Let's get Physiqual: Combining electronic diary data with commercial sensor data.

Electronic diary questionnaire data can give insight into the day-to-day life of an individual. Unfortunately, such data only comprises subjective, self-report data. Combining questionnaire data with objectively measured sensor data can greatly enrich a dataset. We present Physiqual, a platform that allows for combining commercial sensor data with diary data.

Personalized healthcare is gaining attention in several fields of healthcare, including the field of psychiatry. Currently, most psychiatric research is still based on averages drawn from population samples, even though evidence suggests that such averages might not always hold for the individual. Electronic diary data (or Ecological Momentary Assessments, EMA) shows promise in shifting our focus towards the individual in the field of psychiatry. EMA can be used to provide insight into the day-to-day fluctuations of various psychological and physiological variables of a person over time.

As with all questionnaire data, data retrieved from EMA studies can be subjective. Furthermore, due to the number of questionnaires and questions, completing an EMA study can be cumbersome. In order to ease the collection of data and enrich data obtained through EMA studies, we have designed Physiqual, a platform that allows for combining diary study data with objectively measured data from commercially available sensors, such as provided by Fitbit, Google, or Jawbone.

Currently Physiqual can offer data from two service providers, viz. Google Fit and Fitbit. From these platforms we expose five variables, that is, steps, sleep, distance, calories expended and heart rate. We automatically apply transformations to each of the variables, making the data useful for combination with diary study data.

Besides providing access to the sensor data, Physiqual can provide aggregated and resampled datasets. Physiqual automatically serves data in a format that can easily be combined with existing diary study designs. That is, it resamples the data in a way similar to the diary study protocol, and exposes the separate variables in a format useful for the researcher. Furthermore, missing data can be imputed automatically by one of the various imputation algorithms available.

To demonstrate how Physiqual works, we will present a two-subject case study. The subjects participated in a thirty day, tri-daily EMA study for collecting psychological data, while wearing a wearable device or smartwatch. We analyzed the combined dataset using a statistical method known as vector autoregression, to show the temporal relations between the variables.

We expect sensor quality and wearable data to improve significantly in the near future. Physiqual automatically supports this wearable evolution, and allows researchers to take advantage of the best commercial wearable technology available.