Gesture recognition for Healthcare 4.0: a machine learning approach to reduce clinical infection risks
Computer vision to recognize hand-washing and reduce hospital infections, from a Healthcare 4.0 perspective.
Abstract
A study of an integrated system that leverages smart sensors and a digital hospital process-management system (based on flowcharts and BPMN models) for the active surveillance of colonization by antibiotic-resistant bacteria. At the core of the work is a computer-vision-based field sensor that uses machine learning algorithms to recognize hand-washing procedures in real time: hand joints are extracted frame by frame to derive geometric features, which are processed by three independent Random Forests whose outputs feed a final classifier (a simple artificial neural network). The system achieves an accuracy of 73.3%, an encouraging result given the complexity of the gestures analyzed, demonstrating how field sensors can enable new real-time management models in healthcare, in line with Healthcare 4.0 principles.
Authors
B. Lanza, E. Ferlinghetti, C. Nuzzi, L. Sani, A. Garinei, L. Maiorfi, S. Naso, E. Piccioni, F. Bianchi, M. Proietti, A. Marini, S. Speziali, M. Marconi, A. Vispa, M. Lancini
Link
Publication date: 06/06/2023 Last updated: 02/10/2026 17:20