Criterion Validity of a Consumer Wearable for Step Counting and Activity Intensity Classification in Adults With Lung...

Area of Science:

• Exercise Oncology
• Biomedical Engineering
• Wearable Technology Validation

Background:

• Consumer wearable activity monitors are increasingly used in exercise-oncology trials.
• Their accuracy is unvalidated in lung cancer patients, where gait may challenge algorithms.

Purpose of the Study:

• Assess Fitbit Charge 6 validity against direct observation in lung cancer patients.
• Evaluate step-count agreement across durations and speeds.
• Compare active/sedentary minute classification and spurious step detection.

Main Methods:

• 14 lung cancer adults participated in an in-laboratory validation study.
• Fitbit Charge 6 worn during walking trials (variable duration/speed) and nonwalking tasks.
• Bland-Altman analysis, MAPE, ICCs, and confusion matrix used for validation.

Main Results:

• Fitbit undercounted steps, with poor relative agreement and low ICC.
• Undercounting was greatest at slow gait speeds (<0.6 m/s) and short bouts (5s).
• High sensitivity but low specificity for activity intensity misclassified sedentary minutes as active.

Conclusions:

• Fitbit Charge 6 shows meaningful error during short bouts and slow speeds common in lung cancer patients.
• Systematic overestimation of active minutes due to low specificity.
• Findings impact interpretation of exercise-oncology interventions using wearable data.