connect with us
FORESIGHT: A novel GPT-based pipeline trained on NHS data
February 7, 2023
Zeljko Kraljevic
outlines how these foundation models for medicine can provide the potential for a diverse integration of medical data that includes electronic health records, images, lab values, biologic layers such as the genome and gut microbiome...
Over the past four years, the AI world has surged ahead with large language models (LLMs), also known as “foundation models” which can be adapted to achieve many linguistic tasks. You’ve probably seen a plethora of articles in the media recently about some of these models (ChatGPT, Dalle-2), that can write coherent essays, write code, but also generate art and films, and many other capabilities.
With the NHS at breaking point, a critical question is whether these AI approaches could be used to improve care. Hospital records hold detailed information about each patient's health status and general clinical history, a large portion of which is stored within the unstructured text. Temporal modelling of this medical history, which considers the sequence of events, could be used to forecast and simulate future events, estimate risk, suggest alternative diagnoses or forecast complications.
I have developed
Foresight
as part of the
CogStack
platform, a novel GPT-based pipeline that is trained on NHS data to forecast future medical events such as disorders, medications, symptoms and interventions.
On tests in two large King’s Health Partner hospitals (King’s College Hospital, South London and Maudsley) and the US MIMIC-III dataset Foresight performed well when set challenges by clinicians. The model is being used for many uses including real-world risk estimation, virtual clinical trials and clinical research to study the progression of diseases, simulate interventions and counterfactuals, and for educational purposes.
Preprint:
https://arxiv.org/abs/2212.08072
Share
We’re pleased to announce that Charles Friedman from the University of Michigan Medical School , will deliver our March Seminar Series with his talk, "Why AI and Learning Health Systems Need Each Other " . Charles will begin by advancing the idea that, while both are extremely important: AI is a means and Learning Health Systems (LHS) are an end--and why it is most important to maintain that distinction. He will then introduce the socio-technical infrastructure required for high-functioning learning systems and argue that this infrastructure provides a framework, actually a schematic, for successfully implementing AI into healthcare. Charles Friedman is Professor of Learning Health Sciences at the University of Michigan Medical School, where he directs the Knowledge Systems Laboratory. He was formerly Founding Chair of the Department of Learning Health Sciences and the Josiah Macy Jr. Professor of Medical Education. He holds joint appointments in the Schools of information and Public Health. He is editor-in-chief of the open-access journal Learning Health Systems and co-chair of the multi-national movement to Mobilize Computable Biomedical Knowledge. Throughout his career, Friedman has developed and studied methods to improve health, education, and research through innovative applications of information technology. Most recently, Friedman has focused his academic interests and activities on the concept of Learning Health Systems that improve health by marrying discovery to implementation, and the socio-technical infrastructure required to sustain these systems. Friedman is a Distinguished Fellow of the American College of Medical Informatics, and a founding fellow of the International Academy of Health Sciences Informatics. He holds an honorary doctorate from the University of Lucerne in Switzerland for his contributions to the science of Learning Health Systems. Prior to coming to Michigan, Friedman held executive positions at the Office of the National Coordinator for Health IT (ONC) in the U.S. Department of Health and Human Services. Immediately prior to his work in the government, he was Associate Vice Chancellor for Biomedical Informatics, and Founding Director of the Center for Biomedical Informatics at the University of Pittsburgh. Seminar Series Event: "Why AI and Learning Health Systems Need Each Other" Date and Time: Wednesday 26 March 2025, 10:00 – 11.00 hrs (BST) Location: The Anatomy Museum, King's Building, Room K6.36, Strand Campus, Strand, London, WC2R 2LS Attendance: Mandatory for all DRIVE-Health students, therefore please accept the calendar invitation. Registration: Alumni and wider King's College London research community all welcome - please email drive-health-cdt@kcl.ac.uk to let us know if you would like to attend.
It was a pleasure to welcome C hris Tomlinson from LifeArc , who delivered our February Seminar Series with his talk, "Translational Clinical Data Science: from patient data to patient impact " . Chris gave an overview of LifeArc, a self-funded translational research charity, seeking to deliver patient benefit and address unmet needs. As UK Health Data & AI lead, he focuses on how they harness data science and AI to fulfil their aim: to ‘make life sciences, life changing’. Chris is a clinician by background, specialising in Anaesthesia & Intensive Care, before transitioning to full-time research. His work leverages electronic health records, epidemiology and artificial intelligence at scale to advance our understanding of health and disease, and address the fundamental challenges of precision medicine. His research has been featured in top medical journals and informed both policy and clinical practice internationally. Seminar Series Event: "Translational Clinical Data Science: from patient data to patient impact" Date and Time: Thursday 27 February 2025, 15:00 – 16.00 hrs (BST) Location: Hodgkin Building, Classroom 6, Guy's Campus Attendance: Mandatory for all DRIVE-Health students Registration: Students, alumni and wider King's College London research community, please email drive-health-cdt@kcl.ac.uk to register.
