JMIR Publications

ABSTRACT

Background:

Wearable fitness trackers have become a valuable tool in public health research due to their ability to collect large-scale, individual-level data at a low cost. However, their use has been largely limited to high-income settings, and a major challenge remains: high rates of missing data. These gaps may be exacerbated in low-resource environments, where logistical and operational barriers further complicate data collection.

Objective:

The purpose of this study is to characterize the patterns and frequency of missing data, and to determine the underlying missingness mechanism.

Methods:

For this study, 300 women in slum communities of Kampala, Uganda were equipped with Garmin smartwatches that collected continuous data over five days. Approximately 30% of nighttime data was missing. The nature of missingness was assessed through four methods: pattern analysis, Little’s test, a random forest classification model, and a logistic regression classification model.

Results:

Approximately 30% of nighttime data were missing. Three missingness patterns were identified that occurred in over 10% of the nights of data. Additional missingness patterns occurred in fewer nights of data. Pattern analysis and Little’s test indicated that the data were not missing completely at random (MCAR). Both the random forest (AUC=0.7) and logistic regression models suggested that the data were missing at random (MAR).

Conclusions:

Based on the evidence provided by the classification models and the likelihood that device and battery failure contributed considerably to missingness, it was concluded that the missingness was consistent with MAR. Potential causes of missingness include device removal, battery failure, and technical malfunctions. These findings have important implications for both wearable device users and future research. Understanding the mechanisms behind missing data can inform strategies to improve compliance and data quality, particularly in low-resource settings.