Adaptive Thresholds and Statistical Methods for Automatically Identifying Central and Mixed Apnea Sequences in Minimal-Contact Biomedical Signals

Sleep-disordered breathing affects a significant portion of the population, with prevalence increasing alongside age and body weight. Given the growing need for effective diagnostic and treatment methods in sleep medicine, innovative telemedical approaches are essential. To address this challenge, more precise out-of-center testing could significantly enhance the detection of sleep-disordered breathing in home environments. This involves utilizing smart, non-invasive, and minimal-contact diagnostic devices supported by advanced automatic data processing algorithms. In this paper, we analyze the signal from a minimal-contact piezoelectric sensor placed on the mattress to identify events of sleep-disordered breathing. Our approach employs adaptive thresholds and statistical methods to achieve superior accuracy in detecting central and mixed apnea events compared to conventional techniques. One of the key advantages of our method is its adaptability; it can be easily customized for individual patients based on their specific characteristics and needs. This personalization enhances the effectiveness of detection while ensuring that interventions can be tailored appropriately. Preliminary results demonstrate promising classification outcomes, indicating not only high accuracy but also robustness across diverse patient profiles. The ability to accurately detect apnea events using minimal contact opens up new avenues for long-term monitoring and management of sleep-disordered breathing outside conventional clinical settings, thereby improving patient accessibility to care.