Logo image
Back
Clinical notes: An untapped opportunity for improving risk prediction for hospitalization and emergency department visit during home health care
Journal article   Open access   Peer reviewed

Clinical notes: An untapped opportunity for improving risk prediction for hospitalization and emergency department visit during home health care

Jiyoun Song, Mollie Hobensack, Kathryn H. Bowles, Margaret V. McDonald, Kenrick Cato, Sarah Collins Rossetti, Sena Chae, Erin Kennedy, Yolanda Barrón, Sridevi Sridharan, …
Journal of biomedical informatics, Vol.128, pp.104039-104039
04/2022
DOI: 10.1016/j.jbi.2022.104039
PMCID: PMC9825202
PMID: 35231649
url
https://doi.org/10.1016/j.jbi.2022.104039View
Published (Version of record) Open Access

Abstract

[Display omitted] •Early identification of HHC patients at risk can prevent hospitalizations/ED visits.•Valid indicators for hospitalizations/ER visits were generated by NLP approaches.•Combining clinical notes with structured data improves risk prediction. Between 10 and 25% patients are hospitalized or visit emergency department (ED) during home healthcare (HHC). Given that up to 40% of these negative clinical outcomes are preventable, early and accurate prediction of hospitalization risk can be one strategy to prevent them. In recent years, machine learning-based predictive modeling has become widely used for building risk models. This study aimed to compare the predictive performance of four risk models built with various data sources for hospitalization and ED visits in HHC. Four risk models were built using different variables from two data sources: structured data (i.e., Outcome and Assessment Information Set (OASIS) and other assessment items from the electronic health record (EHR)) and unstructured narrative-free text clinical notes for patients who received HHC services from the largest non-profit HHC organization in New York between 2015 and 2017. Then, five machine learning algorithms (logistic regression, Random Forest, Bayesian network, support vector machine (SVM), and Naïve Bayes) were used on each risk model. Risk model performance was evaluated using the F-score and Precision-Recall Curve (PRC) area metrics. During the study period, 8373/86,823 (9.6%) HHC episodes resulted in hospitalization or ED visits. Among five machine learning algorithms on each model, the SVM showed the highest F-score (0.82), while the Random Forest showed the highest PRC area (0.864). Adding information extracted from clinical notes significantly improved the risk prediction ability by up to 16.6% in F-score and 17.8% in PRC. All models showed relatively good hospitalization or ED visit risk predictive performance in HHC. Information from clinical notes integrated with the structured data improved the ability to identify patients at risk for these emergent care events.
 Home Health Care Clinical deterioration Natural language processing Nursing informatics Predictive modeling Risk assessment

Details

Metrics

18 Record Views
32 Times Cited - Web of Science
Logo image