Background

Adult obesity rates in the UK are rising. Adults with obesity can access interventions through the National Health Service (NHS) but patients experience high individual variability in response to treatment, and this variability disproportionately affects minority ethnic groups1-3. Causal artificial intelligence (AI) integrates machine learning with causal inference to predict individualised treatment responses from observational data (e.g., electronic health records)4-6. Causal AI could help to reduce variability in patient outcomes from obesity interventions; however, formal assumptions about the underlying causal problem must be tested before models can be developed and implemented. 

Novelty & Importance

Causal AI could guide personalised treatment decisions that improve patient outcomes and reduce health inequalities in obesity. Causal AI is yet to be explored in obesity or tested in practice.

Aims & Objectives

Overall aim: To improve the validity and reliability of causal AI for routine obesity care.

Objective 1: Develop and validate a taxonomy of obesity interventions delivered in multidisciplinary weight management services to inform future causal learning. 

Objective 2: Formulate the causal structure of the problem (i.e., variability in obesity treatment outcomes)

Objective 3: Estimate causal quantities and assess the plausibility of underlying assumptions of treatment effects in obesity interventions  

Objective 4: Develop and test methods and or interventions to improve the plausibility of underlying assumptions to reduce bias in causal AI models

References:

1.        Brown TJ, O'Malley C, Blackshaw J, et al. Exploring the evidence base for Tier 3 weight management interventions for adults: a systematic review. Clin Obes. Oct 2017;7(5):260-272. doi:10.1111/cob.12204

2.        Dent R, McPherson R, Harper ME. Factors affecting weight loss variability in obesity. Metabolism. Dec 2020;113:154388. doi:10.1016/j.metabol.2020.154388

3.        Hazlehurst JM, Logue J, Parretti HM, et al. Developing Integrated Clinical Pathways for the Management of Clinically Severe Adult Obesity: a Critique of NHS England Policy. Curr Obes Rep. Dec 2020;9(4):530-543. doi:10.1007/s13679-020-00416-8

4.        Bica I, Alaa AM, Lambert C, van der Schaar M. From Real-World Patient Data to Individualized Treatment Effects Using Machine Learning: Current and Future Methods to Address Underlying Challenges. Clinical Pharmacology & Therapeutics. 2021/01/01 2021;109(1):87-100. doi:https://doi.org/10.1002/cpt.1907

5.        Feuerriegel S, Frauen D, Melnychuk V, et al. Causal machine learning for predicting treatment outcomes. Nature Medicine. 2024/04/01 2024;30(4):958-968. doi:10.1038/s41591-024-02902-1

6.        Sanchez P, Voisey JP, Xia T, Watson HI, O'Neil AQ, Tsaftaris SA. Causal machine learning for healthcare and precision medicine. R Soc Open Sci. Aug 2022;9(8):220638. doi:10.1098/rsos.220638