A Machine Learning-Based Screening Model for Obstructive Sleep Apnea Diagnosis and Severity Prediction

<bold>Introduction:</bold> Obstructive sleep apnea syndrome (OSAS) is associated with serious health risks, making early diagnosis and severity assessment essential. This study explores an innovative approach by integrating machine learning with clinical parameters, including sleepiness scale, age, gender, comorbidities, and anthropometric measurements, to diagnose and classify OSAS severity. <bold>Methods:</bold> This multicenter, retrospective cross-sectional study included patients over 18 years with or without PSG-confirmed OSAS. PSG results were categorized into Stage-0 to Stage-3 based on severity. An artificial neural network (ANN) model was developed with 25 input variables and 4 output classes. The hidden layer contained 50 neurons with RELU activation, while SOFTMAX was used in the output layer. The dataset was split into 80% training and 20% testing. <bold>Results:</bold> A total of 1,954 participants from 9 centers were included; 70.16% were male (n=1,371). The mean age was 47.51 years. Distribution of sleep stages: 39.15% Stage 3, 29.43% Stage 1, 16.38% Stage 0, and 15.05% Stage 2. The model’s performance was assessed via ROC analysis, yielding an AUC of 0.97, indicating strong predictive ability. PPV was 97.56%, and NPV was 87.11%. <bold>Conclusion:</bold> The ANN model demonstrated high accuracy in OSAS diagnosis, suggesting potential for clinical application. Additionally, the applicability of this approach in community screening may facilitate more efficient management of healthcare resources.

Authors

Institutions

Publication Details

Journal
Sleep Science
Published
2025-09-27
DOI
https://doi.org/10.1183/13993003.congress-2025.pa3543
Primary Topic
Obstructive Sleep Apnea Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

A Machine Learning-Based Screening Model for Obstructive Sleep Apnea Diagnosis and Severity Prediction

Aziz Gümüş, Fatma Gülsüm Karakaş, Emre Bendeş, Abdurrahman Kotan et al.
Sleep Science
Obstructive Sleep Apnea Research
article

A Machine Learning-Based Screening Model for Obstructive Sleep Apnea Diagnosis and Severity Prediction

Aziz Gümüş, Fatma Gülsüm Karakaş, Emre Bendeş, Abdurrahman Kotan, Ömer Topaloğlu, Bilge Kara, Ünal Şahin, Abdullah Kansu, Gizem Işık, Burcu Baran Ketencioğlu, Nur Aleyna Yetkin, Songül Özyurt, Önder Öztürk, Mehmet Karadağ, Yılmaz Bülbül, Nurhan Sarıoğlu, Muhammed Emin Akkoyunlu, Özge Aydın Güçlü, Neslihan Özçelik, Elvan Şentürk Topaloğlu, Mustafa Çolak, Hadice Selimoğlu Şen, Merve Nur Yıldız, Ali Erdem Özçelik, Hilal Türkmen Kaya, Sibel Doğru
article en

Abstract

<bold>Introduction:</bold> Obstructive sleep apnea syndrome (OSAS) is associated with serious health risks, making early diagnosis and severity assessment essential. This study explores an innovative approach by integrating machine learning with clinical parameters, including sleepiness scale, age, gender, comorbidities, and anthropometric measurements, to diagnose and classify OSAS severity. <bold>Methods:</bold> This multicenter, retrospective cross-sectional study included patients over 18 years with or without PSG-confirmed OSAS. PSG results were categorized into Stage-0 to Stage-3 based on severity. An artificial neural network (ANN) model was developed with 25 input variables and 4 output classes. The hidden layer contained 50 neurons with RELU activation, while SOFTMAX was used in the output layer. The dataset was split into 80% training and 20% testing. <bold>Results:</bold> A total of 1,954 participants from 9 centers were included; 70.16% were male (n=1,371). The mean age was 47.51 years. Distribution of sleep stages: 39.15% Stage 3, 29.43% Stage 1, 16.38% Stage 0, and 15.05% Stage 2. The model’s performance was assessed via ROC analysis, yielding an AUC of 0.97, indicating strong predictive ability. PPV was 97.56%, and NPV was 87.11%. <bold>Conclusion:</bold> The ANN model demonstrated high accuracy in OSAS diagnosis, suggesting potential for clinical application. Additionally, the applicability of this approach in community screening may facilitate more efficient management of healthcare resources.

Sleep Science
Bursa Uludağ Üni̇versi̇tesi̇ (TR), Recep Tayyip Erdoğan University (TR), Karadeniz Technical University (TR), Süleyman Demirel Üniversitesi (TR), Istanbul Medipol University (TR), Nevşehir Hacı Bektaş Veli University (TR), Erzurum Regional Training and Research Hospital (TR), Kayseri Eğitim ve Araştırma Hastanesi (TR), Pulmonary Associates (US), Gaziantep University (TR), Bolu Abant İzzet Baysal University (TR), Erciyes University (TR)
Openalex Percentile: Top 55%
Obstructive Sleep Apnea Research
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.