The current applications for cognitive training only target memory improvement for older adults. The reason for developing this application, BrainTrain, is to make cognitive training through games more widely applicable, including in rural areas where the rate of MCI patients is up to 62%. Therefore, the cognitive games in BrainTrain will target various cognitive domains affected by MCI, including memory, attention, language, and math. In addition to designing the application, we will evaluate its effectiveness through neuropsychological tests and EEG data analysis to assess changes in cognitive performance and neurological activity in patients before and after intervention. From the analysis of these changes, we hope to elucidate the mechanism by which cognitive training games impact different cognitive domains in MCI patients. In terms of AI models, previous studies have used EEG data to build models to classify normal, MCI, or AD patients, or to detect abnormalities in the data to diagnose AD early. In this study, we will use EEG data combined with cognitive scores, demographics, medical history, and gaming data from patients in the first month of using BrainTrain to build a predictive model of intervention effectiveness. This predictive model will help doctors and patients save time in finding a more suitable intervention method.