A novel infection screening method using a neural network and k-means clustering algorithm which can be applied for screening of unknown or unexpected infectious diseases

摘要

We recently reported in this journal on a novel mass screening method for influenza patients using a non-contact screening system.1 The method enabled the screening of infection within tens of seconds using non-contact monitored parameters, i.e., heart rate, respiratory rate and average facial temperature via linear discriminant analysis (LDA). The method achieved higher sensitivity (88%) than conventional methods using only thermography.2 However, our method using LDA indicated two limitations. The one was a limitation derived from being "a linear method", the other was the fact that LDA could not respond to "unknown or unexpected infectioufdiseases". In the present paper, we developed a non-linear screening method which determines the perfect screening condition using a neural network (self-organizing map) and k-means clustering algorithm (a non-linear clustering algorithm).