This study compares the performance of two outlier detection algorithms, specifically the Local Outlier Factor (LOF) and Isolation Forest, and how they can be used in medical research, particularly in identifying potential health risks or medical errors in diabetes datasets. The study aims to compare the performance of these methods in detecting outliers in the Pima Indians Diabetes Dataset using f1 score, precision, and recall. According to the results, the LOF algorithm detected the fewest outliers, while the isolated forest method detected the most. The combined LOF-Isolation Forest method detected outliers slightly more frequently than the Isolation Forest method. Despite differences in the number of outliers detected, all three algorithms had similar weighted average recall ratings, indicating that they all detected a similar proportion of true outliers This information sheds light on outlier detection applied to diabetes datasets and has the potential to increase diagnostic predictability.