Comparative assessment of multimodal sensor data quality collected using Android and iOS smartphones in real-world settings

Abstract Healthcare researchers are increasingly utilizing smartphone sensor data as a scalable and cost-effective approach to studying individualized health-related behaviors in real-world settings. However, to develop reliable and robust digital behavioral signatures that may help predict disease trajectory early and future prognosis, there is a critical need to quantify potential technical variability that may be present in underlying sensor data due to variations in smartphone hardware and software used by large populations. Using sensor data collected in real-world settings from 3000 participants' smartphones for up to 84 days, we compared differences in completeness, correctness, and consistency of the three most common smartphone sensors — accelerometer, gyroscope, and GPS across Android and iOS devices. Our findings show considerable variation in sensor data quality within and across Android and iOS devices. Sensor data from iOS devices showed significantly lower levels of anomalous point density (APD) compared to Android across all sensors (p < 1e-4). iOS devices showed a considerably lower missing data ratio (MDR) for the accelerometer compared to the GPS data (p < 1e-4). Notably, the device type could be predicted with up to 0.98 accuracy 95% CI [0.977, 0.982] using the quality features derived from raw sensor data across devices. With a high degree of heterogeneity in smartphone devices, including iOS and Android, it's crucial to comprehend, measure, and address technical variations in underlying sensor data for creating robust, representative, and clinically actionable digital behavioral signatures linked to health outcomes.