Estimation of Symptom Severity During Chemotherapy From Passively Sensed Data: Exploratory Study

Area of Science:

• Oncology
• Digital Health
• Wearable Technology

Background:

• Chemotherapy often causes physical and psychological symptoms in cancer patients.
• Real-time symptom monitoring can enhance patient outcomes.
• Mobile phone sensors and wearable activity trackers offer passive monitoring solutions with low patient burden.

Purpose of the Study:

• To investigate if passively collected mobile phone and Fitbit data can estimate daily symptom burden during chemotherapy.
• To assess the feasibility of using wearable sensors for remote symptom tracking in oncology patients.

Main Methods:

• 14 gastrointestinal cancer patients undergoing chemotherapy participated in a 4-week study.
• Participants used Android phones and Fitbit devices, completing daily symptom severity ratings.
• Data included mobility, activity, sleep, phone usage, and communication patterns, analyzed using rotation random forests with feature selection.

Main Results:

• Mobile phone and Fitbit data correlated with patient-reported symptom burden.
• A population model achieved 88.1% accuracy, with key predictors including sedentary behavior, reduced physical activity, and increased phone screen time.
• Mobile phone data demonstrated higher predictive power than Fitbit data.

Conclusions:

• Passive sensor data from mobile phones and wearables are associated with daily symptom burden in chemotherapy patients.
• This approach supports continuous remote monitoring and facilitates early, adaptive interventions for symptom management.
• Minimal patient burden and potential for improved cancer care outcomes are highlighted.