Assessing fatigue and sleep in chronic diseases using physiological signals from wearables: A pilot study

Figures

• 

  FIGURE 1. The study cycle. Participants went through a baseline visit at the beginning of the study, followed by four technology use periods (the circle) when they were instructed to use different sensors and report certain outcomes with a mobile phone application (example screenshot on the right). This study focuses on the participants using the depicted wearable device, VitalPatch (two alternative wear locations depicted).

• 

  FIGURE 2. Participant flow diagram. The number of participants (N) in total and included in different analyses, presented with the inclusion criteria (black-bordered box).

• 

  FIGURE 3. Average daily coverage of the cleaned digital measures across participants, presented by cohort.

• 

  FIGURE 4. Coverage (%) of PROs per participant group (left) and questionnaire timing (right).

• 

  FIGURE 5. The SD of responses across participant groups. Here, the PROs correspond to the feature aggregates normalized with the participant-mean L5 window parameters.

• 

  FIGURE 6. The participants’ average L5 parameter, presented by participant group, for (top) the mean signal values and (bottom) selected frequency domain HRV features. The figure covers 31 healthy participants, and 21 NDD and 49 IMID patients.

• 

  FIGURE 7. Repeated measures correlation r values between the 2-h feature aggregates and the corresponding PROs. Here, the 2-h feature aggregates have been normalized with the participant-mean L5 parameters. Statistically significant correlation results (p-value<0.05) are emphasized with black borders, other r values have faded annotation.

• 

  FIGURE 8. Repeated measures correlation r values between the 2-h feature aggregates and the corresponding PROs. Here, the 2-h feature aggregates have been normalized with the participants’ latest L5 window parameters. That is, as compared to Figure 7, the normalization parameters are not averaged over the full study periods. Statistically significant correlation results (p-value<0.05) are emphasized with black borders, other r values have faded annotation.

• 

  FIGURE 9. Heart rate recovery (left) in the three participant groups, including one representative value for 21 healthy participants, 18 NDD (9 HD, 10 PD), and 33 IMID patients (11 IBD, 8 PSS, 8 RA, 6 SLE). The adjusted p-values from post hoc analysis with Tukey’s method indicate the significance of the differences. The total duration of each accepted walk leading to a 1-min resting period is presented on the right.