Moon as a test bed for the first human mission to Mars. Current medical operations are heavily dependent on Earth based resource and assistance, future missions are going to need to have what is termed ‘Earth Independent Medical Operations’. We are part of a team at the European Space Agency working with NASA and other international agencies to contribute to the medical planning for human missions to the Moon and Mars. This will be heavily dependent on terrestrial advances in data driven healthcare and application of novel approaches to human health, well-being and performance. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr David Watson

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - Biological systems are fundamentally causal, full of interdependencies that govern response to disease and treatment. Yet standard supervised learning algorithms – e.g., neural networks and random forests – are based on empirical risk minimisation, a strategy that explicitly prioritizes correlation over causation. The result is a black box model that may succeed at labelling new samples but fails to advance our understanding of the underlying process. This is unsatisfying in systems biology, where the goal is to map complex interactions between molecular phenomena. The implications for clinical practice are profound. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Sorin-Cristian Vladescu

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Professor Guy Carpenter

Project Outline:

Background - Meat and dairy alternatives are humanity's top lever to fight climate change, reducing greenhouse gases 11 times more efficiently per investment dollar than zero-emission cars. A major biological barrier to the widespread adoption of plant-based diets is astringency, a loss of oral lubrication while consuming plant-based proteins, leading to lower palatability.

Our current research into plant-based astringency reveals that particles within protein preparations can have a large (30%) effect in causing delubrication of the salivary film on the tongue, leading to the perception of dryness. Our previous work on carbonated beverages also revealed bubbles can affect taste and mouthfeel by affecting the salivary film on the tongue. In this project we believe the properties of the particles and their interaction with the salivary film on the tongue can be harnessed to create novel fat replacement additives. It is known that fat affects mouthfeel and is used in food products to reduce astringency, given the strong correlation between fat percentage and measured friction.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Emmanouil Spyrakos-Papastavridis

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - The most recent stroke burden estimates reveal that stroke remains the second leading cause of death and third leading cause of death and disability globally, with a total of 1.3 million stroke survivors in the UK alone, whose treatment incurs an aggregate cost of £25.6 billion annually. It has been revealed that improved mobility of stroke survivors can be achieved via rehabilitation exercise-induced neuroplasticity, which is traditionally delivered by specialised physiotherapists and clinicians, within rehabilitation centres. On the downside, in addition to the paucity of physiotherapists, the majority of stroke survivors face mobility challenges when travelling to and from rehabilitation centres. Hence, there is a need for the development of exoskeletons for use in residential settings; however, the wider adoption of such “domestic” exoskeletons is currently impeded by an array of technological limitations pertaining to both their mechanical designs and motion control algorithms.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Dafnis Batalle

Department of Forensic & Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Dr Luke Mason

Project Outline:

Background - Autism is a highly diverse and heterogeneous condition affecting ~1% of the population. It is associated with several co-occurring mental health problems and an increased risk of suicide and premature mortality relative to the neurotypical population. Histopathology and neuroimaging studies in children and adults with autism have shown subtle disruptions in the organisation of their neural systems. For instance, using a simple measure of functional connectivity (degree centrality), reproducible alterations in patients with autism have been reported in independent datasets. Brain network analysis techniques have also been used to show a reduction in global communication capacity in brain networks of three-year-old children with a diagnosis of autism. Despite the promise of precision neuroscience, progress on developing pharmacological treatments and effective behavioural support strategies has been slow. Major barriers include lack of objective, reproducible stratification biomarkers able to stratify patients into biologically and clinically meaningful subgroups. Using EEG, we have recently shown that neural responses to faces (N170 latency) was on average slower in autistic individuals compared with matched neurotypical participants, allowing the stratification of a subgroup of autistic individuals with slower responses and poor social prognosis. However, even in young children, we still cannot easily untangle the causes of autism from the secondary or compensatory effects of living with the condition, and we know little about the development of structural (SC) and functional connectivity (FC) networks supporting systems (such as social interaction), which are often atypical in participants with autism. Understanding trajectories of brain circuitry development and subtle alterations associated to neurodivergent development is paramount to accelerate the discovery of new therapeutic strategies for autism and co-occurring conditions.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisors:

Dr Ivana Rosenzweig

Professor Zoran Cvetkovic

School of Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Robert Leech

Professor K Ray Chaudhuri

Professor Peter Goadsby

Professor Steve Williams

Project Outline:

Background - It is envisaged that a PhD candidate with a strong engineering, physics or informatics background will join the highly multidisciplinary team at the Sleep and Brain Plasticity Centre, seeking to develop a system for objective tracking of key risk and progression biomarkers in patients with sleep disorders, such as the idiopathic/isolated REM behaviour Disorder (iRBD). This parasomnia is considered one of the most important predictors for α-synucleinopathies, including Parkinson’s disease and other dementias, such as dementia with Lewy bodies, or multiple system atrophy.  Some studies suggest that about 96% of patients with RBD will develop PD at some point in their lives. Given that Parkinson's disease is one of the fastest growing neurodegenerative diseases, it is especially important to identify individuals with prodromal PD in the general population. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Shan Luo

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - The proposed PhD project integrates health data engineering with human-robot interaction and advanced simulation technology. Focused on unraveling the complexities of grasping scenarios, the research aims to create sophisticated simulations that model the deformation dynamics of both objects and human skin during hand-object interactions. This project aligns closely with health data engineering principles, offering implications for applications in healthcare and assistive technologies.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

ProfessorAdam Hampshire

School of Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Steven Williams

Professor Richard Nicholas

Project Outline:

