Movement quantification in preterm infants: Comparing motion extraction from ECG signals and from pressure sensitive fiber optics mat

Monitoring motion in preterm infants is crucial to evaluating their clinical condition and development. However, no quantitative automated method for continuous motion monitoring is currently clinically available, though innovative methods based on camera or mattress-based sensing are being developed. The aim of this study is to assess the use of ECG-based motion detection methods in NI CU settings. First, we compared the motion derived from the ECG with the manual annotations from video recordings. Then, the ECG-based movement quantification is compared with a fiber mat-based motion detection. The comparison between motion detection based on ECG and the movement quantification based on the fiber mat is made to assess the use of ECG and fiber mat information combined to overcome the limitations of cameras. Comparison of the ECG-based method to annotated video demonstrates a mean (SD) AUC of 0.56 (0.10). Comparison of the ECG-based method to fiber mat motion detection shows a mean (SD) AUC 0.73 (0.18). We observed that the correspondence between the ECG-based method and the manual annotations of video recordings is not high, but it has to do with limitations in both ECG and camera motion detection. Using the fiber mat in addition is likely to give a better capture of movements. In conclusion, the combined use of ECG and optical signals offers the possibility to perform automatic motion measurements lasting days or weeks without requiring time-consuming manual motion annotation.