Global Journal of Medical Research, A: Neurology & Nervous System, Volume 23 Issue 3

Classifier Performance Metrics: CLOXtask1 total score: Gini Index: 0.876, Max K-S Statistics: 0.759, Cutoff: 12.0000; CLOXtask2 total score: Gini Index: 0.880, Max K-S Statistics: 0.834, Cutoff: 14.0000; Both CLOX task1 total score and CLOX task 2 total score demonstrate strong performance in differentiating between positive and negative actual states, with CLOX task 2 having a slightly higher Area Under the ROC Curve. The Gini Index quantifies the disparity in the distribution of predicted probabilities. A higher Gini Index indicates a more effective classifier. In our case, the Gini Index for CLOX1.The total is 0.876, whereas it is 0.880 for CLOX2.Total. These values indicate that both classifiers have a high capacity for discrimination in predicting the actual positive state of Health Control. The K-S Statistics measure the greatest disparity between the cumulative distribution functions of the positive and negative categories. It indicates the classifier's capacity to distinguish between the two groups. CLOX task 1 total score and CLOX task 2 total score have Max K-S values of 0.75 and 0.83, respectively. The greater the Max K-S value, the more efficient the classifier. In both instances, the Max K-S values reported are associated with particular cut-off values. If multiple cut-off values exist, the largest is reported. These Max K-S values indicate that both the scores classifiers are able to distinguish between positive and negative actual states (please see the Figure 3 and 4). Figure 3: ROC curve Figure 4: Model Quality 64 Year 2023 Global Journal of Medical Research Volume XXIII Issue III Version I ( D ) A © 2023 Global Journals Reliability and Validity Evaluation of the ‘’CLOX: An Executive Clock Drawing Task’’ in a Greek Population with Neurological and Autoimmune Diseases ROC curves and CLOX task1 and Task2 In accordance with the above findings, have run a ROC curve analysis as it is a valuable tool in psychometric validation, providing insights into the discriminatory power, optimal cut-off point, comparative analysis, and diagnostic accuracy of the CLOX as a tool. The Area Under the ROC Curve (AUC) is a commonly employed performance metric for classifiers and diagnostic tests. It measures the test's ability to distinguish between positive and negative cases. In our case, the AUC values for the variables represent test results. The test result variables being evaluated based on the ROC curve analysis are CLOX task 1 total score and CLOX task 2 total score. The actual condition being evaluated is Health Control. The Area Under the ROC Curve for CLOX task 1 was 0.938 and for CLOX task 2, it was 0.940. The AUC values indicate that both scores have a high ability to differentiate between genuine positive and negative states. The Classifier Evaluation Metrics provide additional information regarding the efficacy of the classifiers based on the Gini Index and the Kolmogorov-Smirnov (K-S) statistics.