Background - The UK Multiple Sclerosis (MS) Register has been tracking thousands of patients for more than a decade, bringing together clinical record, survey and biomarker data that can provide unique insights into the progression of motor, mental health and cognitive symptoms in this heterogeneous disorder. To date, studies have primarily focused on applying relatively simple statistical methods to derive insights from these rich data. This project will develop and apply more advanced computational modelling methods that are capable of handling the type of very large, sparse and multivariate data that digital patient registers produce in order to better understand (a) population variability in the progression of MS, with a particular focus on the development of cognitive and motor deficits, and (b) how lifestyle, symptom targeted and disease modifying interventions affect that progression. The successful candidate will acquire skills in handling big data, causal modelling and brain imaging.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Adam Hampshire

School of Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Steven Williams

Dr Peter Hellyer

Project Outline:

Background - Cognitive assessments are a cornerstone of healthcare and have historically been conducted under supervised conditions by trained professionals. Advances in digital technologies have highlighted the potential for conducting cognitive assessments remotely via automated website and app-based software, opening the possibility of large-scale screening and longitudinal monitoring of clinical populations. A key outstanding challenge is that the performance of cognitive assessments, using either traditional supervised approaches or online, can be sensitive to cultural factors, for example, language and education level. Computerised assessments can also be sensitive to the devices that people are tested on. These sensitivities limit the accuracy of cognitive assessments when detecting cognitive problems, for example, when screening for the onset of dementia or after a stroke. The candidate undertaking this project will develop computational modelling methods that produce more precise and sensitive assessments that mitigate these limitations.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Michael Berthaume

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Nicola Bailey

Project Outline:

Background - Globally, an estimated 31.6 million people suffer from lower-limb loss (LLL; Cieza et al., 2020), with incidences of major amputation ranging from 3.6 to 600 per 100,000 (Moxey et al., 2011). Within the LLL community, mobility, enabled by functional prosthetics, is highly correlated to quality of life. Prosthetic feet, provided by prosthetic and orthotic (P&O) facilities within the NHS in the UK, are the only portion of the prosthetic that regularly interact with the external environment, making them particularly prone to fracture and mechanical wear. Oddly, in the UK, prosthetic feet are often replaced because they are out of warranty, and therefore the patient is at risk of the prosthetic foot breaking, and not because of any issues with the prosthetic foot. The feet are then donated to low-and middle-income countries (LMICs) where they are worn by others with LLL for years

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Charlotte Tye

Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Nicholas Cummins

Dr Michael Absoud

Project Outline:

Background - The aim of this project is to identify predictors of neurodevelopmental outcomes in early-onset epilepsy, by combining advances in machine learning techniques with linkage of multimodal and routine maternal and infant medical records and prospective measurement of young children with epilepsy over the first three years of life. This is an exciting opportunity for a prospective student with an interest in neurodevelopment to learn from a unique interdisciplinary supervisory team lying at the intersection between developmental psychology, neuroscience and health informatics, while interacting with the digital technology partner, vCreate.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Adelaide de Vecchi

Department of Biomedical Engineering, School of Biomedical Engineering & Imaging Sciences

Co-Supervisors:

Dr David Lloyd

Professor Pablo Lamata

Project Outline:

Background - Coarctation of the aorta (CoA), a narrowing of the aorta below the left subclavian artery, represents 7% of all congenital cardiac defects, making it one of the most common defects. In neonates, it is due to the abnormal closure of the ductus arteriosus and therefore occurs during the first two weeks of life. The aetiology of CoA is rooted in the interplay of anatomy and flow in the foetus and is still incompletely understood  [1]. Specific patterns of haemodynamic forces play a key role in vascular remodelling via mechanobiological action on the endothelial cells that line the vessel and have been linked to the closure defect [2]. Detecting CoA antenatally is vital as it enables timely treatment, lowering adverse outcomes. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Kim Goldsmith

Biostatistics and Health Informatics Department, Institute of Psychiatry, Pscyhology and Neuroscience

Co-Supervisor:

Dr Ewan Carr

Project Outline:

Background - To improve patient outcomes, we need to understand how interventions work and for whom. Such mechanistic questions can be addressed using mediation and moderation analysis. The identification of mediator and moderator variables is essential to translate scientific findings into effective interventions. Understanding mechanisms allows continuous, evidence-based refinement of interventions and theory, leading to rapid translation, personalisation, and, ultimately, improved patient outcomes. This is vital given recent meta-analyses showing that psychological interventions have moderate effect sizes that differ for different types of people 1,2. 

There is growing interest in tailored, personalised interventions that reflect individual circumstances, moving beyond the ‘one-size-fits-all’ approach of standard therapy3. Understanding more about how interventions work differently for different groups will help us best tailor interventions to the individual. However, despite this interest, personalised or stratified interventions have not yet delivered on their early promise. New methods are now needed to realise the potential of personalised treatment. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Oliver Canfell

Department of Nutritional Sciences, School of Life Course & Population Sciences

Co-Supervisors:

Dr Alastair Duncan

Dr Oli Williams

Project Outline:

Background - Most adults in the UK now live with overweight or obesity. Specialist weight management services are essential to improving health outcomes for adults living with obesity. This project will address two systemic inequities that remain pervasive in weight management services: 

(1) inequity of access in men and people of non-White heritage; and 

(2) inequity of health outcomes. 

This PhD project aims to leverage routinely collected ‘big data’ from electronic medical records to understand weight management service inequities across time, place, and person. This new intelligence will be used to create a learning health system that identifies inequities in care provision so that tailored interventions can be continuously and routinely implemented to improve inequalities in access and health outcomes.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Luis C. Garcia Peraza Herrera

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - The goal of this PhD project is to develop an innovative framework for generating synthetic surgical videos through command prompts. This research aims to advance the field of surgical simulation by creating realistic and diverse datasets for training and evaluating computer vision models in surgery.  Our ultimate aspiration is to establish a system akin to DALL-E, whereby we can seamlessly request the generation of synthetic surgical videos on-demand.

