A Data-Driven Framework for Clinical Decision Support Systems in Positive Airway Pressure and Oxygen Titration

Area of Science:

• Respiratory Medicine
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Manual positive airway pressure (PAP) titration for obstructive sleep apnea (OSA) presents challenges, including complex interactions and variable SpO2 data accuracy.
• Patients with co-occurring chronic hypercapnia require precise oxygen titration, which is difficult with current methods.

Purpose of the Study:

• To develop and evaluate a Clinical Decision Support System (CDSS) utilizing Markov decision processes (MDPs) for optimizing PAP and oxygen titration in OSA patients.
• To address limitations in manual titration and improve management of OSA patients, particularly those with hypercapnia.

Main Methods:

• A study involving adults with OSA-induced hypoxemia undergoing one-night PAP titration with supplemental oxygen, adhering to data protection regulations.
• Utilized Markov decision processes to define three treatment actions for disorder management, guided by real-time patient metrics and expert medical insights.
• Monitored vital signs and physiological parameters, adjusting CPAP pressure, BiLevel transitions, and oxygen levels based on patient data.

Main Results:

• Significant reduction in Apnea-Hypopnea Index (AHI) from 61.8 to 18.0 events/hour (p < 0.0001) and increase in SpO2 from 79.7% to 89.1% (p < 0.0003) post-titration.
• Identified hypercapnia aggravation in 50% of patients during CPAP pressure increase; transitioning to BiPAP effectively reduced pCO2 and increased SpO2.
• MDP analysis demonstrated realistic patient behavior, minimal apnea, and good tolerance to high CPAP pressure under stable conditions.

Conclusions:

• A Markov decision process framework for PAP and oxygen titration algorithms shows promise for improving pCO2 and SpO2 values in OSA patients.
• Despite challenges with data quality, this approach offers substantial benefits for patient management and care optimization in telemedicine and respiratory healthcare.
• This represents an innovative frontier in optimizing respiratory support for OSA patients.