Potential of Machine Learning for Predicting Sleep Disorders: A Comprehensive Analysis of Regression and...

Area of Science:

• Computational neuroscience
• Medical informatics
• Machine learning applications in healthcare

Background:

• Sleep disorders present significant emotional and physical challenges, impacting daily functioning and well-being.
• Existing diagnostic and predictive methods for sleep disorders may benefit from advanced computational approaches.
• Machine learning offers powerful tools for analyzing complex health datasets and predicting disease outcomes.

Purpose of the Study:

• To leverage machine learning for accurate prediction of sleep disorders.
• To identify the optimal regression and classification models for sleep disorder datasets.
• To determine the most effective learning strategy for sleep disorder prediction tasks.

Main Methods:

• Evaluation of twenty-three regression models and multiple classification models on two distinct sleep disorder datasets.
• Utilized various performance metrics relevant to both regression and classification tasks.
• Compared six different learning strategies to assess their predictive efficacy.

Main Results:

• MultilayerPerceptron, SMOreg, and KStar demonstrated superior performance as regression models.
• IBK, RandomForest, and RandomizableFilteredClassifier emerged as top-performing classification models.
• The Function learning strategy exhibited the highest predictive accuracy across both datasets and most metrics.

Conclusions:

• Machine learning models, particularly MultilayerPerceptron, SMOreg, KStar, IBK, and RandomForest, show significant promise in predicting sleep disorders.
• The Function learning strategy is highly effective for sleep disorder prediction.
• These findings can inform the development of advanced, data-driven tools for sleep disorder diagnosis and management.