Developing clinical risk prediction models to better target suicide prevention care

Suicide prevention is a global public health concern. Identifying who might be at increased risk can increase a health system's ability to provide care and intervention to those who need it most. I will discuss the development and evaluation of suicide risk prediction models using electronic health records. To estimate these models, we used data on over 25 million mental health visits made by 3 million people in 7 health systems across the US to train and evaluate models. I will present results comparing artificial neural network, logistic regression with lasso, random forest, and ensemble models with 1500 temporally defined predictors to logistic regression models with fewer less detailed predictors. Dr. Shortreed is a Senior Investigator & Manager in the Biostatistics Division at Kaiser Permanente Washington Health Research Institute and an Affiliate Professor in the Biostatistics Department at the University of Washington. Her research brings together statistics and machine learning methods to address health science and biomedical problems, with a special emphasis on analyzing complex longitudinal data and overcoming missing-data challenges. Much of her methodological work is focused on developing and evaluating statistical inference approaches for observational data, such as data collected from electronic health care records. She is also interested in developing new machine learning methods and extending current best-practice methods, specifically for creating individualized treatment strategies and selecting which pieces of information are important to include in statistical analyses. She collaborates with scientists in a broad range of areas including cancer screening, chronic pain, depression and suicide prevention. November 19, 2024 The University of Washington is committed to ensuring digital accessibility in our services, programs, and activities. If you encounter accessibility barriers using videos found on this channel, please contact UW Video at uwvideo [at] uw [dot] edu.