Abstract
BACKGROUND: Obsessive-compulsive disorder (OCD) is a debilitating psychiatric disorder characterized by intrusive thoughts or repetitive behaviors. Clinicians use serotonin reuptake inhibitors (SRIs) for OCD treatment, but 40%-60% of the patients do not respond to them adequately. Here, we described an association rule mining approach for treatment response prediction using an Iranian OCD data set. PATIENTS AND METHODS: Three hundred and thirty OCD patients fulfilling DSM-5 criteria were initially included, but 151 subjects completed their pharmacotherapy which was defined as 12-week treatment with fluvoxamine (150-300 mg). Treatment response was considered as >35% reduction in the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score. Apriori algorithm was applied to the OCD data set for extraction of the association rules predicting response to fluvoxamine pharmacotherapy in OCD patients. We considered the association of each attribute with treatment response using interestingness measures and found important attributes that associated with treatment response. RESULTS: Results showed that low obsession and compulsion severities, family history of mental illness, illness duration less than 5 years, being married, and female were the most associated variables with responsiveness to fluvoxamine pharmacotherapy. Meanwhile, if an OCD patient reported a family history of mental illness and his/her illness duration was less than 5 years, he/she responded to 12-week fluvoxamine pharmacotherapy with the probability of 91%. We also found useful and applicable rules for resistant and refractory patients. CONCLUSION: This is the first study where association rule mining approach was used to extract predicting rules for treatment response in OCD. Application of this method in personalized medicine may help clinicians in taking the right therapeutic decision.