We aim to explore methods that use command prompts as a guiding mechanism, investigating the integration of procedural commands to control the content, complexity, and variability of the simulated surgeries. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Miao Guo

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisors:

Professor Mark Ide

Dr David Moyes

Dr Saeed Shoaie 

Project Outline:

Background - Microbiomes play crucial roles in the homeostasis of various ecosystems (including environmental and human). The vast majority of the microbial capabilities in these ecosystems have yet to be unlocked. Metagenomic technologies have enabled greater insights into microbial biodiversity and distributions in diverse microbiome ecosystems. However, the substantial information and patterns hidden in these high-dimensional data are yet to be discovered. Notably, by exploring these patterns, we could 1) predict microbiome-environment (or host) interaction, and microbiome composition; 2) classify microbial features and identify ‘core’ microbiota which could diagnose ecosystem stability and health; 3) understand spatial and temporal characteristics of microbiomes to further optimise microbial systems. Furthermore longitudinal studies provide temporal information for complex trajectories of microbes within a community and offer valuable insights e.g. microbiome evolution. Previous work by supervisor team (Dr David Moyes and Dr Saeed Shoaie) has identified specific clusters of persistent and transient microbes longitudinally with associations with general health status. Temporal dynamics and evolution of the microbiome along time-series remain largely unexplored due to the resolution of time-series data being often constrained by the experimental design and sampling-points capacity. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Huajie Jin

Health Service and Population Health Research, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor James MacCabe

Dr Angus Roberts

Project Outline:

Background - Healthcare interventions, such as antipsychotic medication, have lasting effects on patients' health outcomes and resource utilisation beyond the period typically considered in clinical trials. To evaluate the long-term cost-effectiveness of these interventions, health technology assessment agencies, including National Institute for Health and Care Excellence (NICE) in the UK, recommend developing economic models with a sufficiently extended time horizon (1). This allows for the extrapolation of short-term outcomes over a long term or even a lifetime horizon, so that cost effectiveness and cost benefit analyses can consider medium- and long-term impacts of interventions. 

The primary constraints of economic models are (a) the quality of the data informing the parameters that are fed into the model (e.g. treatment costs, probabilities of transitioning from one health state to another), (b) the assumptions of the model (e.g. homogeneous population, rational prescribing, full treatment adherence, stable treatment costs, unchanging healthcare systems) and (c) the complexity of the model (e.g. the structure adopted by most existing models only considered three health states (relapsed, remitted, deceased) which are crude representations of the complex reality) (2). 

Current economic models encounter challenges in various domains, which may contribute to a degree of instability, potentially yielding results that are less precise or unbiased, accompanied by broader error margins (3). Consequently, confidence in these models among healthcare stakeholders, including buyers, providers, and consumers, may be diminished, possibly influencing resource allocation decisions that are perceived as less robust.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Michael Gomez

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisors:

Dr Matthew Howard

Dr Adam Shortland

Project Outline:

Background - The healthcare landscape is rapidly evolving with the integration of digital technologies, which offer innovative solutions for patient health monitoring, rehabilitation, and early detection of disorders such as Parkinson’s disease, essential tremor and Huntingdon’s disease [1]. In particular, the increasing availability of inexpensive, millimetre-sized inertial sensors, such as accelerometers and gyroscopes, has enabled the development of small, wearable sensors for measuring patient body movement. In being lightweight and easily attachable to clothing, these sensors allow for continuous, non-invasive collection of large amounts of movement data, both within and outside the lab/clinic, which can be processed for various monitoring and diagnostic purposes.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Katie Dalrymple

Department of Nutritional Sciences, School of Life Course & Population Sciences

Co-Supervisor:

Professor Laura Magee

Project Outline:

Background - This proposal seeks to address the knowledge gap of medicines and supplements used in pregnancy and the long-term effects on maternal and infant health outcomes. While many pregnant women require medications to manage various health conditions, there is limited research on the safety and effectiveness of these medications and their potential impact on long-term outcomes, particularly physical, psychological, and developmental outcomes in the infant. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Martin Chapman

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisor:

Dr Steffen Zschaler

Project Outline:

Background - Novel techniques are needed to better understand population health and health systems, the inequalities that exist within these systems, and the efficacy of interventions designed to address these inequalities. This understanding can then be factored into the decisions made by policymakers and ultimately shape population-level medicine and health.

Artificial Intelligence (AI) simulation is one such technique, allowing for digital environments to be constructed that model the individuals within a population, their behaviours, and the potential impact of proposed interventions upon them. To date, AI simulation techniques have typically drawn on traditional, model-based approaches (e.g. agent-based modelling). However, data-based AI, such as machine learning (which also encompasses Large Language Models (LLMs)), has shown great promise in recent years. Hybrid approaches -- which sit at the intersection of both model and data-based AI, drawing on the strengths of each (representing complex systems and identifying complex patterns in data, respectively) -- are therefore of great interest. With access to model-based AI that also draws from recent advances in data-based AI, even more powerful simulation models could be built that can augment the investigation of complex phenomena such as health systems and interventions.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Kelly Diederen

Department of Psychosis Studies, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Dr Tom Spencer

Project Outline:

Background - Theoretical computational neuroscience has significantly advanced our understanding of psychiatric symptoms and illnesses over the past decade. This interdisciplinary field uses mathematical models and computational simulations to provide a more mechanistic understanding of mental health, adding rigour to the field and paving the way for clinical translation(1). Unlike purely data-driven approaches, these models offer interpretability, linking observed deficits to our existing knowledge about the system.

