A pilot study at McLean Hospital used artificial intelligence to predict the therapeutic response to antidepressants: an approach that could be extended to many other therapeutic areas, in the future.

Big data and artificial intelligence (AI) represent a real turning point. Initially tested in drug design, AI now has been evaluated to determine, a priori, the therapeutic efficacy of a certain intervention.

Researchers at McLean Hospital (Massachusetts, a leading psychiatric hospital within the Harvard Medical School) investigated the efficacy of antidepressant drugs, and developed and tested an algorithm for predicting patient response before the beginning of the therapy. The results were published on Psychological Medicine.

Compare the prediction with the actual clinical data

Preliminary data acquired ex-ante on the clinical and demographic characteristics of participants in the Embarc multi-site clinical study on the efficacy of some antidepressant drugs, were used as a basis for the development of the algorithm. Its functioning was tested in a pilot study conducted in parallel to the clinical efficacy study of the drug, in which patients were randomized against placebo. According to the authors, in about a third of the cases the prediction of the algorithm on more responsive patients coincided with the clinical observation.

Artificial Intelligence: next steps

The next goal of American researchers is to extend the experimentation in real world conditions, where to compare the prediction of the algorithm to the real therapeutic response even in conditions of comparison of different drugs, or the use of antidepressants with respect to psychotherapy.

“These results bring us closer to the identification of patients who are most likely to benefit from selective serotonin reuptake inhibitor (Ssri) treatment and we could achieve the goal of customizing the choice of antidepressant treatment,” said the coordinator of the study, Madhukar Trivedi.

Possible impact on the market

Therefore the impact of such an approach could help revitalize the antidepressant market which, according to a recent report by GlobalData, has currently about thirty products, most of which are generic drugs. The last few years have seen the patent expiry of historical products such as duloxetine hydrochloride and aripiprazole, which have been offset by the launch of new products such as vortioxetine and brexpiprazole. The advanced pipelines also count six other products close to the market, all dedicated to the treatment of patients resistant to therapy.

Most important, the possibility of predicting in advance how the patient could respond to a certain drug would allow, according to GlobalData’s Neurology Analyst, Rahael Maladwala, to find an effective remedy for that third of patients who is not currently responding to the first drug therapy that is prescribed for depression. In addition to the optimization of care, the use of artificial intelligence for the correct prescription of drugs could have a major impact, both in terms of cost optimization and intervention, for health systems.