Abstract :

Frequent Breathing Pauses During Sleep Are The Hallmark Of Sleep Apnoea, A Common And Potentially Fatal Sleep Disease. Serious Health Issues Like Weariness, Decreased Cognitive Function, And Cardiovascular Disorders Might Result From These Periods. Although Current Deep Learning Models Have Demonstrated Good Performance In Apnoea Diagnosis, They Are Less Appropriate For Real-time Or Edge-based Healthcare Applications Due To Their High Computational Resource Requirements And Lack Of Transparency. In This Work, We Provide A Machine Learning-based Method That Uses Extreme Gradient Boosting (XGBoost) To Effectively And Precisely Identify Sleep Apnoea Episodes From Physiological Signals.A Strong Feature Engineering Pipeline Is Used In The Suggested System To Extract Statistical, Temporal, And Frequencydomain Characteristics From Biosignals Include Oxygen Saturation (SpO₂), Airflow, And ECG Data. To Reduce Redundancy And Improve Classification Performance, These Characteristics Are Chosen And Improved Utilising Correlation Analysis And Feature Importance Techniques. In Order To Address Class Imbalance And Enhance Generalisation, The Collected Features Are Subsequently Fed Into An XGBoost Classifier, Which Is Optimised Using Cross-validation And Hyperparameter Tuning. In Comparison To Deep Learning Models, Our Approach Shows Competitive Accuracy While Drastically Lowering Computational Cost When Tested On The Sleep Heart Health Study (SHHS) Dataset. Additionally, The Model Uses Feature Importance Analysis To Produce Interpretable Outputs That Help Clinical Experts Comprehend Choice Variables. Because Of This, The Method Is Ideal For Explainable AI-driven Diagnoses, Portable Device Deployment, And Real-time Monitoring In The Context Of Sleep Apnoea And Related Diagnosis.

Published:

14-8-2026

Issue:

Vol. 26 No. 8 (2026)

Page Nos:

911-918

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite