Bio-inspired Approach for Early Diabetes Prediction and Diet Recommendation

Diabetes mellitus is one of the hyperglycemic diseases. To meet with an objective of early prediction of diabetes, the paper comprises of case studies of diabetes patients, the existing working models used to predict diabetes in patients. This study aimed to use the nature-inspired metaheuristic algorithms like ant colony optimization, Bat Algorithm, Cuttlefish Algorithm, Elephant Herd Optimization Algorithm, and Artificial Bee Algorithm, etc. which are usually utilized for numerical analysis such as accuracy and other performance metrics. The objective was to develop a model that accurately recognizes diabetes by employing algorithms influenced by nature on a particular dataset. Diabetes was detected using several classification algorithms, and the accuracy of the classifiers was improved by tuning their hyperparameters using Hybrid Bat Algorithm. Most of the classifiers in use have their performance improved using various techniques. With the maximum accuracy of 98%, the voting classifier along with Smote and Bat Algorithm exceeded the competition, which is shown and discussed in the paper. The focus is on diabetes prediction, and then delve into the dietary recommendation aspect.