Development and validation of a machine learning–based model for predicting fall-related injury risk in hospitalized patients

Clicks: 14
ID: 326593
2026
Article Quality & Performance Metrics
Overall Quality
Not rated
Combines reader engagement with the AI quality analysis. This article has not been analysed, so there is no overall score — reader engagement is measured and shown alongside.
AI Quality Assessment
Not analyzed
Readership in this journal
Emerging

Ranked #65 of 114 articles by views in International journal for quality in health care : journal of the International Society for Quality in Health Care

Most read Least read

Bar heights use a square-root scale.

Mint this article as an NFT
Not yet minted

Create a permanent, verifiable on-chain record of this article on the Scimatic Network. The NFT is held in your Journament account, and you can withdraw it to your own wallet at any time.

5 SUSD one-off · no wallet required
Abstract
Abstract Background Falls are among the most common adverse events in hospitalized patients, with about 30% leading to injury. We developed a machine learning model to predict which in-hospital fall events would lead to patient injury, thereby supporting post-fall risk stratification. Methods We retrospectively analyzed data from 410 patients who had experienced falls at a tertiary general hospital in China. Among them, 134 patients (32.7%) had fall-related injuries. The dataset was divided into training and test sets at a 7:3 ratio by outcome-stratified random sampling. Least absolute shrinkage and selection operator regression was used for feature selection to identify relevant predictors. Four machine learning models—logistic regression, random forest, extreme gradient boosting, and support vector machine—were developed and assessed. Model performance was evaluated in the test set using the area under the receiver operating characteristic curve, Brier score, and calibration curves. According to model performance, the optimal model was selected, and multivariable logistic regression analysis was then performed to determine independent risk factors. Results In the test cohort, the logistic regression model showed the strongest predictive ability (AUC = 0.863, 95% CI: 0.785–0.941; Brier score = 0.142). Five independent risk factors were detected, including impaired consciousness (OR = 3.35, 95% CI: 1.58–7.10), reduced muscle strength (OR = 3.93, 95% CI: 1.87–8.26), use of high-risk medications (OR = 10.07, 95% CI: 5.05–20.07), ward-related environmental hazards (OR = 3.30, 95% CI: 1.63–6.69), and hypocalcemia (OR = 2.10, 95% CI: 1.05–4.19). Based on this model, a nomogram was developed, and decision curve analysis indicated a positive net clinical benefit within the threshold probability range of 0.10–0.80. Conclusion A prediction model based on five routinely collected clinical variables was developed to estimate fall-related injury risk after an in-hospital fall. The logistic regression model showed a favorable balance among predictive performance, calibration, interpretability, and clinical feasibility. This model may help support post-fall injury risk stratification, triage for further assessment, and monitoring decisions in hospitalized patients who have already experienced a fall.
Reference Key
openalex_W7204554574 Use this key to autocite in the manuscript while using SciMatic Manuscript Manager or Thesis Manager
Authors Shiyu Wang, Xiaomei Zhang, Yi Qin, Xiuqun Xu, Linjing Du, Panpan Xu, Jiahui Yu, Yanqing Li, Lingyan Yang
Journal International journal for quality in health care : journal of the International Society for Quality in Health Care
Year 2026
DOI
10.1093/intqhc/mzag128
URL
Keywords Keywords not found

Citations

No citations found. To add a citation, contact the admin at info@scimatic.org

No comments yet. Be the first to comment on this article.