Computational models have substantially enhanced our understanding of positive psychotic symptoms, such as hallucinations and delusions, by providing a perspective through maladaptive belief updating (2). Belief updating, a process that integrates and balances anticipated and novel information, is fundamental to the formation and revision of our beliefs. This process is intimately associated with dopamine (3), the primary neurotransmitter implicated in psychosis. This association establishes a direct correlation between behavioural information and neurobiological data. In this approach, computational models, encompassing reinforcement learning and Bayesian models, are applied to behavioural outcomes, such as choice behaviour.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Richard Jefferson-Loveday

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Jack Lee

Project Outline:

Background - The project will focus on using reduced order CFD modelling to enhance magnetic resonance imaging for haemodynamic studies.  Four-dimensional flow cardiovascular magnetic resonance (CMR) imaging allows the direct measurement of blood flow velocity fields both spatially and temporally for detailed assessment of haemodynamics.  The spatial and temporal resolutions of the approach however can be coarse and may also be heavily corrupted by noise, which precludes their use in building a digital twin of the patient.  There is also no pressure field information associated with this technique that characterise the severity and/or outcome of intervention for e.g. valve implantations and aortic coarctation which are known to involve complex haemodynamics.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Johnny Downs

Department of Child & Adolescent Psychiatry, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Dr Zina Ibrahim

Project Outline:

Background - Large Language Models (LLMs) such as GPT-4 possess the capability of "universal structuring," allowing efficient abstraction of patient information from clinical text on a large scale. In biomedical applications, GPT-4 has demonstrated remarkable performance even without domain specific fine-tuning. For example, LLM can achieve high grades when taking expert-level medical exams, such as the  USMLE, without the need for costly task- self-refinement.1  This suggests that applying it is potentially feasible to apply LLMs directly to medical record text to rapidly synthesise information within a patient record, for example asking “has XXX patient has ever had an adverse reaction to a medication, or across a set of medical records, “how many patients in XXX team have evicted from their home in the last year”  within substantial domain specific training. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Oya Celiktutan

Department of Engineering, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Professor Lorna Fraser

Project Outline:

Background - Paediatric Palliative Care (PPC) concerns the medical, psychosocial, spiritual, and economic needs of children with life-limiting conditions (LLCs). Within England, the population of such children is expected to be between 67.0 and 84.2 per 1000 by the year 2030.  Life for young patients can be a very lonely and distressing experience as they are suffering, vulnerable, often socially isolated and subjected to unpleasant medical procedures. The field of Paediatric Palliative Care attempts to provide holistic care for children suffering from various life-limiting oncological and neurological disorders (Fraser et al., 2012) by addressing their specific needs for optimizing the patients’ quality of life (QoL) (Meghani, 2004). A systematic review on the experiences of children with LLCs found that they often preferred staying at home (Castor et al., 2018), leading to mental-health concerns likes loneliness. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Nicholas Cummins

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Ewan Carr

Dr Judith Dineley

Project Outline:

Background - 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Thomas Booth

Department of Cancer Imaging, School of Biomedical Engineering & Imaging Sciences

Co-Supervisor:

Dr Marc Modat

Project Outline:

Background - The burgeoning workload in radiology departments has reached a critical point, necessitating innovative solutions. In this context, artificial intelligence (AI) emerges as a beacon of hope. Our centre, home to a vast repository of over 200,000 MRI scans and corresponding reports from various UK hospitals, provides an exceptional PhD research opportunity in developing an 'AI Radiologist.' This project aims to create a comprehensive, automatic diagnostic framework for MRI analysis, tackling key real-world challenges.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Alfredo Iacoangeli

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Professor Andrea Malaspina

Project Outline:

Background - Dysregulation of gut-brain axis (GBA) crosstalk and significant gastrointestinal comorbidities have been observed in patients with neurodegenerative diseases (NDs) such as Parkinson’s and Alzheimer’s disease [1,2], with degeneration occurring first in the gut before spreading to the CNS. Similarly, changes observed in the composition of the gut microbiome in MND patients is thought to promote or enhance an immune/inflammatory response which could not only alter bowel motility but also drive the changes of both innate and adaptive immune responses known to impact on the progression of ALS [3]. With regard to gastrointestinal symptoms, constipation is commonly seen in ND patients, and delayed gastric emptying and colonic transit times have been reported even in the absence of impaired motor symptoms, suggesting autonomic dysfunction

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Ammar Al-Chalabi

Department of Basic and Clinical Neuroscience, IoPPN

Co-Supervisors:

Dr Alfredo Iacoangeli

Dr Ahmad Al Khleifat

Project Outline:

Background - Motor neuron disease (MND, ALS) is a neurodegenerative disease in which the nerve cells controlling voluntary movement are progressively lost. The result is a spreading and worsening weakness that leads to complete paralysis, with 50% of people dead within two years because the breathing muscles have been affected. MND kills 1 in every 300 people, making it as common as multiple sclerosis in the UK, but the high rate of death means it appears rare. There is no cure, and the only treatment currently available in the UK is Riluzole, which slows the disease almost imperceptibly. Our understanding of what causes MND is rapidly improving, and as a result there are many new potential treatments that need to be tested in clinical trials. To be accepted as valid evidence for licensing a new therapy, a clinical trial needs to give some people a placebo, rather than the active drug. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Ewan Carr

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Professor Kimberley Goldsmith

Project Outline:

Background - The digital transformation of healthcare is generating ever-larger amounts of high-dimensional, high-frequency, and multimodal data. This includes data from routine appointments as well as remotely collected data from smartphones and wearables.

These emerging data streams have the potential to transform the management of long-term conditions such as diabetes or psoriasis. Long-term conditions are experienced by one-third of people and account for half of hospital admissions.

