A recent study published by VivoSense Senior Data Scientist Yaya Zhai PhD, in the Karger Digital Biomarkers journal, demonstrated how everyday activity tracking through consumer wearables can provide valuable insights into cancer patient outcomes by offering a methodological breakthrough while establishing a compelling link between daily step counts and clinical events in cancer patients.
How Wearable Tech is Transforming Cancer Care
Wearable technology is transforming cancer care by providing continuous, real-time data that captures patients’ daily physical activity outside clinical settings. Traditional cancer care relies heavily on infrequent clinical visits, leaving significant gaps in understanding patients’ daily function and early warning signs between appointments. Wearable technology allows for the identification of early warning signs of adverse events, the tracking of treatment side effects, and the monitoring of recovery progress, all while reducing patient burden by enabling remote monitoring. Ultimately, these digital biomarkers have the potential to personalize treatment plans, optimize clinical interventions, and improve patient outcomes by integrating seamlessly into their daily lives.
The Challenge: Gaps in Data from Consumer Wearables
There are inevitable gaps in data collection with consumer wearables, specifically the Apple Watch, which poses challenge for data analysis. The gaps are caused by various factors like sedentary behaviors, device removal or technical glitches. This is particularly relevant in oncology, where patients may experience periods of severe fatigue or treatment side effects that lead to elongated sedentary time or interrupt consistent device usage.
Key Findings: Making Sense of the Data Gaps
Our research developed an innovative framework to estimate daily step counts from consumer-grade wearable devices, addressing a critical gap in handling missing data due to device non-wear or inactivity. We introduced a robust preprocessing technique (Method 3) that categorizes data gaps based on duration and context, distinguishing between periods of inactivity, non-wear time, and sleep. This method yielded more reliable step count estimates comparable to research-grade sensors in previous studies, averaging about 5,000 steps daily among cancer patients.
The results were compelling:
• Predictive Power of Daily Steps:
o Daily step counts on days with sufficient wear time (10+ waking hours) were a strong predictor of time to first clinical event, with higher step counts correlating with reduced hazard of adverse clinical events and mortality.
o Patients taking more than 2,510 steps daily were less likely to experience adverse outcomes, including emergency department visits, hospitalizations, or referrals to palliative care.
o Natural participant clusters emerged based on step patterns, with the most active group showing the lowest hazard probability for death or clinical events.
• Data Quality Matters: Comparisons between raw and preprocessed step data highlighted the importance of robust data cleaning—unprocessed data failed to predict clinical events, emphasizing the need for tailored analysis methods.
Implications for Cancer Care
This work underscores the feasibility of integrating consumer wearables into oncology research, offering a patient-centered, real-time view of physical activity that holds promise for informing clinical decision-making. As healthcare becomes increasingly decentralized, this means a shift towards a more proactive and personalized approach to cancer care, with these digital biomarkers acting as early warning systems for clinicians, potentially allowing for timely interventions before a patient’s condition deteriorates.
Looking Ahead: Transforming Clinical Trials
Our research redefines clinical trials by leveraging wearable technology to collect real-world data, focusing on accurately estimating daily step counts while simultaneously assessing data quality. This approach shifts from episodic, site-based assessments to continuous, remote monitoring, offering a clearer, more reliable picture of patient health and its relationship to clinical events. By integrating digital biomarkers, clinical trials become more patient-centered and data-driven, reducing clinic visits while enhancing both accessibility and analytical precision. These advancements hold the potential to accelerate drug development, tailor interventions in real-time, and ultimately improve patient outcomes.
Join us in pushing the boundaries of digital health research—because every step counts.
To explore the detailed insights, read the full article here