International Journal of Environmental Research and Public Health · 4 authors, 3 centres
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
This systematic review analyzed 34 studies on food recommender systems for diabetic patients, categorizing them into semantic, optimization, rule-based/classification, and interaction-based approaches. The main finding is that while various computational methods exist, the field lacks a consolidated research framework, robust evaluation protocols, and sufficient diabetes-specific datasets, limiting reproducibility and progress.
This systematic review, following the PRISMA 2020 framework, aimed to characterize the strengths and weaknesses of food recommender systems (RS) for diabetic patients. The authors searched Web of Science, Scopus, and PubMed (January 2010–November 2022), identifying 34 relevant papers after screening and quality assessment. Key results indicate that while semantic technologies are a common default, all categories suffer from weaknesses such as limited evaluation, lack of shared datasets, and insufficient personalization. The review highlights the absence of a unified research framework and the need for better integration of medical knowledge, user preferences, and explainable AI. Limitations include the focus on computer science literature and the exclusion of clinical studies without computational methods. Implications suggest future research should develop consolidated frameworks, exploit medical guidelines, enhance personalization, and incorporate fuzzy tools and explainability to advance food RS for diabetes management.
