Classification vs Regression by CNN for Handwashing Skills Evaluations in Nursing Education

Abstract

Nosocomial infections have been of interest to experts in the medical field due to their ability to put human life at risk. Thus, nosocomial infection control is extremely important and methods vary. Handwashing methods can be applied to medical practice where there is a need for handwashing skills. Many studies have proposed handwashing skills evaluation methods. Today, quantitative evaluation methods make it possible to evaluate handwashing skills for the entire hand. However, these methods cannot provide numerical feedback on which part of the hand is left unwashed. Additionally, it is unclear whether handwashing skills are maintained after handwashing training is finished. In this paper, we propose a new quantitative evaluation method based on deep learning and investigate whether handwashing skills continue to take root after education is complete. Our experimental subjects were one hundred sixty-four students in the School of Nursing at the University of Toyama who were handwashing skills beginners. Collected image data were divided into training data and test data and a convolutional neural network (CNN) was trained and validated. The results show the CNN F-value was 93.0%. Next, we collected data from 23 students from the School of Nursing who had finished handwashing lectures and 16 handwashing beginners. Experienced handwashers demonstrated almost the same skills as those of beginners.