Validade de critério concorrente da pulseira interativa para adesão à higiene das mãos

Abstract

It is well known that Healthcare-Related Infections (HAIs) are a worldwide public health problem, since they affect the quality of care provided in healthcare services. To this end, even though hand hygiene (HH) is the most effective primary action to prevent HAIs, its adherence is lower than recommended worldwide. This study sought to validate and develop an innovative electronic wristband for healthcare professionals, with the aim of increasing adherence to HH. It should be noted that the proposal presented here is the subject of a patent application for a utility model, of the patent privilege type, under protocol number BR2020220012674 at the National Institute of Industrial Property (INPI). Objective: To evaluate the concurrent criterion validity of the wristband related to the correct hand hygiene technique, regarding the steps performed and the time spent. Methodology: Methodological study to validate competing criteria for a technology proposed to increase adherence to hand hygiene among healthcare professionals. Data collection took place in two stages, the first between April and September 2023, in a research laboratory at the Faculty of Nursing of the Federal University of Mato Grosso (FAEn/UFMT), with four participants, and 240 techniques of the 3 steps of HH were collected, resulting in 120 correct and 120 incorrect. The raw data was pre-processed, segmented, characterized and classified to obtain the first validated version of the algorithm. The gold standard for calculating concurrent criterion validity was the evaluation recorded on a validated instrument by an observer trained by carefully filming the techniques performed. The second stage took place in a realistic simulation environment, with the participation of healthcare professionals using the sensor and performing the three-step MH technique, to analyze the concurrent criterion validity of the first version of the algorithm to be used in the bracelet. Results: The correct HM technique collections, in the classification phase, obtained data divided into 10% for validation, 15% for testing and 75% for training the algorithm. The values resulting from the test classification matrix show that 14% of the data was classified as false negatives; on the other hand, 6.3% was classified as false positives in the training matrix; 15.9% was classified as false positives and 7.7% was classified as false positives in the training matrix. After classifying the data from the techniques correctly, 89.3% accuracy was obtained for the test confusion matrix with the application of the artificial neural network. Discussion: Studies have shown significant reductions in the occurrence of HAIs with the application of new technologies, emphasizing the relevance of investments in innovative products in the health area. That said, the algorithm created to read the MH technique has demonstrated accuracy, with similar validity to other studies classifying MH movements. Conclusion: The results obtained are promising and encourage the study to continue. It is important to carry out further analyses with incorrect movement data and in different clinical scenarios to improve the algorithm.