Three-class obstructive sleep apnea severity assessment: a parallel AHI and ODI explainable artificial…

AI / informaticsSleep & breathing = Schlaf & Atmung· Jul 2026

Three-class obstructive sleep apnea severity assessment: a parallel AHI and ODI explainable artificial intelligence framework using craniofacial-enriched clinical data.

Zahra Ameli Mazandarani, Mohammad Behnaz, Hamed AmiriFard … Shahab Kavousinejad

Craniofacial & CleftAI / informaticsDiagnosisQ1IF 2.7

AI model using craniofacial features classified obstructive sleep apnea severity with 87.2% accuracy.

Three-class obstructive sleep apnea severity assessment: a parallel AHI and ODI explainable artificial intelligence framework using craniofacial-enriched clinical data.

Zahra Ameli Mazandarani … Shahab Kavousinejad

Sleep & breathing · Jul 2026

Background

Machine learning models for Obstructive Sleep Apnea (Obstructive Sleep Apnea) diagnosis have largely inherited some structural limitations: reliance on generic, opportunistically collected feature sets; use of the Apnea-Hypopnea Index (Apnea-Hypopnea Index) as the sole ground truth; poor performance in multi-class severity grading; and predictions that offer clinicians no mechanistic insight.

Methods

In this single-center study, 233 treatment-naïve adults from a tertiary referral cohort (61.8% severe Obstructive Sleep Apnea prevalence) underwent in-laboratory polysomnography (polysomnography).

n = 47

Results

correctly graded severity in most patients, higher is better

87.2%

Anatomy-inclusive mode

72.3%

Ablation model

More results

The Oxygen Desaturation Index-based model achieved 76.6% accuracy, offering reliable exclusion of severe desaturation burden (No/Mild specificity: 0.94).

More results

Univariate analyses confirmed significant associations between Obstructive Sleep Apnea severity and STOP-BANG score, age, BMI, neck circumference, observed apnea, loud snoring, high blood pressure, Cervico-Mental Angle, Mentocervical Distance, and submental fat (all p ≤ .034 for both indices).

“

Conclusion

As a proof-of-concept, this study demonstrates that an interpretable ML framework integrating craniofacial and intraoral assessments with standard clinical predictors can classify Obstructive Sleep Apnea severity across three classes and provide feature-level explanations to support clinical reasoning.

Three-class obstructive sleep apnea severity assessment: a parallel AHI and ODI explainable artificial intelligence framework using craniofacial-enriched clinical data.

Three-class obstructive sleep apnea severity assessment: a parallel AHI and ODI explainable artificial intelligence framework using craniofacial-enriched clinical data.

Related papers