Table 4 Recommendations for reporting in studies using passive data, based on the findings from this review
From: A systematic review of passive data for remote monitoring in psychosis and schizophrenia
Study and participants | How many participants were in the study? How long was passive data collected for? How many participants who entered the study did not complete the full period of passive data collection? Is this study using newly collected data or using an existing dataset? (e.g. CrossCheck) |
Devices | What devices were used, including manufacturer and model? What instructions were given to participants about wear? (e.g. location of device, any removal periods) What sensors did the devices use to collect data? What device settings were selected? (e.g. epoch length, if applicable) |
Data pre-processing | Was data extracted raw, or processed by device algorithm? What pre-processing was done? (e.g. removing outliers, splitting data into time windows, identifying and removing (or replacing) non-wear time) If all data recorded was not analysed, how was data sampled or selected for analysis? |
Data quality | Was there any assessment of data quality? (e.g. whether sampling rate was as expected) Was a threshold of quantity of data applied? (e.g. a minimum of 10 h of data per day) If so, how much data was included/excluded for each participant? |
Feature extraction | What specific features were extracted from raw data? Including their unit of measurement, data channels contributing to them, and the time window (e.g. number of steps per day, or number of outgoing phone calls per hour) How were these extracted from raw data? Were any open-source algorithms used? |
Analysis methods | How was missing data handled? How were repeated observations utilised? (e.g. were they summarised over hours, days or weeks?) Was any feature or variable selection performed? |