However, this potential can only be realised if new data streams are paired with appropriate statistical techniques that reflect the complexity of heterogeneous disease trajectories, fluctuating symptoms, and alternating periods of remission and relapse.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Mariam Molokhia

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisors:

Dr Mariam Molokhia

Dr Kathleen Steinhofel

Dr Kate Bramham

Project Outline:

Background - Chronic kidney disease (CKD) is common, found in 14% of people aged 65-74 and 33% of people aged ≥75 years in England [1, 2], and the prevalence increases with age. [3] In some patients, kidney function continues to slowly decline, resulting in end stage kidney disease (ESKD). Patients with ESKD either choose to forgo renal replacement therapy and be managed conservatively for symptom control and quality of life,  or choose  dialysis. Currently, we are unable to identify patients who are being managed conservatively, although there is significant variation [4] and suboptimal care.  As the majority of these conservative care patients are elderly, the lack of data limits our ability to provide renal care and to understand management of these patients and their outcomes.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Oleg Glebov

Department of Psychological Medicine, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Dr Rossano Schifanella

Project Outline:

Background - Psychotropic drugs are routinely used for treatment of mental health disease. Since their advent in the 1960s, antidepressants (ADs) have remained by far the most common psychotropic drugs, accounting for 83.4mln prescriptions in the year 2021/22 in the UK alone. As a result of their ubiquitousness, low cost, and favourable safety profile, ADs are highly prominent not only as therapeutics against depression but also as prime candidates for drug repurposing.

I am a cell biologist/neuroscientist using a multi-disciplinary approach for repurposing of psychotropic drugs for wider therapeutic applications (https://febs.onlinelibrary.wiley.com/doi/full/10.1111/febs.15369, https://www.frontiersin.org/articles/10.3389/fphar.2021.787261/full, https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-023-02877-9). To this end, I have recently begun to use retrospective analysis of anonymised clinical data for the purpose of  identifying associations between drug prescription and disease protection. Firstly, I collaborate with NIHR Maudsley Biomedical Research Centre investigators, using the Clinical Record Interactive Search (CRIS) system (https://www.maudsleybrc.nihr.ac.uk/facilities/clinical-record-interactive-search-cris/) to extract data from the mental health outpatient records. As a complementary approach, together with Dr. Rossano Schifanella (Dept. of Computer Science, University of Turin), we utilise a custom pipeline for nation-wide integration of open-source data at the local geography level.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Sarah Marzi

Department of Basic and Clinical Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Jernej Ule

Dr Ahmad Al Khleifat

Project Outline:

Background - Amyotrophic Lateral Sclerosis (ALS) is a devastating neurodegenerative disorder characterized by motor neuron degeneration. The dysregulation of RNA splicing and the activation of transposable elements have been implicated in ALS pathogenesis, with RNA-binding proteins playing a pivotal role. This bioinformatics PhD project aims to unravel the complex interplay between epigenetic modifications, RNA processing, and RNA-binding proteins in sporadic ALS cases, utilizing diverse genomic datasets and in silico analyses.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Daniele Quecia

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - Acknowledging the significant impact of environmental factors on public health, as exemplified by the UK's 2022 heatwaves, our project leverages Earth Observation (EO) tools to investigate the correlation between environmental conditions and health outcomes. The complexity of EO data has historically hindered its integration into health research. We have developed the MedSat dataset in collaboration with the AI4EO Lab, Germany, showcased at NeurIPS 2023. This dataset combines health data with satellite imagery from 2019-2020, covering 33,000 areas in England. It also includes extensive prescription data for various health conditions, ranging from diabetes to asthma to depression, collected monthly from 2010 to the present. Furthermore, our dataset offers the potential to be integrated with other health datasets in England, such as the individual-level data from BioBank. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Daniele Quercia

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Project Outline:

Background - In the rapidly advancing field of Artificial Intelligence (AI) in healthcare, the critical imperative is to develop responsible AI solutions that adhere to ethical guidelines. This PhD thesis proposes a specialized and efficient methodology for integrating ethical considerations into AI development within the healthcare domain. The approach centers on the use of prompt cards, collaboratively designed by a diverse team comprising AI developers and experts in standardization at Nokia Bell Labs Cambridge, and healthcare professionals at King's College London. These prompt cards offer a succinct, context-specific, and actionable framework for embedding ethical principles in AI algorithms designed for healthcare applications. Unlike conventional checklists, prompt cards provide a user-friendly and agile solution, addressing the unique challenges and ethical nuances inherent in healthcare AI. By streamlining the integration of ethical guidelines, this research aims to contribute to the responsible and ethical advancement of AI technologies in healthcare, ensuring they align with the highest standards and uphold patient welfare.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Jernej Ule

Department of Basic and Clinical Neuroscience, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Alfredo Iacoangeli

Dr Sarah Marzi

Project Outline:

Background - Gene therapy for brain disorders holds great promise, yet the irreversible nature of most treatment strategies presents significant health risks. Therapies for slowly progressing developmental and age-dependent brain disorders such as neurodegeneration are particularly challenging, because small subsets of specific cell types are affected by pathogenic mechanisms that depend on certain contexts. This project will interrogate multiple types of transcriptomic data generated from postmortem brain samples of patients with neurodegeneration and age-matched controls, as well as human induced pluripotent stem cell (hiPSC)-derived neurons, glia and organoid models of neurodegeration. This will serve to identify transcriptomic changes that represent the early stages of neurodegeneration, focusing on the alternatively spliced mRNA isoforms. Next, computational models of the multi-layered RNA regulatory programmes will be built to identify the cis-regulatory RNA signatures that could explain why specific mRNA isoforms change their expression in brain disorders. Finally, knowledge of these signatures will be harnessed to computationally design minimal RNA sensor-autoregulatory units that could drive context-dependent and spatially localised expression of gene therapy. This will be key to ensure therapeutic delivery to the small subsets of neurons that undergo pathogenic changes, while avoiding side-effects of gene therapy in healthy cells. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Marina Cecelja

Department of Twin Research & Genetic Epidemiology, School of Life Course and Population Sciences

Co-Supervisor:

Professor Claire Steves

Project Outline:

