JMIR Publications

ABSTRACT

Background:

Preventive care aids patients by helping them identify diseases that can cause medical problems before they become serious. The internet provides a wealth of data online about available preventative measures. Unfortunately, Humans’ working memory and reasoning ability cannot process all the data online; therefore, recommender systems assist in processing and providing recommendations from these data. Publications in the recommender system research area are domain-specific and are dominated by service and retail industries with limited publications based in the healthcare context

Objective:

This paper suggests practice-based empirical propositions for developing recommender systems. We also describe a study design, methods for developing a survey, and conducting an analysis

Methods:

We propose using a survey to collect data from approximately 600 participants on Amazon’s M-Turk, then using SAS, STATA, R, or Python to analyze the research model. Researchers should perform a principal component analysis, Harman Single Factor test, exploratory factor analysis, correlational analysis, examine the reliability and convergent validity of individual items, test if multicollinearity exists, and complete a confirmatory factor analysis.

Results:

Data collection and analysis can start once IRB approval is obtained.

Conclusions:

Examining recommender systems for preventative care can be vital in achieving the quadruple aims by advancing the steps toward precision medicine and applying best practices.