Not all therapeutics are created equal in regards to individual patients. The problem of identifying which compound will work best with which patient is a significant burden across all disease contexts. In the context of autoimmune diseases such as alopecia areata, several formulations of JAK/STAT inhibitors have demonstrated efficacy in clinical trials. All of these compounds demonstrate different rates of response, and here we observed that this coincided with different molecular effects on patients undergoing treatment. Using these data, we have developed a computational model that is capable of predicting which patient-drug pairs have the highest likelihood of response. We achieved this by integrating inferred mechanism of action data and gene regulatory networks derived from an independent patient cohort with baseline patient data prior to beginning treatment.
Keywords: Alopecia areata; Artificial intelligence; Drug prediction; Personalized medicine; Systems biology.