In recent years, the clinical decision support system (DSS) has emerged as an important area in medical sciences to assist clinicians in medical diagnosis. Health records classification is based on learning from various health datasets to improve the better quality of DSS in health care. The main objective of this investigation is to establish a system for the successful classification of health data. OLPP orthogonal local preservation projection has been used to obtain promising outcomes in the classification of medical data. This is a high-dimensional data input package. A feature-reduction tool is then used to reduce the functionality space without compromising the calculation accuracy. The Artificial Neural Network shall be used as a classifier. We used an optimization algorithm to boost efficiency. The "artificial bee colony algorithm" is a bio-based optimization algorithm used by a neural network. The strategy for optimizing the NN artificial bee colony's weight during the learning process improves the performance of classification. Our proposed solution findings have shown that, in comparison to other types, our optimal classifier findings are better. The medical datasets used to represent the average improvement in the proposed system classification quality with the current form is around 15%.