Background - Approximately 25% of the general population carries at least one ε4 allele of the Apolipoprotein E (APOE ε4) with a major role in atherosclerotic cardiovascular disease. Additionally, it is and important genetic risk factor for late onset Alzheimer’s disease and other forms of dementia, such as vascular dementia, and cognitive decline in the general population. As well as its known action on lipid metabolism, recent research has implicated APOE in vascular ageing with APOE protein accumulating in extracellular matrix components of vascular tissue in association with senescent vascular smooth cells. In particular, APOE may drive amyloid aggregation in the extracellular matrix of vascular tissue which is strongly implicated not only in Alzheimer’s disease but with accelerated systemic vascular ageing (i.e. atherosclerosis and arteriosclerosis). As such, APOE may mediate the observed association between cardiovascular disease and cognitive decline through its effects on vascular tissue (1). However, many individuals show resilience against dementia and CV disease despite APOE-ε4 polymorphism. Identifying individuals whose brain and vascular ageing path remains unaffected by known stressors (such as APOE) offers a unique opportunity to identify protective factors and resilience pathways for these highly prevalent chronic conditions. The aim of the present PhD project is to identify sources of vascular and brain ageing resilience in the context of a APOE phenotype.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Claire Steves

Department of Twin Research & Genetic Epidemiology, School of Life Course and Population Sciences

Co-Supervisors:

Dr Nicholas Cummins

Dr Jonathan Birns

Dr Peter Sommervile

Dr David Smithard

Dr Salma Ayis

Project Outline:

Background - Pneumonia due to aspiration is a common, but often late-diagnosed cause of morbidity and death in older people, and comes about due to a combination of swallowing problems and altered immune response.  Research has shown that this can be improved by exercises to strengthen the tongue, suprahyoid complex and anterior neck muscles. Up to 50% of older adults admitted to hospital have swallowing problems but they are often not identified in routine clinical practice, and when they are it is often when it is too late to intervene.

This project's overall objective is to develop a simple tool which could be used at home, in the clinic, or emergency department to detect subtle signs of muscle strength loss which could then be modulated through use of preventative exercises. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Claire Steves

Department of Twin Research & Genetic Epidemiology, School of Life Course and Population Sciences

Co-Supervisors:

Dr Michela Antonelli

Professor Katherine Sleeman

Dr Adrian Hopper

Professor Catherine Evans

Dr Peter May

Project Outline:

Background - Many people die in hospital after escalating secondary care utilisation, with burdensome unplanned care. Our initial work suggests that, using machine learning with time-series hospital data, this can be anticipated, providing an opportunity for better planning ahead.  Palliative care, and end of life teams have skills to facilitate care which focuses on patient needs and choices, and could enable very different journeys at the end of life, guided by compassionate and holistic planning, started well in advance.  This project aims to co-develop with patients, risk stratification using machine learning which can inform palliative care teams of patients at risk of end-of life, preventing patient inequity by evaluating need for care based on personalised data tracks.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Claire Steves

Department of Twin Research & Genetic Epidemiology, School of Life Course and Population Sciences

Co-Supervisors:

Dr Kerrin Small

Dr Michela Antonelli

Dr Ruth Hackett 

Project Outline:

Background - This project aims to use the unique TwinsUK "real twins" dataset to understand changes in healthcare utilisation over time and age using timeseries in health care record data.  Using these data we will generate simulated twins, with different lifestyles as a way to explore the effect of personal interventions on health, such as diet, exercise at different points across the lifecourse, within the bounds of the same genotype. As well as giving insights to health care providers and policy makers, this could have impact for patients themselves, through design of direct to patient interfaces.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Zina Ibrahim

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Richard Dobson

Professor James Galloway

Dr Rosita Zakeri

Professor James Teo

Project Outline:

Background - Deep Learning Early Warning Systems respond to the national need for proactive AI-driven clinical early-warning systems in NHS hospitals [1]. The objective is to enhance the recognition of patient deterioration beyond current ad-hoc tools like NEWS2 [2], employing deep learning early warning systems that leverage multimodal longitudinal patient data (e.g., repeated measurements of blood results, vital signs, clinical notes, medical examinations). Despite their potential transformative impact, these systems often face challenges related to bias, interpretability and reduced performance in settings lacking continuous data streams such as general hospital wards, where their insights could be most valuable. This project's goal is to overcome these limitations by developing better predictive AI models which draw on multimodal knowledge accumulated in EHRs, in addition to clinical guidelines and known causal pathways.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Zina Ibrahim

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Honghan Wu

Professor James Teo

Professor Ammar Al-Chalabi

Project Outline:

Background - Neurodegeneration is defined as the progressive loss of specific neuronal components or functions. The ageing population corresponds to an increasing prevalence of neurodegenerative diseases. This demographic shift places a growing burden on healthcare systems, leading to resource strain and financial implications. The annual toll of neurodegenerative diseases imposes high strains on the NHS. The urgency for early diagnosis in neurodegenerative diseases has been underscored by research, emphasising its potential impact on enabling early interventions and facilitating patient enrolment in clinical trials. Despite this, a considerable challenge persists as patients often remain asymptomatic for an extended period or face misdiagnosis, leading to delayed referral to the appropriate specialty, hindering timely intervention and representing a missed opportunity to mitigate disease progression effectively.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Jie M. Zhang

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Martin Chapman

Project Outline:

Background - Skin diseases can have markedly different appearances across various skin tones. Dermatology has historically been limited by a lack of diversity in medical education resources. In the UK, most dermatological training and resources are skewed towards conditions as they appear on lighter skin, leaving a significant gap in the accurate and fair diagnosis and treatment of skin conditions in patients with Asian or dark-coloured skin.

