Scholarlink

Introduction <p>Sleep Apnea (SA) is a rapidly spreading sleep-disorder. A subject with SA can observe repeated interruptions or pauses in breathing during sleep which significantly affects subjects’ life. SA increases the risk of chronic illness like Hypertension, Cardiovascular diseases etc. For effective treatment and management of SA, it's early-stage diagnosis is very important. This research article explores application of Machine Learning (ML), Optimization techniques, dimensionality reduction techniques for the identification of SA using a structured dataset.</p> Methods <p>We utilized Sleep Health &amp; Lifestyle Dataset (SHLD) which consisting of 560 patients records with 13 attributes, mainly including demographic, physiological and lifestyle related features. We implemented a machine learning based SA diagnostic pipeline which includes preprocessing, normalizing and label encoding. We have used many ML classifiers like Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), Multi-Layer-Perceptron (MLP) on SHLD dataset. To mitigate data imbalance, we have used Mathews Correlation co-efficient (MCC) as evaluation parameter. In order to increase prediction accuracy of ML classifiers, along with hyperparameter tunning we have employed Principal Component Analysis (PCA) and Particle Swarm Optimization (PSO) as well.</p> Results <p>Among all the classifiers we tested, the hybrid PSO-SVM model demonstrated the best performance with an accuracy of 95.24% against individual SVM model (92.86%) and PCA-SVM model. We observed that, the integration of PCA and PSO with SVM has improved overall performance of model.</p> Conclusion <p>This research study confirms that the combination of PCA/PSO with SVM significantly improves its ability to detect Sleep Apnea. Also, MCC proves to be effective performance measure while dealing with imbalanced dataset.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Harnessing Machine Learning and Optimization Techniques for Identifying Risk Factor of Sleep Apnea by Exploring Sleep Health and Lifestyle Dataset

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • Mubashir Khan,
  • Yashpal Singh,
  • Harshit Bhardwaj

摘要

Introduction

Sleep Apnea (SA) is a rapidly spreading sleep-disorder. A subject with SA can observe repeated interruptions or pauses in breathing during sleep which significantly affects subjects’ life. SA increases the risk of chronic illness like Hypertension, Cardiovascular diseases etc. For effective treatment and management of SA, it's early-stage diagnosis is very important. This research article explores application of Machine Learning (ML), Optimization techniques, dimensionality reduction techniques for the identification of SA using a structured dataset.

Methods

We utilized Sleep Health & Lifestyle Dataset (SHLD) which consisting of 560 patients records with 13 attributes, mainly including demographic, physiological and lifestyle related features. We implemented a machine learning based SA diagnostic pipeline which includes preprocessing, normalizing and label encoding. We have used many ML classifiers like Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), Multi-Layer-Perceptron (MLP) on SHLD dataset. To mitigate data imbalance, we have used Mathews Correlation co-efficient (MCC) as evaluation parameter. In order to increase prediction accuracy of ML classifiers, along with hyperparameter tunning we have employed Principal Component Analysis (PCA) and Particle Swarm Optimization (PSO) as well.

Results

Among all the classifiers we tested, the hybrid PSO-SVM model demonstrated the best performance with an accuracy of 95.24% against individual SVM model (92.86%) and PCA-SVM model. We observed that, the integration of PCA and PSO with SVM has improved overall performance of model.

Conclusion

This research study confirms that the combination of PCA/PSO with SVM significantly improves its ability to detect Sleep Apnea. Also, MCC proves to be effective performance measure while dealing with imbalanced dataset.