Food recommender system for cancer patients based on sentiment analysis.

Abstract

Cancer is a serious disease characterized by abnormal and irregular cell development in any part of the body, in the form of a tumor. It is considered as the second-leading cause of death in the world. Efforts to find a successful cancer therapy have led to the effective use of various treatments such as chemotherapy, and surgery to eliminate dangerous tumors. However, these treatments may affect the patient's immune system by damaging blood cells that protect the body from disease. Therefore, it is strongly recommended that cancer patients consume nutrient-rich foods to increase their strength to better cope with the side effects of treatment. In this dissertation, a hybrid food recommendation system which considers the patient's emotional state and dietary preferences and needs to help them predict the foods which can be consumed have been proposed. The approach involve to use a content-based system that filters recipes according to user needs and sentiment scores. Additionally, a rule-based sentiment analysis method was employed to identify sentiment from text reviews determining which of the foods were liked or disliked. The efficacy of this proposed approach was rigorously assessed, and the results yielded promising insights. Notably, combining content based, sentiments analysis led to a marked improvement with a precision of 97%.