JMIR Publications

ABSTRACT

Background:

It is challenging to maintain focus on health-related activities due to our lifestyle choices, work commitments, and varying levels of motivation towards being healthy. Applying machine learning methods to data generated by ubiquitous devices like smartphones present an opportunity to enhance the quality of healthcare and diagnostics. Smartphones are ideal for providing quick feedback on diagnoses and proposing interventions for health improvement.

Objective:

As this is an area of growing interest, we reviewed existing literature to gather studies that have used machine learning techniques with smartphone-derived data for the prediction and diagnosis of health anomalies.

Methods:

A comprehensive search of PubMed and IEEE Xplore databases was conducted, and a keyword screening method was used to filter the papers based on their title and abstract. Afterwards, the papers that studies about the three areas: voice, skin and eye, were extracted from all the available literature, and analyzed based on how data for machine learning models were extracted, i.e., either by the use of publicly available databases, or through experiments. Machine learning methods used in each study were also noted.

Results:

As a result, a total of 49 studies were selected as relevant to the topic of interest, among which there were 31 different databases along with 24 different machine learning methods.

Conclusions:

These details provide a better understanding of how smartphone data are collected for predicting different diseases, and what kind of machine learning methods are used on them. The findings of the paper could be used or improved in future studies as a reference to conduct similar studies, experiments or statistical analysis.