Belal Hossain - Performance of high-dimensional disease risk scores in risk prediction modelling

Developing clinical risk prediction models with investigator-specified risk factors using health administrative databases often provides poor predictions since admin databases generally contain no data on smoking, alcohol, height, weight, physical activity, and diet. We developed a cohort of immigrants in BC. Adding routinely collected healthcare variables from the linked databases with the investigator-specified covariates, we developed high-dimensional disease risk scores that improve the predictive performance without overfitting. Our model will guide researchers in improving the performance of risk prediction models even if some important risk factors are unavailable.

WEBCAST VIEWING REQUIREMENTS TEST LINK, Click to confirm browser and connectivity to watch this event

10/10/2023 8:05:00 PM