JMIR Publications

ABSTRACT

Background:

Health recommender systems (HRS) offer great potential to motivate and engage users to change their behavior by offering better choices and actionable knowledge based on observed user behavior.

Objective:

We review HRS for non-medical professionals (laypersons) to better understand the current state-of-the-art and identify both the main trends as well as the gaps with respect to current implementations.

Methods:

We conducted a systematic literature review according to the PRISMA guidelines and synthesized the results; 73 published studies reporting both implemented and evaluated HRS targeted to laypersons were included in the review and analyzed.

Results:

Recommended items are classified into four major categories: lifestyle recommendations, nutritional recommendations, providing general healthcare information or recommending actions for specific health conditions. Hybrid recommender algorithms are the most popular technique. Evaluations of HRS vary greatly. While many of the studies only evaluate the algorithm and thus lack a user-centered evaluation approach, some studies performed full-scale randomized controlled trials (RCT) or conducted in the wild studies to evaluate the impact of HRS, showing the field is slowly maturing. Based on our review, we argue that it should always be clear what the HRS is recommending and to whom these recommendations apply. Recommendations should be presented in a visual manner. Finally, studies should report the dataset and algorithms that were used to calculate the recommendations.

Conclusions:

There is a significant opportunity for HRS to inform and guide health actions. We promote discussion of ways to augment HRS research by recommending design guidelines.