Machine Learning for Hypothesis Generation in Multimodal Physical and Psychosocial Healthcare Data: A Proof-of-Concept Study

Hypothesis

Unexpected hypotheses will be generated and robustly tested. Musculoskeletal outcomes in FU1 can be predicted using data from the Baseline.

Summary

Applying artificial intelligence and machine learning techniques to the ADVANCE dataset (starting with just musculoskeletal data to lean on current expert knowledge within the team) with the intention of producing publications. Collaborative funding has been awarded by Imperial College London for a small collaborative project with the Computer Science department.

The project aims to: (1) Apply unsupervised ML algorithms to ADVANCE’s baseline musculoskeletal data to act as an initial proof-of-concept study, (2) use these hypotheses to set priorities for future research, (3) test a newly generated hypothesis using typical statistical modelling. The algorithms applied during baseline will be further applied to FU1 with the new aim: 1) To develop and validate machine learning models that can predict 3-year musculoskeletal health outcomes of UK military personnel. 2) To improve the clinical applicability of these models to ensure they can be used in early risk detection and intervention planning.

Keywords

Artificial Intelligence, AI, Machine Learning, Musculoskeletal


Back to Research Proposals