OBJECTIVE: The identifying schizotypal trait in obsessive-compulsive disorder (OCD) patients is important to predict clinical course, since those patients are hardly overcome through conventional intervention methods. This paper presents the trial of classification method of obsessive-compulsive disorder with schizotypal trait using Frontal Lobe Function Test (FLFT).

METHODS: 110 OCD patients are divided into two groups:27 pure OCD patients, and 83 non-pure OCD patients. After training artificial neural network (ANN) using frontal-lobe function test data of train data (schizophrenia, pure OCD, and normal group), we classify test data (non-pure OCD patients) into one of the three groups.

RESULTS: Among the total 83 test data (non-pure OCD patients), 44 patients were classified as schizophrenia, 32 patients as normal, and 7 patients as pure OCD. With respect to the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) data of those classified patients, ordering score in compulsion was significantly different between three groups. Moreover, cluster A socre (Schizoid, Schizotypal) of Personality Diagnostic Questionnaire (PDQ) data showed significant difference between them.

CONCLUSION: The results presented that those OCD patients who are classified as schizophrenia using generated model with machine learning technique is tend to have compulsive symptom of arrangement and schizotypal personality disorder.