Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations

Abstract Current clinician-based and automated risk assessment methods treat the risk of suicide-related behaviors (SRBs) as static, while in actual clinical practice, SRB risk fluctuates over time. Here, we develop two closely related model classes, Event-GRU-ODE and Event-GRU-Discretized, that can...

Full description

Saved in:
Bibliographic Details
Main Authors: Yi-han Sheu, Jaak Simm, Bo Wang, Hyunjoon Lee, Jordan W. Smoller
Format: Article
Language:English
Published: Nature Portfolio 2025-03-01
Series:npj Digital Medicine
Online Access:https://doi.org/10.1038/s41746-025-01552-y
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850039572537278464
author Yi-han Sheu
Jaak Simm
Bo Wang
Hyunjoon Lee
Jordan W. Smoller
author_facet Yi-han Sheu
Jaak Simm
Bo Wang
Hyunjoon Lee
Jordan W. Smoller
author_sort Yi-han Sheu
collection DOAJ
description Abstract Current clinician-based and automated risk assessment methods treat the risk of suicide-related behaviors (SRBs) as static, while in actual clinical practice, SRB risk fluctuates over time. Here, we develop two closely related model classes, Event-GRU-ODE and Event-GRU-Discretized, that can predict the dynamic risk of events as a continuous trajectory across future time points, even without new observations, while updating these estimates as new data become available. Models were trained and validated for SRB prediction using a large electronic health record database. Both models demonstrated high discrimination (e.g., Event-GRU-ODE AUROC = 0.93, AUPRC = 0.10, relative risk = 13.4 at 95% specificity in a low-prevalence [0.15%] general cohort with a 1.5-year prediction window). This work provides an initial step toward developing novel suicide prevention strategies based on dynamic changes in risk.
format Article
id doaj-art-8fc19ecff35c4ed29332199df8d86028
institution DOAJ
issn 2398-6352
language English
publishDate 2025-03-01
publisher Nature Portfolio
record_format Article
series npj Digital Medicine
spelling doaj-art-8fc19ecff35c4ed29332199df8d860282025-08-20T02:56:17ZengNature Portfolionpj Digital Medicine2398-63522025-03-018111110.1038/s41746-025-01552-yContinuous time and dynamic suicide attempt risk prediction with neural ordinary differential equationsYi-han Sheu0Jaak Simm1Bo Wang2Hyunjoon Lee3Jordan W. Smoller4Center for Precision Psychiatry, Massachusetts General HospitalDepartment of Electrical Engineering, KU LeuvenCenter for Precision Psychiatry, Massachusetts General HospitalCenter for Precision Psychiatry, Massachusetts General HospitalCenter for Precision Psychiatry, Massachusetts General HospitalAbstract Current clinician-based and automated risk assessment methods treat the risk of suicide-related behaviors (SRBs) as static, while in actual clinical practice, SRB risk fluctuates over time. Here, we develop two closely related model classes, Event-GRU-ODE and Event-GRU-Discretized, that can predict the dynamic risk of events as a continuous trajectory across future time points, even without new observations, while updating these estimates as new data become available. Models were trained and validated for SRB prediction using a large electronic health record database. Both models demonstrated high discrimination (e.g., Event-GRU-ODE AUROC = 0.93, AUPRC = 0.10, relative risk = 13.4 at 95% specificity in a low-prevalence [0.15%] general cohort with a 1.5-year prediction window). This work provides an initial step toward developing novel suicide prevention strategies based on dynamic changes in risk.https://doi.org/10.1038/s41746-025-01552-y
spellingShingle Yi-han Sheu
Jaak Simm
Bo Wang
Hyunjoon Lee
Jordan W. Smoller
Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
npj Digital Medicine
title Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
title_full Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
title_fullStr Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
title_full_unstemmed Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
title_short Continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
title_sort continuous time and dynamic suicide attempt risk prediction with neural ordinary differential equations
url https://doi.org/10.1038/s41746-025-01552-y
work_keys_str_mv AT yihansheu continuoustimeanddynamicsuicideattemptriskpredictionwithneuralordinarydifferentialequations
AT jaaksimm continuoustimeanddynamicsuicideattemptriskpredictionwithneuralordinarydifferentialequations
AT bowang continuoustimeanddynamicsuicideattemptriskpredictionwithneuralordinarydifferentialequations
AT hyunjoonlee continuoustimeanddynamicsuicideattemptriskpredictionwithneuralordinarydifferentialequations
AT jordanwsmoller continuoustimeanddynamicsuicideattemptriskpredictionwithneuralordinarydifferentialequations