The primary motivation for this project is to address the significant healthcare inequality faced by patients with non-white skin tones in the United Kingdom. Studies and reports have highlighted the challenges and inaccuracies in diagnosing skin conditions in these populations, primarily due to the lack of training and resources focusing on diverse skin tones. By leveraging advanced AI techniques, this project aims to bridge this gap, providing dermatologists with tools that are well-versed in identifying and diagnosing conditions across a wide spectrum of skin tones. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Alfredo Iacoangeli

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Petroula Proitsi

Dr Oliver Pain

Project Outline:

Background - Identifying genetic factors shared by multiple traits related to ageing and neurodegeneration, can help elucidating their underlying biology by highlighting fundamental mechanisms, and can offer opportunities for the development and repurposing of treatments. Local genetic correlation based on Genome Wide Association Studies (GWAS) data enables to identify pleiotropy across multiple traits; however, statistical fine-mapping and colocalization have shown that the variants in these loci often differ between traits suggesting the role of distinct biology despite being shared loci. We will explore the use of functional genomics summary data to characterise these distinctive mechanisms within a statistical fine-mapping and colocalization framework. This new methodology will be able to provide biological insight about the relationship between genetic loci shared by multiple traits and their distinct genetic causes. Understanding such mechanisms is essential if we want to translate GWAS findings into biological and medical applications leading to the development of effective treatments.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Alfredo Iacoangeli

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Professor Ammar Al-Chalabi

Project Outline:

Background - There is increasing recognition of heterogeneity within and overlap between ALS and other neurodegenerative diseases both in terms of clinical phenotype and underlying biological mechanisms. This is well exemplified by the established spectrum of disease that exists between FTD and ALS. Such a complex landscape is an obstacle to the identification of relevant subgroups for personalised treatment. Understanding the relationship between ALS and FTD, their sub-phenotypes and underlying causes is therefore key to the development of targeted approaches to therapy.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Alfredo Iacoangeli

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisor:

Dr Edoardo Milanetti

Project Outline:

Background - Neurodegenerative diseases (NDs), such as AD, FTD, MND, Ataxias or PD, are incurable and debilitating conditions that result in progressive degeneration and/or death of nerve cells. Genetic variants are responsible for a large proportion of ND patients. However, among several millions of variants present in a given genome, the vast majority has no consequence on the person’s health. Being able to predict NDs causing variants is fundamental for our research. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Tim Neate

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Dr Iain Marshall

Project Outline:

Background - Patient portals are a patient-facing, web-based technology offered by healthcare organisations to provide patients with access to their electronic health record (EHR) data and also often provide additional features to engage patients in their healthcare. Such access has been found to improve medication adherence, blood pressure and glycaemic control, improve health ownership, reduce high-cost healthcare utilisation in patients with long-term conditions, and decrease primary care and speciality care visits. Patients have reported improved access to information, better insight into their conditions, reduced anxiety, improvements in health behaviours and better engagement with care.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Iain Marshall

Department of Population Health Sciences, School of Life Course and Population Sciences

Project Outline:

Background - Health professionals need frequent, up-to-date, and reliable information to guide their practice; for example, GPs typically have a clinical uncertainty needing answered for every second patient they see. Ideally, these uncertainties should be resolved by consulting systematic reviews: high quality syntheses of health research.

Systematic reviews provide the gold standard of health research evidence to support decision making. However they are labour intensive, and time-consuming, and may take two years to complete. Many are already out of date by the time they are published; and there are many questions arsing in clinical practice for which no systematic review exists. There is a burgeoning research field in how to use machine learning/natural language processing to speed up the conduct of systematic reviews.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Vasa Curcin

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisor:

Dr Iain Marshall

Project Outline:

Background - Artificial Intelligence (AI) has received much attention globally following the release of multiple Large Language Models (LLMs) which boast significant capabilities and knowledge bases from their voluminous training data sets. The application of these LLMs have been experimented with in various science, technology, engineering, and mathematics (STEM) applications with an aim to aid or replace human-led applications. Within the healthcare sector, LLM have been widely researched and praised for their ability to processing healthcare knowledge and formulating clinician level responses ​(1,2)​. They have also been experimented with for their ability to assist in potentially alleviating frontline health worker workloads and reducing error rates ​(3,4)​. 

Recent advances in LLMs have not been adequately explored for their potential to improve health policy and guideline formulation. Augmentation of evidence scanning and policy interrogation by LLMs poses a great potential area of research with tangible effects resulting in reduced delays in current guideline and policy formulation processes in many LMICs

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Vasa Curcin

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisor:

Dr Nicola Paoletti

Project Outline:

Background - Computable clinical guidelines (CPGs) are the foundation of clinical decision support systems, aiming to improve the quality of care, reduce errors in disease management and provide clinicians with a standardised approach to treatment. However, they are not without their limitations. The most notable is their ability to handle co-morbidities - issues like polypharmacy must be considered to prevent adverse reactions between drug recommendations. In these cases, co-ordination and argumentation is required to resolve conflicts between recommendations and generally require clinical experts and knowledge engineers to resolve. As multi- morbidity is common in patients with chronic disease, this is a significant issue - in the UK, 67% of patients over 65 have two or more chronic conditions.

Electronic Health Record systems (EHRs) present a wealth of data available for research and quality improvement, and are the main setting for clinical decision sup- port systems. However, there are two main issues with these systems: most of the data is unstructured (e.g. free text) making it difficult to extract information; and the lack of standardisation of recording of data between clinicians, organisations & systems often make queries complex and limit portability of CPG implementations. While there are certain standards for recording data, such as SNOMED-CT, they are not always used correctly, or at all.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Steffen Zschaler

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisor:

Professor Richard Dobson

Project Outline:

Background - Hospitals are continuously overstretched. To cope with the ever increasing demand and ensure patient safety, they need to continually identify improvements to processes and resource usage, balancing the impact of such process interventions on costs, efficiency, and patient risk. However, hospitals and their departments are highly complex systems and, therefore, the impacts of any given intervention are difficult to predict accurately. Experimenting with interventions directly in the hospital, thus, becomes a costly and risky strategy. Instead, many hospitals are using computer simulation to explore proposed interventions in silico before implementing anything in the real world. Typically, a new simulation is developed from scratch for every change project.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Ingrid Wolfe

Department of Women & Children's Health, School of Life Course & Population Sciences

Co-Supervisor:

Dr Jeremy Yates

Project Outline:

Background - There is an urgent need to develop new health care models to reduce inequalities and improve outcomes for children. Through a decade of work in the Children and Young People's Health Partnership, we have developed the CHILDS Learning Health System which together with Population Health Management tools enables us to identify children with unmet needs in the community and target early intervention care. 

