Calibrate step detection thresholds for ME/CFS and Long COVID

Determine and validate the sliding‑window length and peak‑threshold parameters of the local variance step‑detection algorithm used to compute Steps/Day from ankle‑mounted accelerometer data in myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and Long COVID populations, since it is currently not certain whether the thresholds calibrated on healthy controls yield accurate step counts for these patient cohorts.

Background

The paper computes Steps/Day using a local variance method that detects peaks in accelerometer data exceeding a threshold within a sliding window. Threshold parameters were tuned using data from healthy controls, and the same settings were applied to ME/CFS and Long COVID participants without a dedicated patient-specific calibration.

The authors note that the resulting step counts appear unrealistically large and explicitly acknowledge that, without a separate calibration and validation paper for ME/CFS and Long COVID, they cannot be certain the parameters are appropriate for these populations, motivating the need to establish validated thresholds.