BabyFM Sensor Displacement Detection System: Performance Analysis with Clinical Dataset

This is the authors' abstract. We don't add key points for this paper.

This paper presents a technical validation with clinical datasets of the Sensor Displacement Detection (SDD) framework for BabyFM, a wearable medical device for continuous axillary temperature monitoring, designed primarily for pediatric and immunocompromised patients; the clinical trial was conducted in adult patients. The study uses two distinct clinical datasets: a 15-patient feasibility cohort (14 analyzed, 7201 valid sensor readings, 97.6 h) used for post-hoc verification of the predefined SDD thresholds, and a 22-patient validation cohort (184,361 sensor readings over 500.05 cumulative hours) used to evaluate the SDD algorithm. The SDD algorithm applies a composite rule—temperature below 35.5 ∘ C combined with a rate of temperature change exceeding |dT/dt| > 1.2 ∘ C/min—to flag displacement artifacts. Across the 22-patient cohort, 2110 displacement triggers were detected, corresponding to a weighted SDD trigger rate of 1.14% (unweighted per-patient mean 1.12%, range 0.00−4.75%), consistent with the prior theoretical displacement-rate prediction of 1.4%; 98.86% of readings remained valid. Consecutive triggers were clustered into 397 discrete displacement episodes (2-min gap) or 236 clinically relevant episodes (5-min gap). The predefined thresholds were independently verified in the 15-patient feasibility cohort (136 triggers; weighted rate 1.87%; range 0.00−4.09%). Sensor accuracy was independently established in the BABYFM-010 trial (658 paired measurements against a gallium-in-glass reference: mean difference −0.09 ∘ C; 80.5% within ± 0.5 ∘ C; Pearson r = 0.830). The contribution lies in the parameterization and clinical evaluation of a rate-and-temperature artifact-rejection strategy for continuous axillary thermometry rather than in the threshold rule itself. The main limitations are the single-center, single-arm design without a randomized control group, and the adult-only cohort, which call for independent multi-center validation and dedicated studies in the intended pediatric and immunocompromised populations.