Official websites use .gov

A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS

A lock ( 🔒 ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Need Help

Precision Risk Modeling to Reduce Opioid Overdose Deaths among Individuals Experiencing Homelessness

Award Number: R21DA065889
ORGANIZATION: NATIONAL INSTITUTE ON DRUG ABUSE
OPDIV: NIH
AWARD CLASS: DISCRETIONARY
AWARD ACTIVITY TYPE: SCIENTIFIC/HEALTH RESEARCH (INCLUDES SURVEYS)
PERIOD OF PERFORMANCE START DATE: 05/15/2026
PERIOD OF PERFORMANCE END DATE: 04/30/2028

Group Awards By:

View Award Description

Precision Risk Modeling to Reduce Opioid Overdose Deaths among Individuals Experiencing Homelessness - Project Summary/Abstract Each year, over 1.25 million people experience homelessness in the US. An estimated 7-16% of these individuals have opioid use disorder (OUD), as compared to less than 1% in the non-homeless population. Drug overdose is the leading cause of death among homeless adults, accounting for 1 in 4 deaths at rates up to 30 times higher than in the general population. Opioids contribute to over 90% of these overdose deaths. Despite the effectiveness of medications for OUD, adherence remains suboptimal in this population, with high rates of early disengagement and elevated mortality risk after discontinuing treatment. Although numerous opioid overdose risk models have been developed in recent years, none have been tailored to or validated in homeless populations. These models also fail to incorporate the dynamic, time-varying treatment patterns and social determinants that drive risk in this group. This project aims to fill these critical gaps by leveraging 20 years of clinical and mortality data on a large cohort of adults experiencing homelessness in Boston to: 1) derive and validate a prediction model for a) opioid overdose mortality and b) all-cause mortality, and 2) characterize 6-month OUD treatment trajectories and examine their association with subsequent a) opioid overdose mortality and b) all-cause mortality. To accomplish Aim 1, we will use an ensemble of 5 supervised machine learning models – regularized logistic regression, extreme gradient boosting, support vector machines, random forest, and a fully connected neural network – to identify predictors of opioid overdose mortality and all-cause mortality among individuals with clinically diagnosed OUD. To accomplish Aim 2, we will use group-based trajectory modeling to identify latent classes of treatment trajectories among adults with OUD over the first 6 months of establishing care based on 3 clinical parameters: buprenorphine adherence, opioid abstinence, and appointment attendance. We will then examine the association between these treatment trajectory classes and subsequent opioid overdose mortality and all-cause mortality over the ensuing 12 months. Both aims will be supported by an existing dataset that includes 71,914 adults (including 10,663 with OUD) seen at Boston Health Care for the Homeless Program (BHCHP) between 2003-2022, linked clinical data on these individuals from the BHCHP electronic data warehouse, and linked Massachusetts death records for this cohort over the same years (revealing 2,769 opioid overdose deaths and 11,605 all-cause deaths). This work will identify individuals at highest risk of opioid overdose death, elucidate how and when risk evolves, and inform real-time clinical decision support tools to reduce preventable deaths in this structurally vulnerable population. Results will provide a foundation for precision prevention efforts in homeless health care systems and offer a scalable analytic framework for future implementation and external validation.


Issue Date FY	Funding FY	Legal Entity Name	Legal Entity Address	Legal Entity City	Legal Entity State	Legal Entity Zip Code	Legal Entity COUNTY	Legal Entity COUNTRY	Assistance Listing	Award Code	Budget Year	Action Date	Action Type	Action Amount

Issue Date FY: 2026 ( Subtotal = $476,808 )
2026	2026	THE GENERAL HOSPITAL CORPORATION	55 FRUIT ST	BOSTON	MA	02114	SUFFOLK	USA	Drug Use and Addiction Research Programs	000	1	5/20/2026	NEW	$476,808
2026	2026	THE GENERAL HOSPITAL CORPORATION	55 FRUIT ST	BOSTON	MA	02114	SUFFOLK	USA	Drug Use and Addiction Research Programs	001	1	6/25/2026	NEW	$0
														Subtotal = $476,808

Grand Total All Awards = $476,808

Top

All Categories

About

Search

Reports

Data Submission

Award Information

Precision Risk Modeling to Reduce Opioid Overdose Deaths among Individuals Experiencing Homelessness

Award Number: R21DA065889

ORGANIZATION: NATIONAL INSTITUTE ON DRUG ABUSE

OPDIV: NIH

AWARD CLASS: DISCRETIONARY

AWARD ACTIVITY TYPE: SCIENTIFIC/HEALTH RESEARCH (INCLUDES SURVEYS)

PERIOD OF PERFORMANCE START DATE: 05/15/2026

PERIOD OF PERFORMANCE END DATE: 04/30/2028

Federal Websites

Department of Health & Human Services

HHS Operating Divisions

HHS Staff Divisions

Download A Document Viewer