Smartphone Application for Automated Measurement of Capillary Refill Time (CRT)

This prospective, monocenter observational study aims to validate a smartphone application (CRTApp) that automatically measures capillary refill time (CRT) from a short video of a standardized finger compression. Sixty adult intensive care unit (ICU) patients will undergo a CRT measurement recorded with a professional Android smartphone. Clinical CRT will be determined from the video by three blinded observers (reference = median of the three measurements). The application will analyze the same video locally to provide an automated CRT measurement. Agreement between automated and clinical reference CRT will be assessed, and predefined subgroup analyses will explore performance across skin tone, ambient lighting, smartphone characteristics, and severity of peripheral hypoperfusion.

Capillary refill time (CRT) is a bedside indicator of peripheral perfusion, but visual assessment is prone to inter-observer variability. In this study, CRT is measured using a standardized method (10-second finger compression with a glass slide). A video of the fingertip is captured without audio; framing is strictly limited to the fingers. Files are pseudonymized at acquisition. Videos are transferred via USB to an AP-HP workstation and stored on an AP-HP server; no cloud or messaging is used, and videos are deleted from the smartphone after import. Automated CRT is computed locally from the video by CRTApp without network transmission and without retaining the video. Statistical analyses will include intraclass correlation coefficient and Bland-Altman analysis for agreement. Secondary analyses will evaluate agreement stratified by skin tone (Monk scale / ITA), ambient lighting (lux), smartphone characteristics, and severity of peripheral hypoperfusion.