# Using machine learning to analyze integrated clinical and geo-social data to inform the design of a decision support system to reduce the risk of stimulant therapy

> **NIH NIH K01** · PENNSYLVANIA STATE UNIV HERSHEY MED CTR · 2024 · $180,896

## Abstract

PROJECT SUMMARY/ABSTRACT
Stimulant use disorder (STUD) and overdose fatalities are devastating public health problems, straining
healthcare and criminal justice systems, and societal productivity. Annually, approximately 55% of stimulant
prescriptions have been prescribed to adults, with nearly one-third of them reporting stimulant misuse or
stimulant-related harm. Despite growing concern that stimulant medications may increase the risk of STUD
and overdose in some adults, risk factors for stimulant-related harm have not been well understood, and no
screening tool is available for assessing the risk/benefit of stimulant therapy at the point of care. This project
hypothesizes that machine learning (ML) techniques, applied to large clinical (Aim 1) and linked census (Aim 2)
datasets, can help identify personalized risk factors for STUD and overdose among adults treated with
stimulants, and assist with the development of clinical decision support system (CDSS) tool (Aim 3), which, in
turn, can help guide clinical decision making when considering stimulant therapy and decrease stimulant-
related risk of harm. To evaluate potential demographic and clinical risk factors for STUD and overdose,
retrospective longitudinal ML-based analysis of electronic health records (EHR) from the TriNetX research
network will be conducted (Aim 1). Because EHRs do not include characteristics of neighborhoods where
patients reside, and social determinants of health (SDOH) can contribute to STUD and overdose risk, the
predictive model from Aim 1 will be further enhanced by geocoding and linking the local health system’s EHR
to the census block group level neighborhood data for each patient to account for the potential impact of SDOH
and enhance the accuracy of the STUD and overdose risk modeling (Aim 2). Risk factors identified through
Aim 1 and Aim 2 analysis, along with prescribing-clinician input, will enable the development of a CDSS tool to
aid clinicians with risk/benefit assessment of stimulant therapy at the point-of-care, based on unique
demographic, clinical and neighborhood characteristics of each patient (Aim 3). The CDSS has been
increasingly utilized in digitally-driven healthcare to improve treatment safety and outcomes, but their
applications toward improving care for adults treated with stimulants or to mitigate addiction and overdose risks
have been lagging. The proposed research leveraging big data and ML has enormous potential for
understanding and predicting health risks of stimulant therapy in a personalized way, and will lay foundation for
future clinical trials evaluating the efficacy of CDSS-based intervention on reducing the risk, and, ultimately,
improving health of adults considered for stimulant treatment. This mentored research scientist development
award (K01) will also enable the PI to pursue additional advanced training in relevant clinical knowledge,
analytical skills, grantsmanship, leadership, and diversity, equity, and inclusiveness to...

## Key facts

- **NIH application ID:** 10984771
- **Project number:** 1K01DA059606-01A1
- **Recipient organization:** PENNSYLVANIA STATE UNIV HERSHEY MED CTR
- **Principal Investigator:** Wen-Jan Tuan
- **Activity code:** K01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $180,896
- **Award type:** 1
- **Project period:** 2024-07-15 → 2029-06-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10984771

## Citation

> US National Institutes of Health, RePORTER application 10984771, Using machine learning to analyze integrated clinical and geo-social data to inform the design of a decision support system to reduce the risk of stimulant therapy (1K01DA059606-01A1). Retrieved via AI Analytics 2026-05-27 from https://api.ai-analytics.org/grant/nih/10984771. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*