An Optimistic K-Nearest Neighbor Algorithm for Detecting Brain Stroke in Comparison with Logistic Regression Algorithm to Improve the Accuracy

The proposed system aims to enhance the resultant accuracy measured from stroke prediction obtained from the Novel Optimistic K-Nearest Neighbor Algorithm (NOKNN) and subsequently compare its performance with Logistic Regression (LR). Materials and Methods: In this research, the Novel Optimistic K-Nearest Neighbor and Logistic Regression algorithms were used to identify brain strokes using machine learning. With a group sample size of 20 per algorithm evaluation, the total sample size of 40 is used for comparison. The purpose of comparing these two sets of data is to enhance the research’s overall accuracy. The research’s accuracy was assessed using the ClinCalc tool, which was configured with an alpha value of 0.8, a 0.8 of G-Power value, 0.05 value of significance, and 95% of Confidence Interval (CI). Result: From the result, it was found that the NOKNN algorithm has produced an accuracy of 98.5% while the Logistic Regression produced an accuracy of 93.05%. After performing an independent sample T-Test, the resulting p=0.000, shows p<0.05 significance level. The results are statistically significant. Conclusion: As technology advances, it becomes increasingly important to develop more accurate and effective supervised solutions for addressing complex problems and helps in decrease of death rate. Therefore the proposed Novel Optimistic KNN gets higher accuracy when compared with the LR.