Although there are many anti-seizure medications (ASMs) on the market for epileptic patients, it is difficult to tailor these medications for individual epileptic patients. This study shows that AI can now be used to predict the most effective ASM for individual epileptic patients. This study comprised all the patients that visited the epilepsy center from 2008 to 2017, which were 2586 patients. The 2586 patients expressed different kinds of epilepsies and different etiologies. Focal epilepsy was the most common at 74.4% of patients and 43.5 % of patients had an unknown etiology. Overall there was a wide group of diverse cohorts used for the study. Generally, depending on the specific epilepsy in the patient, different medicine combinations were used and were varied in number as well. 8874 regimens were described overall and the average number of ASMs per person used was 2.87. To understand the efficiency of each regimen, these ASMs were categorized into 3 classifications: complete response (as in, no seizures), partial response (greater than 50% of seizures decreased), and a poor response (less than 50% of seizures decreased). 2388 regimens were complete responses, 196 regimens were partial responses, and 5124 regimens were poor responses. The dataset was analyzed separately and both Random Forest (RF) and CatBoost (CATB) were used to determine the highest area under the curve and the accuracy. The monotherapy regime was analyzed first, and the results showed that the AUCs varied by regimen and VPA had the highest AUC (out of the 5 common ASMs) of 0.686. Next, AI was used to analyze the dual regimens, which showed that the LEV and CBM had 0.709 and 0.764 for RF and CATB, respectively, suggesting a high prediction power. Overall, this study suggests that AI showed diverse results, but among the monotherapy regimens, the VPA showed better prediction performance using the AI. In the dual regimen therapy, the LEV and CBM were the best predicted by the AI. LEV plus OXC had low AUCs, which although seems like a disappointing result, was a good step towards using AI for understanding the effects of ASM medication on individual patients.