Abstract
Background: Obstructive sleep apnea is a common and sometimes fatal condition characterized by frequent pauses in breathing during sleep. It is estimated to affect between 9% and 38% of all adults. It affects between 5% and 10% of middle-aged adults and up to 20% or 30% of older people (65 years or older), and prevalence increases with age and obesity. Geographic differences exist because different countries and ethnic groups may experience different lifestyles, nutrition and genetic predisposition. According to estimates, up to 80% of people with moderate to severe apnea may not receive treatment. These figures highlight the critical need for accurate, real-time, and easy-to-use diagnostic tools that can be applied outside clinical settings to track sleep quality and identify apnea episodes early, allowing for rapid treatment and intervention.
Approach: This study offers a novel option by using artificial intelligence (AI) for the evaluation of data from noninvasive sleep monitoring devices that record heart rate variability, brain wave activity, and breathing patterns according to the protocol followed by BTI company who leads the project. The AI system is able to detect apnea episodes, variations in typical breathing patterns, and assesses overall sleep quality. This method provides patients, dentists, and other healthcare professionals with useful information and allows for a more complete understanding of sleep quality which is really important for the quality of life of the whole society but, in particular, for elderly people. This system is being developed in cooperation with AI researchers, sleep specialists, and dentists of BTI company. As they are often the first medical professionals to notice sleep apnea symptoms during dental exams, dentists are essential to this effort.This system is training AI models using machine learning classifiers like Random Forest and SVM, deep learning architectures such as LSTMs and CNNs, Transformer models like BERT for sequential data, and exploring quantum machine learning techniques with QML frameworks like Pennylane and TensorFlow Quantum, all on time-series data from 6,317 patients.
Results: By monitoring their sleep habits, users have a better understanding of their sleep health, which they share with dentists and other medical professionals for ongoing care. Additionally, the system’s alert feature is intended to facilitate prompt interventions by quickly informing users when severe cases of apnea occur. This system takes a proactive approach to sleep health management by offering real-time data and prediction capabilities, allowing users to seek treatment when necessary or make appropriate lifestyle changes.
Implications: Integrating AI with sleep monitoring offers the potential to transform sleep health management with an accessible and effective tool for sleep apnea detection and quality assessment. Patients, people and, especially older adults, have an active role in managing their sleep health and it enables healthcare providers to make more informed decisions about treatment options.