The ML model based on the combined genetic data and EEG data achieved better classification accuracy than using either alone. These results indicate that these two modalities might reflect somewhat different aspects of AUD etiopathology, and cannot replace each other in terms of portraying the disease, also confirming previous literature results showing the advantage of a ML model using multiple dimensions to classify a disease 3. Importantly, these results open the door to more personalized approaches to predicting diseases. Models based on different modalities can include features that change over time (i.e., brain structures and functions)40 and over human maturation (i.e., behavior and psychology)41 making it possible to focus on specific groups (such as categorization by age, gender, ancestry, FH, culture, and behavior) to create prediction models where individualization has real value to advance personalized care for patients.
