41 Expected to be the world’s biggest health burden by 2030, depression affects almost 322 million individuals globally. Yet despite decades of research and substantial funding, psychology has yet to understand the root causes of depression, let alone make progress toward a cure. Indeed, though available treatments have demonstrated an average level of efficacy, up to half of all patients who suffer from depression report no real benefit. Increasingly, clinical psychologists and academic researchers are seeking answers in personalized interventions. One promising approach is leveraging AI and machine learning to identify the specific sequence of events that lead to therapeutic improvement in individual patients; in turn, this knowledge can be used to develop more effective treatments on a large scale. At the forefront of this effort are Department of Psychology Profs. Dana Atzil-Slonim and Eva Gilboa-Shechtman, whose Psychotherapy Research Lab harnesses a vast, diverse dataset and collaborates with mental-health and AI/machine learning experts to uncover novel insights into depression-treatment strategies. “Years before it was fully embraced in mental-health research, [Bar-Ilan President] Prof. Arie Zaban insisted that data analytics are the future, and can revolutionize mental healthcare. He encouraged us to collaboratewithBar-Ilan data scientists Profs. YoavGoldberg and SharonGannot for the collectionand analysis of our clinical work, and almost five years, 2000 patients, and 30,000 psychotherapy sessions later, our lab’s initial findings only underscore his point,” says Atzil-Slonim. She explains that by integrating multi-modal data from psychotherapy sessions in the Psychology Department’s Community Clinic—including texts, audio, facial expressions, body movements, and even physiology, along with subjective reports from therapists and patients—the lab can identify patterns in patients and therapists' interactions that predict positive treatment outcomes. “These technologies allow us to zoom in on clinical treatment sessions and measure every aspect for its negative or positive effect,” says Atzil-Slonim, whose lab works with two of the world’s leading institutes for data science and AI: the Alan Turing Institute in London and the UKP lab in Darmstadt Technical University in Germany. “In one such project, we use state-of-the art natural language processing (NLP) techniques to automatically recognize in transcripts specific therapists’ interventions and patients’ responses, speech turn by speech turn, throughout the course of treatment. Then, we use data mining techniques to identify sequences of therapists’ interventions and clients’ emotional responses that are predictive of positive outcomes.” Another benefit of large enough datasets, continues AtzilSlonim, is the ability to cluster patients on the basis of predictive factors. By recognizing that patients with certain personal characteristics and behaviors tend to respond in certain ways to a given treatment, therapists can begin to personalize interventions for better and faster results. In the coming years, the Psychotherapy Research Lab hopes to validate its AI-derived best practices and methodologies in randomized clinical trials. In the long term, it also hopes to adapt its findings across different cultures and languages. Atzil-Slonimconcludes that psychology has only recently begun to realize the immense potential of collaborating with data scientists and software engineers. While computers can learn much faster than humans, she believes that rather than replace them, they will “complement therapists’ abilities to help their clients achieve better well-being.” Letting Data do the Lifting Psychotherapy Research Laboratory Prof. Dana Atzil-Slonim Prof. Eva Gilboa-Schechtman
RkJQdWJsaXNoZXIy NDU2MA==