Year: 2025
Type: conference-abstract
Abstract: Background A significant proportion of patients with Obstructive Sleep Apnea (OSA) are unable or unwilling to adhere to conventional treatments such as continuous positive airway pressure (CPAP). Emerging therapies like optogenetic stimulation offer a novel, personalized approach by activating upper airway dilator muscles immediately prior to inspiration, thereby preventing apnoeic events. Methods We developed a wearable intraoral sensor array embedded in a custom mouthguard to predict optimal stimulation timing. The device integrates photoplethysmography (PPG), head acceleration, and intraoral acoustic sensing. Data were collected during overnight in-lab polysomnography (PSG). A convolutional neural network (CNN) was trained to fuse multimodal signals and detect ideal stimulation points within the respiratory cycle. The study targets 20 participants (OSA, n = 10); this interim analysis includes data from the first 8 participants (OSA, n = 5). Results The model was trained on 16 960 seconds of synchronized sensor and PSG data and evaluated on 1040 seconds of held-out test data. The CNN achieved a precision of 92%, recall of 86%, and an F1 score of 0.89 in identifying optimal stimulation windows. Ground truth validation was performed using PSG pressure cannula and thermistor waveforms. Conclusion These preliminary results demonstrate the feasibility and accuracy of using multimodal intraoral sensing and deep learning to deliver precision-timed optogenetic stimulation. This approach holds significant promise as a non-invasive, personalized therapy for patients who are unable to tolerate or benefit from standard OSA treatments.
Source: SLEEP Advances
Institutions: Neuroscience Research Australia, UNSW Sydney
Language: English
Subfield: Physiology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,054
OpenAlex ID: W4414793914
Cites:
Cited by:
Related to: