Bayesian networks for the development of an interdisciplinary decision support system in screening for the risk of laryngotracheal aspiration

Abstract

Screening for the risk of laryngotracheal aspiration in hospitalized adults is a complex and interdisciplinary task that involves multiple variables and causes. The objective of this study was to develop and assess the applicability of a clinical decision support system based on Bayesian Network to assist the interdisciplinary hospital team. Methods: A Bayesian network was constructed using a systematic literature review to identify and categorize the main variables required for assessing the risk of laryngotracheal aspiration. These variables were grouped into the patient’s current diagnosis, history of swallowing disorders, swallowing-related aspects, and basic epidemiological characteristics. An application was developed and subsequently evaluated by nine specialists from an interdisciplinary hospital team. The evaluation focused on both the internal and external quality of the application. Results: The system was developed for Android mobile devices and was evaluated by a group of specialists (n=9) from various disciplines including physicians, nurses, nutritionists, and speech pathologists. The specialists considered the software product to be functional (86.1%), reliable (83.3%), usable (84.4%), efficient (92.6%), maintainable (90.4%), and portable (88.9%). Conclusions: The interdisciplinary screening system developed in this study was found to be valuable and feasible for use by specialists. Study registration: The validation of this study involved specialists from the fields of medicine, nursing, nutrition, and speech pathology. As a result, it was necessary to obtain evaluation and approval from the Ethics Committee in Research with Human Beings of the Federal University of Health Sciences of Porto Alegre – UFCSPA (# 44920121.0.0000.5345).