Diabetes Mellitus Prediction.

Abstract

After evaluating the performance of our six models, it is evident that the ACO-LSTM model is the superior choice, achieving the highest accuracy of 97.1%. This model's remarkable performance underscores the potential of hybrid approaches that integrate swarm intelligence with deep learning techniques. The consistent high performance of the PSO-LSTM and ACO-GRU models further validates the efficacy of combining optimization algorithms with advanced neural networks. As we move forward, we are committed to refining these models, addressing their limitations, and expanding their applications to a broader range of medical conditions, thereby contributing to the advancement of predictive healthcare technologies.