JMIR Publications

ABSTRACT

Background:

Inpatient falls are a significant concern for healthcare providers and are associated with negative outcomes for patients. Automated detection of falls using machine learning algorithms may aid in improving patient safety and reducing the occurrence of falls.

Objective:

This study aimed to develop and evaluate a machine learning algorithm for inpatient fall detection using Multidisciplinary Progress Record (MPR) notes and a pre-trained BERT language model.

Methods:

A cohort of 4,323 adult patients admitted to four acute care hospitals in Calgary, Alberta, Canada from 2016 to 2021 was randomly sampled. Trained reviewers determined falls from patient charts, which were linked to electronic medical records (EMR) and administrative data. The Bidirectional Encoder Representation from Transformers (BERT)-based language model was pre-trained on clinical notes, and a fall detection algorithm was developed based on a neural network binary classification architecture.

Results:

To address various usage scenarios, we developed three different AHN-BERT models: a high sensitivity model (sensitivity= 97.7, IQR: 87.7- 99.9), a high PPV model (PPV=85.7, IQR: 57.2-98.2), and the high F1 score model (F1=64.4). Our proposed method outperformed three classical machine learning algorithms and an International Classification of Diseases (ICD) code-based algorithm for fall detection, showing its potential for improved performance in diverse clinical settings.

Conclusions:

The developed algorithm provides an automated and accurate method for inpatient fall detection using MPR notes and a pre-trained BERT language model. This method could be implemented in clinical practice to improve patient safety and reduce the occurrence of falls in hospitals.