# Diet-Related Health Recommender Systems for Patients With Chronic Health Conditions: Scoping Review

## TL;DR

This review explores diet-related health recommender systems for chronic disease patients, highlighting their potential and current limitations.

## Contribution

The study provides a comprehensive scoping review of diet-related HRSs for chronic conditions, identifying research gaps and future directions.

## Key findings

- Most diet-related HRSs target diabetes and hypertension patients, using hybrid recommendation methods for personalization.
- Current systems lack long-term evaluation of their impact on dietary behavior and health outcomes.
- Hybrid approaches and data from professional sources are commonly used, but user-centered design and standardized evaluation are needed.

## Abstract

Diet-related Health Recommender Systems (HRSs) have gained attention for their potential to provide personalized dietary guidance, particularly for patients with chronic conditions. However, studies on diet-related HRSs in health care are relatively limited.

This scoping review aims to present the state of current research on diet-related HRSs for patients with chronic health conditions, identify existing gaps, and suggest future research directions.

The scoping review was conducted following the Arksey and O’Malley framework and was reported in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. The literature search was conducted in October 2024 across 6 English databases (PubMed, Medline, Embase, Web of Science Core Collection, IEEE Xplore, and CINAHL) and 4 Chinese databases (SinoMed, CNKI, Wanfang, and VIP). Studies focusing on diet-related HRSs for patients with chronic conditions were included.

Fifteen studies published between 2010 and 2024 from 9 countries were included. Diet-related HRSs mainly target adults with chronic diseases, with 9 systems (60%) including users with diabetes and 6 (40%) including users with hypertension. Nine studies (60%) described functional structures, which were categorized into 4 components: user information, food or diet recommendations, knowledge and decision support, and data management with additional functions. Recommended content was categorized into 5 types: food (n=6, 40%), recipes (n=4, 26.67%), diet plans or meal plans (n=3, 20%), recipes and food (n=1, 6.67%), and meals (n=1, 6.67%). Recommendation methods included constraint-based (n=6, 40%), focusing on patients’ dietary restrictions; preference-based (n=5, 33.33%), considering patients’ food preferences; and hybrid (n=4, 26.67%), combining both approaches. Of all recommendation technologies, most studies (n=13, 86.67%) applied hybrid approaches, enabling more robust personalization. For the data used for training, 13 studies (86.67%) explicitly mentioned the data sources, and 10 studies’ (66.67%) data came from professional organizations and websites. The recommendation process followed a structured workflow. Twelve studies (80%) evaluated diet-related HRSs using either online or offline methods, while accuracy (n=9, 60%) has been the most common evaluation criterion. However, no studies went deeper into how these systems affected users’ dietary behaviors over time.

Diet-related HRSs have the potential to deliver personalized dietary support for patients with chronic diseases, but current systems show key gaps. Future development must adopt user-centered design, provide practical and actionable dietary guidance, and use hybrid recommendation techniques to increase precision and clinical relevance. Standardized evaluation methods and real-world, long-term studies are essential to evaluate the impact of diet-related HRSs on dietary behavior and health outcomes. Addressing these needs will enable diet-related HRSs to become reliable tools for chronic disease management and patient-centered care.