Complex modeling with detailed temporal predictors does not improve health records-based suicide risk prediction.

Autor: Shortreed SM; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA. susan.m.shortreed@kp.org.; Department of Biostatistics, University of Washington, 1705 NE Pacific St, Seattle, WA, 98195, USA. susan.m.shortreed@kp.org., Walker RL; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA., Johnson E; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA., Wellman R; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA., Cruz M; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA.; Department of Biostatistics, University of Washington, 1705 NE Pacific St, Seattle, WA, 98195, USA., Ziebell R; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA., Coley RY; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA.; Department of Biostatistics, University of Washington, 1705 NE Pacific St, Seattle, WA, 98195, USA., Yaseen ZS; U.S. Food and Drug Administration, Silver Spring, MD, USA., Dharmarajan S; U.S. Food and Drug Administration, Silver Spring, MD, USA., Penfold RB; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA., Ahmedani BK; Center for Health Policy & Health Services Research, Henry Ford Health System, 1 Ford Place, Detroit, MI, 48202, USA., Rossom RC; HealthPartners Institute, Division of Research, 8170 33rd Ave S, Minneapolis, MN, 55425, USA., Beck A; Kaiser Permanente Colorado Institute for Health Research, 2550 S. Parker Road, Suite 200, Aurora, CO, 80014, USA., Boggs JM; Kaiser Permanente Colorado Institute for Health Research, 2550 S. Parker Road, Suite 200, Aurora, CO, 80014, USA., Simon GE; Kaiser Permanente Washington Health Research Institute, 1730 Minor Avenue, Ste 1600, Seattle, WA, 98101, USA.
Jazyk: angličtina
Zdroj: NPJ digital medicine [NPJ Digit Med] 2023 Mar 23; Vol. 6 (1), pp. 47. Date of Electronic Publication: 2023 Mar 23.
DOI: 10.1038/s41746-023-00772-4
Abstrakt: Suicide risk prediction models can identify individuals for targeted intervention. Discussions of transparency, explainability, and transportability in machine learning presume complex prediction models with many variables outperform simpler models. We compared random forest, artificial neural network, and ensemble models with 1500 temporally defined predictors to logistic regression models. Data from 25,800,888 mental health visits made by 3,081,420 individuals in 7 health systems were used to train and evaluate suicidal behavior prediction models. Model performance was compared across several measures. All models performed well (area under the receiver operating curve [AUC]: 0.794-0.858). Ensemble models performed best, but improvements over a regression model with 100 predictors were minimal (AUC improvements: 0.006-0.020). Results are consistent across performance metrics and subgroups defined by race, ethnicity, and sex. Our results suggest simpler parametric models, which are easier to implement as part of routine clinical practice, perform comparably to more complex machine learning methods.
(© 2023. The Author(s).)
Databáze: MEDLINE
Externí odkaz: