Modeling disease screening, progression and prediction using machine learning techniques
Approved Research ID: 86027
Approval date: July 27th 2022
The goal of the proposed research project is to develop models that either predict the current disease status or future disease risk using known risk factors and symptoms and/or medical history data from Electronic Health Record (EHR). We plan to adopt state of the art statistical and machine learning techniques, such as graph models and deep learning to build algorithms using the UK Biobank data. These methods are well suited to handling the uncertainty and missingness typically seen in healthcare datasets. Once deployed on our cloud based platform, the algorithms will be able to serve our user population by either surfacing the risk prediction outcome to the clinicians who manage specific patient populations or triggering personalized health, nutritional and fitness related content to our users via mobile app. These medical and behavioral interventions will help them reduce their disease risks and better manage their existing conditions.