Enhancement of Accuracy Measure by Detecting a Clinical Notes Using Support Vector Machine

This study evaluates the effectiveness of the Support Vector Machine (SVM) algorithm over the Naive Bayes algorithm in classifying clinical notes by medical specialty. We introduced a novel Support Vector Classifier to perform accuracy and loss assessments using the DATA dataset from the Kaggle repository, consisting of 20 samples. These samples were analyzed using both algorithms. The SVM achieved a notable accuracy of 86.58% with a loss of 13.61%, outshining the Naive Bayes’ accuracy of 73.25% and loss of 26.74%. The significant superiority of SVM in classifying medical specialties was statistically validated through a two-tailed independent sample T-Test, yielding a p-value of 0.001 at a 95% confidence level. This study underscores the SVM's enhanced proficiency in the categorization of medical specialties from clinical notes in comparison to Naive Bayes.