Predicting Individual-Level Mortality with Traditional and Machine Learning Methods
Posted: 1/4/2023 (CSDE Seminar Series)
This week’s seminar lecture will be given by Nicholas Irons (Statistics). Irons’ presentation investigates different approaches to predicting individual-level mortality. Irons argues that individual-level mortality prediction is a fundamental challenge with implications for individuals and societies, enabling life planning, targeting of high-risk individuals, and organization of social interventions. Demographers have been primarily concerned with mortality analyses at a macro level, leveraging strong regularities in mortality rates. Besides clinical settings, individual-level mortality predictions have been largely overlooked. Irons and colleagues use the US Health and Retirement Study, a representative survey of people over 50, and estimate 12 statistical and machine learning models using over 100 predictors. Machine learning and traditional models report comparable accuracy and relatively high predictive and discriminative performance, particularly when including time-varying information (best integrated Brier Score 0.110 and mean Area Under the Curve 0.874). They observe consistent inequalities in lifespan predictability, with predictions for men, people of color, and low-educated respondents being less accurate than for their respective counterparts. Finally, they find minimal variation in the top features across groups, with age, diabetes, and smoking behaviors relevant predictors.
Date: 01/06/2023
Time: (12:30-1:30 PM PT)
Location: 101 Hans Rosling Center (Also available via zoom [insert link])