Prediction of activity mode with global positioning system and accelerometer data.

Area of Science:

• Wearable technology
• Physical activity recognition
• Biomedical engineering

Background:

• Accurate physical activity monitoring is crucial for health research.
• Distinguishing between different activity modes presents a challenge for current wearable sensors.
• Integrating multiple sensor types may improve classification accuracy.

Purpose of the Study:

• To evaluate the efficacy of combining global positioning system (GPS) and accelerometer data for predicting physical activity modes.
• To determine the optimal combination of sensor-derived variables for accurate activity classification.

Main Methods:

• Ten adults simultaneously wore GPS units and accelerometers during various activities.
• Discriminant function analysis was employed to classify activity modes based on sensor data.
• Data were analyzed using both calibration and validation datasets.

Main Results:

• Accelerometer data alone (median counts and steps) correctly classified 90% of activity bouts.
• Combining accelerometer data with GPS speed improved classification accuracy to 93% in calibration data.
• In validation datasets, the combined approach achieved 91-98% accuracy for classifying activity modes, with walking and bicycling being most accurately identified.

Conclusions:

• The integration of GPS data with accelerometer monitoring offers a modest improvement in physical activity mode classification.
• Further research with larger sample sizes and a wider range of activities is recommended to validate these findings.
• This combined sensor approach shows promise for enhanced physical activity identification in free-living settings.