Developing a Multisensor-Based Machine Learning Technology (Aidar Decompensation Index) for Real-Time Automated...

Area of Science:

• Cardiology
• Pulmonology
• Infectious Diseases
• Biomarkers
• Digital Health

Background:

• Post-COVID-19 condition presents a significant global health challenge with persistent symptoms beyond 3 months.
• Severe COVID-19 survivors face increased hospitalization risks due to poorly understood pathophysiological mechanisms.
• Long-term effects on chronic cardiovascular and pulmonary diseases are anticipated in COVID-19 patients.

Purpose of the Study:

• To develop and validate the Aidar Decompensation Index (AIDI) for predicting health decompensation.
• To utilize the integrated MouthLab device and a cloud-based analytics engine for real-time health monitoring.
• To assess AIDI's efficacy in patients with a history of severe COVID-19.

Main Methods:

• Enrollment of 200 participants with severe COVID-19 history and chronic conditions.
• Daily physiological data capture via MouthLab device and monthly symptom surveys.
• Development of the AIDI based on physiological signals and clinical characteristics to predict decompensation events (DEs).

Main Results:

• Recruitment of 204 patients initiated in January 2023.
• Study completion and full results publication are anticipated in 2025.
• The AIDI model is expected to achieve >80% sensitivity and >70% positive predictive value.

Conclusions:

• Identifying predictor variables from biosensor-derived physiological features can capture diverse post-COVID-19 complications.
• The AIDI has the potential to enable early detection and management of health decline in severe COVID-19 survivors.
• This approach may improve long-term outcomes for individuals affected by the post-COVID-19 condition.