Researchers from the University of California, San Francisco developed a machine-learning model trained on patient-generated health data from wearable fitness trackers. In a trial in patients before and during chemoradiation therapy (CRT), the model demonstrated it could predict hospitalizations based on daily step counts, they wrote.
"The step counts immediately preceding the prediction window ended up being generally more predictive than clinical variables," said senior author Dr. Julian Hong in a news release from ASTRO. "The dynamic nature of the step counts, the fact that they're changing every day, seems to make them a particularly good indicator of a patient's health status."
Estimates suggest up to 20% of patients who receive outpatient radiation or chemoradiation therapy will need acute care, and these unplanned hospitalizations can cause treatment interruptions that impact clinical outcomes. Early identification and intervention for patients at higher risk of complications can prevent these events, according to Hong and colleagues.
Previously, the researchers demonstrated that a machine-learning algorithm using data such as cancer history and treatment plan could identify patients at higher risk of emergency department visits during cancer treatment and that additional surveillance from their providers can reduce acute care rates for these patients.
In this study, the group collected data from 214 patients in three previous clinical trials during which they wore commercial fitness trackers continuously before and during curative-intent CRT for multiple cancer types. The team integrated step counts and other data to develop a logistic regression model to predict the likelihood that a patient would be hospitalized in the next week, based on their previous two weeks of data.
Hong and colleagues trained the model using data on 70% of the patients with and without step count-derived features and then validated the model using the remaining 30% of patients (63 people).
According to the findings, the model that integrated step counts was strongly predictive of hospitalization the following week (area under the curve [AUC] = 0.8), and it significantly outperformed the model without step counts (AUC = 0.46), the researchers said.
Specifically, the top predictive variables in the model included step counts from each of the past two days, as well as the relative changes in maximum step count and step count range over the past two weeks.
"One of the unique parts of this model is that it's designed to be a running prediction," said lead author Isabel Friesner, a clinical data scientist at UCSF. "You can run the algorithm on any given day and have an idea of a patient's risk level one week out, giving you time to provide that additional support they need."
Ultimately, the approach could help reduce hospitalizations, perhaps by indicating the need for more frequent follow-ups or a change in the patient's treatment plan, the researchers suggested. Specific next steps will include clinical validation of the model in patients undergoing CRT for lung cancer with or without daily step count monitoring.
"As more people begin to use wearables, the question of whether the data they are collecting could be useful arises. Our study shows there is value in having our patients collect their own health data during their everyday lives, and that we can use this data to then monitor and predict their health status," Friesner said.
Copyright © 2022 AuntMinnie.com