Symptom Monitoring in Oncological Patients Using Wrist-Worn Wearables: A Machine Learning Approach.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Oncology

Background:

• Continuous psycho-physical symptom monitoring is crucial for enhancing cancer patients' quality of life.
• Traditional methods like questionnaires lack continuous monitoring capabilities.
• Wearable sensors and AI offer a novel approach for pervasive health data collection.

Purpose of the Study:

• To evaluate machine learning (ML) algorithms for predicting nine common cancer patient symptoms.
• To utilize physiological signals from wearable sensors and self-rated symptoms for prediction.
• To assess the feasibility of in-home, continuous symptom monitoring.

Main Methods:

• Extracted features from electrodermal activity (EDA), skin temperature (TEMP), and accelerometer (ACC) data.
• Applied principal component analysis (PCA) to reduce feature dimensionality.
• Trained and compared logistic regression (LogReg), support vector machine (SVM), and random forest (RF) algorithms.
• Employed a bootstrap approach for robust result evaluation.

Main Results:

• SVM and RF algorithms outperformed LogReg in predicting cancer patient symptoms.
• Tiredness was predicted with the highest F1-score (91.68% ± 2.81%) using SVM.
• Other symptoms including malaise, drowsiness, anxiety, appetite, nausea, and pain achieved F1-scores above 70%.

Conclusions:

• The integration of wearable devices and ML shows feasibility for continuous psycho-physical symptom monitoring in cancer patients.
• This approach has the potential to significantly improve cancer patient care through early intervention and personalized strategies.
• Further research with larger sample sizes and consideration of time-of-day effects is warranted.