Tracking Daily Step Counts May Predict Hospital Admissions During Cancer Treatment

Tracking daily step counts may predict, and help prevent, hospital admissions for acute events among patients receiving chemoradiotherapy for cancer.

Physical Activity: One of the Most Important Indicators of Health

Step-count data from a wearable device were shown to predict which patients would be admitted to hospital in subsequent weeks.

Author Julian Hong, from Radiation Oncology and Health Sciences at the University of California (UCSF), said: “If you can predict a patient's risk of an unplanned hospital admission, you can modify cancer treatment and reduce the likelihood of emergency department visits or hospitalisation.”

The research was presented on 24 October at the American Society for Radiation Oncology (ASTRO) 2022 Annual Meeting.

Izzy Friesner, a clinical data scientist at UCSF, said: “As more people start using wearable devices, the question is whether the data collected can be useful. Our study shows that it is valuable for patients to collect their own health data during everyday life and that we can then use these data to monitor and predict their health status.”

“The advantage of using data from a patient's own device is that it can be used outside a clinical setting, at home or during daily life. However, one challenge of working with real-world wearable data is the economic and racial differences affecting ownership of such devices.” she added.

Study Details

For this research, the team collected data on 124 cancer patients participating in three prospective clinical studies.

The study included patients with various cancers, including head and neck cancer (30%) and lung cancer (29%), who wore activity trackers for several weeks.

The team randomly assigned 70% of patients to a training group and the remaining 30% to a holdout group. In addition to step count, the information covered age, ECOG performance status, cancer type and clinical characteristics, providing extensive information on potential predictors of an acute event.

They calculated a 3-day average of daily step counts, defined a prediction window 1 week before the first hospital admission, then collected step-count data from 2 weeks earlier.

The research strongly predicted hospital admissions a week later.

Author Julian Hong said: “Step counts were more predictive than other commonly known clinical variables.”

“The dynamic nature of step counts, changing every day, appears to make them a good indicator of a patient's health.” he added.

The most predictive features for hospital admission were step counts in the previous 2 days and the relative change in the weekly maximum step count.

The team's model will now be tested in a subsequent trial that randomly assigns patients receiving chemoradiotherapy for lung cancer to treatment with or without daily step-count monitoring.