Diabetic Food Recommendation System According to Patients Behaviors

Abstract

Diabetic patients need to control their blood sugar levels to be close to normal by adjusting their diet to control sugar intake. The objective of this research is to control sugar intake based on the eating behavior of patients. In this research, we propose food recommendations based on the eating behavior of patients. The research employs two filtering techniques: content-based filtering and collaborative filtering. Recommended food is adjusted based on the sugar levels of the food that patients consume, and the energy used from their activities is also calculated. The energy is calculated as grams of sugar burned per kilocalorie per minute to encourage patients to consume more food variety. From the test results, it is found that the content-based filtering recommendation system has a higher precision value of 0.7462 and a higher recall value of 0.8163 than the collaborative filtering recommendation system.