Activity Number:
|
142
- Fairness and Equity in Clinical Risk Prediction: Healthcare Data for the Public Good
|
Type:
|
Invited
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 11:50 AM
|
Sponsor:
|
Biometrics Section
|
Abstract #309350
|
|
Title:
|
Assessing Racial and Ethnic Fairness of a Suicide Risk Prediction Model
|
Author(s):
|
Yates Coley* and Eric Johnson and Susan Shortreed
|
Companies:
|
Kaiser Permanente Washington Health Research Institute and Kaiser Permanente Washington Health Research Institute and Kaiser Permanente Washington Health Research Institute
|
Keywords:
|
machine learning;
random forest;
lasso;
fairness;
prediction;
mental health
|
Abstract:
|
Recent research has identified race- and ethnicity-based bias in algorithms to guide health care delivery. Use of prediction models that perform poorly in under-served populations may exacerbate existing health disparities and deny care to patients with greatest need. This talk will consider the racial and ethnic fairness of a prediction model to identify patients at risk of suicide death after a clinic visit. We explore possible sources of poor performance in racial and ethnic subgroups, including measurement error and risk modification, consider statistical methods to improve performance in small subgroups, and evaluate model discrimination and calibration within subgroups. Our presentation also generalizes knowledge gained in this research and provides guidance to statisticians considering developing prediction models for clinical use.
|
Authors who are presenting talks have a * after their name.