Contact-Free Monitoring of Physiological Parameters in People With Profound Intellectual and Multiple Disabilities

Gasper Slapnicar, Erik Dovgan, Pia Cuk, Mitja Lustrek; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 0-0

This paper presents a contact-free method for physiological parameter estimation in people with profound intellectual and multiple disabilities (PIMD). We used an existing state-of-the-art algorithm Plane-Orthogonal-to-Skin (POS) in order to obtain an initial remote photoplethysmogram (rPPG) reconstruction from facial videos. We enhanced this signal by applying a long-short-term-memory (LSTM) neural network to the initial PPG reconstruction. Evaluation of our method on a public database DEAP has shown heart rate (HR) error of 8.09 beats-per-minute, suprpassing the state-of-the-art POS algorithm implementation, which had error of 13.36 BPM. More importantly, a good correlation between our predictions and ground-truth HRs has been observed. The method is currently being implemented as part of a system which aims to monitor people with PIMD in real time in order to obtain information about their physiological and psychological state and in turn increase their quality of life.