In this paper, Li et al. explored how efficiently AI techniques can classify prediction when combined with functional magnetic resonance imaging, in order to determine if bringing them into clinical environments would help in the diagnosis process. In order to complete this evaluation of various algorithms, Li et al. brought in 64 participants, each that suffer with either migraines with aura, migraines without aura, or Hemicrania Continua, as being able to quickly differentiate between the three is pivotal to improving the diagnosis process. Neuroimaging was completed for each of the participants using an MRI system that scanned their brains for 6 minutes, while they were instructed to close their eyes, remain awake, and avoid active thought. Li et al. used different algorithms in conjunction to acquire and process data from the brain scans that were taken from the participants. Values like the regional intrinsic neuronal activity (ALFF), and regional functional connectivity strength (RFCS) were calculated using these AI models. After comparing all the statistical data received from the AI models, Li et al. found that all the models tested exhibited an improvement in classification accuracy from current classification techniques by greater than 84%. The AI model “GoogleNet” was deemed the boat accurate out of the three that were tested in this study, and when used to find RFCS - the best indicator of classification - researchers were able to get the most efficient outcome. Future research is underway to obtain more data from outside sources to determine if these AI models stay consistent across various populations.