An interpretable data-driven approach to optimizing clinical fall risk assessment
6 affiliations
- 1Center for Health Systems and Policy Modeling, Department of Health Policy and Management, Johns Hopkins University, Baltimore, Maryland, United States of America
- 2Malone Center for Engineering in Healthcare, Johns Hopkins University, Baltimore, Maryland, United States of America
- 3Center for Systems Science and Engineering, Department of Civil and Systems Engineering, Johns Hopkins University, Baltimore, Maryland, United States of America
- 4Department of Physical Medicine and Rehabilitation, School of Medicine, Johns Hopkins University, Baltimore, Maryland, United States of America
- 5Department of Physical Therapy, University of Nevada, Las Vegas, Las Vegas, Nevada, United States of America
- 6Department of Nursing, The Johns Hopkins Hospital, Baltimore, Maryland, United States of America
Abstract
In this study, we aim to better align fall risk prediction from the Johns Hopkins Fall Risk Assessment Tool (JHFRAT) with clinical expert fall risk perception via a data-driven modelling approach. We conducted a retrospective cohort analysis of 54,209 inpatient admissions from three Johns Hopkins Health System hospitals between March 2022 and October 2023. In the absence of a true fall risk ground truth, we apply proxy labels based on clinician choices for the application of targeted preventative interventions, resulting in a total of 20,208 high-risk encounters and 13,941 low-risk encounters. We employed constrained score optimization (CSO) models to recalibrate the JHFRAT scoring weights, while preserving its additive structure and clinical thresholds. Recalibration refers to adjusting item weights so that the resulting score can order encounters more consistently by the study’s risk labels, and without changing the tool’s form factor or deployment workflow. The CSO model demonstrated significant improvements over the current JHFRAT in classification alignment with the proxy labels (CSO AUC-ROC = 0.91, JHFRAT AUC-ROC = 0.86). This model performance translates to a weekly average of an additional 35 Johns Hopkins Health System patients who are perceived as high risk (per our proxy labels) being classified by JHFRAT as high risk. The ablation analyses also suggest that the CSO model, though outperformed in prediction metrics by the benchmark black-box XGBoost model, is more robust than XGBoost to variations in risk labeling. Our evidence-based approach provides a robust foundation for understanding risk factor contributions to various indicators of clinician-perceived fall risk. Future research can build upon this foundation to improve risk assessment utility as a decision-support tool in clinical practice. Author summary: Inpatient falls are a major safety concern in hospitals, often leading to injuries and longer recovery times. To prevent these events, nurses use standardized tools like the Johns Hopkins Fall Risk Assessment Tool to identify vulnerable patients and provide extra care, such as more frequent check-ins or specialized equipment. However, these tools are often based on general clinical expertise rather than precise data, and their accuracy can be hidden because the very interventions they trigger prevent the falls they are trying to predict. We addressed this challenge by analyzing over 54,000 hospital admissions using a data-driven approach that accounts for the preventative care patients were assigned by nurses. By recalibrating the original assessment tool with advanced mathematical modeling, we significantly improved its alignment with clinician intuition of fall risk patients without making the tool more difficult for nurses to use. Analyzing fall risk assessment in this manner is a crucial first step to improving and standardizing clinical practice to simultaneously promote patient mobility and reduce the number of preventable falls.
Related
- EXPLAINABLE AI-DRIVEN CLINICAL DECISION SUPPORT SYSTEMS IN PRECISION ONCOLOGY: INTERPRETABLE MODELS FOR MULTIMODAL CANCER CARE
- An interpretable cell-centric representation of pancreatic ductal adenocarcinoma matches multiple-instance learning for clinical outcomes and leads to novel pathologist-driven histological biomarkers
- Improving Post‐Fall Assessment Documentation in Older Inpatients Using a Structured Falls Proforma: A Retrospective Cohort Study
- Clinical Decision Support for Fall Prevention in Primary Care : The eSTEPS Cluster Randomized Clinical Trial
- Artificial Intelligence-Driven Dietary Assessment Software: A Structured Narrative Review of Technological Advances, Persistent Measurement Error and Barriers to Clinical Translation
BibTeX
@article{ganjkhanloo2026interpretable,
title = {{An interpretable data-driven approach to optimizing clinical fall risk assessment}},
author = {Fardin Ganjkhanloo and Emmett Springer and Erik H. Hoyer and Daniel L. Young and Holley Farley and Kimia Ghobadi},
journal = {PLOS Digital Health},
year = {2026},
month = sep,
doi = {10.1371/journal.pdig.0001631},
url = {https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001631},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - An interpretable data-driven approach to optimizing clinical fall risk assessment
AU - Fardin Ganjkhanloo
AU - Emmett Springer
AU - Erik H. Hoyer
AU - Daniel L. Young
AU - Holley Farley
AU - Kimia Ghobadi
PY - 2026
DA - 2026/09/01
JO - PLOS Digital Health
DO - 10.1371/journal.pdig.0001631
UR - https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001631
AB - In this study, we aim to better align fall risk prediction from the Johns Hopkins Fall Risk Assessment Tool (JHFRAT) with clinical expert fall risk perception via a data-driven modelling approach. We conducted a retrospective cohort analysis of 54,209 inpatient admissions from three Johns Hopkins Health System hospitals between March 2022 and October 2023. In the absence of a true fall risk ground truth, we apply proxy labels based on clinician choices for the application of targeted preventative interventions, resulting in a total of 20,208 high-risk encounters and 13,941 low-risk encounters. We employed constrained score optimization (CSO) models to recalibrate the JHFRAT scoring weights, while preserving its additive structure and clinical thresholds. Recalibration refers to adjusting item weights so that the resulting score can order encounters more consistently by the study’s risk labels, and without changing the tool’s form factor or deployment workflow. The CSO model demonstrated significant improvements over the current JHFRAT in classification alignment with the proxy labels (CSO AUC-ROC = 0.91, JHFRAT AUC-ROC = 0.86). This model performance translates to a weekly average of an additional 35 Johns Hopkins Health System patients who are perceived as high risk (per our proxy labels) being classified by JHFRAT as high risk. The ablation analyses also suggest that the CSO model, though outperformed in prediction metrics by the benchmark black-box XGBoost model, is more robust than XGBoost to variations in risk labeling. Our evidence-based approach provides a robust foundation for understanding risk factor contributions to various indicators of clinician-perceived fall risk. Future research can build upon this foundation to improve risk assessment utility as a decision-support tool in clinical practice. Author summary: Inpatient falls are a major safety concern in hospitals, often leading to injuries and longer recovery times. To prevent these events, nurses use standardized tools like the Johns Hopkins Fall Risk Assessment Tool to identify vulnerable patients and provide extra care, such as more frequent check-ins or specialized equipment. However, these tools are often based on general clinical expertise rather than precise data, and their accuracy can be hidden because the very interventions they trigger prevent the falls they are trying to predict. We addressed this challenge by analyzing over 54,000 hospital admissions using a data-driven approach that accounts for the preventative care patients were assigned by nurses. By recalibrating the original assessment tool with advanced mathematical modeling, we significantly improved its alignment with clinician intuition of fall risk patients without making the tool more difficult for nurses to use. Analyzing fall risk assessment in this manner is a crucial first step to improving and standardizing clinical practice to simultaneously promote patient mobility and reduce the number of preventable falls.
SN - 2767-3170
ER -