With the rapid expansion of knowledge and technology and a health care system that performs below acceptable levels for ensuring patient safety and needs, a gap has developed between what recommendations and practice. 

This proposal is about using data to target care for children with unmet needs as a mechanism for improving children's health care equity. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Iain Marshall

Department of Population Health Sciences, School of Life Course and Population Sciences

Co-Supervisor:

Dr Marina Soley-Bori

Project Outline:

Background - Stroke is a leading cause of death and disability globally, disproportionately affecting people experiencing socioeconomic disadvantage (1). Although the impact of social determinants of health (SDoH), like gender, ethnicity, and income, on stroke outcomes is recognised, the underlying causal pathways remain poorly understood using conventional statistical modelling (5). 

There is an opportunity to use novel causal analysis methods to make sense of these health inequalities; which in turn will allow us to intervene to improve the situation.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Grainne Mary McAlonan

Department of Forensic and Neurodevelopmental Science, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Lucilla Poston

Dr Johnny Downs

Project Outline:

Background - There is evidence from epidemiological and preclinical research that exposure to maternal immune activation during prenatal life can increase the likelihood of neurodevelopmental difficulties (such as autism), in some (not all) offspring. Inflammation in pregnancy is common but what factors increase likelihood of childhood difficulties and what influences protect or confer resilience are poorly understood. In the Covid pandemic pregnant mothers were exposed to SARS-2-Coronavirus at the same rates as the general population. This provides us with an opportunity, indeed a responsibility, to investigate factors which influence outcomes for the generation in utero during the pandemic. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Professor Mohammad Reza Mousavi

Department of Informatics, Faculty of Natural, Mathematical & Engineering Sciences

Co-Supervisors:

Professor Adenilso Simao

Project Outline:

Background - Cancer is partially caused by mutations accumulated over time that corrupt cell pathways. Other factors such as smoking, sunlight exposure, age, and phenotype can influence the disease's emergence. The increase in public health data enabled the development of many statistical studies that explore the relationships between mutations, pathways, phenotypes, and risk factors. Causal discovery can enhance those relationships into causal associations. Causal discovery methods from observational data tend to be complex and computationally expensive, often leading to reimplementation of similar ideas in specific ways to facilitate efficient causal analysis. We develop a compositional causal discovery for cancer genomics, by partitioning the data and developing a component-based causal model for cancer genomics. We plan to make the pipeline developed in this project publicly available and use Drive Health to disseminate and explore its application for other genetic diseases and conditions. 

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Angus Roberts

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Peter Schofield

Project Outline:

Background - Electronic Health Records (EHRs) contain a full record of patient's health and care. As such, they are a valuable resource for both research and patient care. If we are to reuse the information contained within EHRs, we must contend with the fact that the bulk of this information is contained within the natural language, free text narrative portion of the record: notes and letters written by clinicians. Natural Language Processing (NLP) has proved adept at extracting much of this information, from medications to symptoms and some contextual information. Other information has, however, proved more challenging. This includes information about patient recovery in mental health records, a genre in which social context and narrative description of highly variable and patient-specific symptoms further complicate extraction of information.

Extracting and understanding patient mental health recovery is critical to studies of treatment efficacy, the role of social factors in recovery, and disease progression. Whilst there are some statistical studies using survey measures of recovery, these usually have only small samples, and are therefore limited in what they can tell us about wider social patterns. Extracting information about recovery from EHR text would open a much greater quantity of rich information about recovery.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Angus Roberts

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Dr Anwar Alhaq

Project Outline:

Background - Prediction models are routinely used in healthcare to identify patients at risk of a particular clinical outcome, such as a heart attack, stroke, or diabetes. Most of these models are based on relatively small numbers of variables, often collected specifically for the purpose of running the model. The model will typically use traditional statistical techniques, although machine learning and neural network models are increasingly being employed. In contrast to these clinical models, administrative models predict outcomes of utility for the efficient delivery of patient care across hospitals, such as length of hospital stay, delayed discharge, and likelihood of readmission. In addition to relevance for individual patients, such models can help with the efficient delivery of healthcare by highlighting areas experiencing problems, helping with resource allocation, and increasing efficiency of workflow.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.

Supervisor:

Dr Angus Roberts

Biostatistics and Health Informatics Department, Institute of Psychiatry, Psychology and Neuroscience

Co-Supervisors:

Professor Robert Stewart

Project Outline:

Background - Electronic Health Records (EHRs) contain a full record of patient's health and care. As such, they are a valuable resource for both research and patient care. If we are to reuse the information contained within EHRs, we must contend with the fact that the bulk of this information is contained within the natural language, free text narrative portion of the record: notes and letters written by clinicians. Natural Language Processing (NLP) has proved adept at extracting much of this information, from medications to symptoms and some contextual information. Other information has, however, proved more challenging. This includes information about the timing and ordering of events. Understanding temporality of events recorded in the EHR text is important for any reuse of the EHR. Despite this, temporality is often expressed in vague and imprecise language, with the writer assuming that the reader has both a cultural understanding of time, and medical domain expertise when relating recorded events to each other.

For full details on project background, aims and outcomes email us at drivecdt@kcl.ac.uk.