Impact of Diet and Nutritional Factors in Controlling Type 2 Diabetes Through Machine Learning

Diabetes mellitus poses a significant challenge globally, necessitating an urgent need for effective solutions that deviate from the conventional route of medication. It is one of the world's leading causes of mortality yet taken less seriously in comparison with other similar life-threatening diseases. The standard approach of medications for diabetes involves receiving prescribed drugs such as synthetic insulin, which pose adverse negative effects on the body over time. Hence, it is crucial to challenge the notion that this approach is the only available treatment option. With a focus on glycaemic index (GI) and ketosis, this research proposes a model to determine the suitability of food items for diabetic patients and hence evaluate its healthiness. This can help people formulate their own personalized meal plans. The existing scientific research of how low GI foods can help promote ketosis and how it is the recommended type of nutrition for diabetic patients is used. Utilizing data from Openfoodfacts, a dataset of 625 records was preprocessed and analyzed, revealing key features influencing the glycaemic index. Through a machine learning model, this research aims to predict the GI value for any food item as closely as possible. By testing various classifiers, including Random Forest, K Neighbors, and Decision Tree models, we achieved an optimal accuracy of 99.5% with the Random Forest Classifier. Our research highlights the need for a shift from medication-centric approaches to diet-centric solutions in managing Type 2 Diabetes.