Disease Prediction Using Machine Learning on Smartphone-Based Eye, Skin, and Voice Data: Scoping Review.

Area of Science:

• Health Informatics
• Machine Learning Applications
• Mobile Health (mHealth)

Background:

• Ubiquitous smartphone data offers opportunities for enhanced healthcare and diagnostics.
• Smartphones facilitate easy data collection, rapid diagnostic feedback, and health improvement interventions.

Purpose of the Study:

• To review literature on machine learning (ML) models using smartphone data for health anomaly prediction and diagnosis.
• To categorize studies based on data acquisition (experiments vs. public databases) and ML model application.
• To provide researchers with insights into databases, experiments, and ML models in the mHealth domain.

Main Methods:

• Comprehensive literature search of PubMed and IEEE Xplore databases.
• In-house keyword screening of titles and abstracts for article filtering.
• Analysis of selected studies focusing on voice, skin, and eye data, distinguishing between experimental and public database usage for ML model data extraction.

Main Results:

• Identified 49 relevant studies.
• Cataloged 31 distinct databases and 24 different machine learning methods used in the reviewed literature.

Conclusions:

• The findings enhance understanding of smartphone data collection for disease prediction and the ML methods employed.
• Publicly available smartphone-based datasets for disease diagnosis are highlighted.
• The review's methodology and findings serve as a reference for future mHealth research and analysis.