The Depression Meter Impact Center
Research Groups: Psychology, Computer Science, Engineering
Depression ranks among the three worst global health burdens. It is a mental disease that causes suffering and leads to suicide. At the social level, it is also related to increased healthcare costs and losses in productivity. In industrialized countries, including Israel, nearly 20% of women and 10% of men suffer from depression; yet only a fraction of those who suffer from depression seek treatment, and many who do, receive inadequate care. Despite the prevalence of the condition, our methods of diagnosing and treating it are poor — so poor that depression is on course to become the #1 global health burden in just 12 years.
At Bar-Ilan University, we are striving to radically improve diagnosis of depression by creating a new technology based on an enormous database and sophisticated Machine Learning techniques.
The Path Forward
We aim to make a depression meter, a system that will continuously gather linguistic, vocal, facial, kinesthetic, and physiological data from patients, and integrate the data to provide real-time feedback to patients and therapists on the severity of the patient’s condition.
The technology is based on machine learning techniques and using big data sets to “teach” deep artificial neural networks, like those used by Google, Facebook, Amazon, and others, to recognize depression and make subtle distinctions in its varieties and degrees.
Creating an objective measure for depression will transform diagnosis and allow for breakthroughs in treatment.
Bar-Ilan is uniquely positioned for such a challenge. Our Department of Psychology has an in-house treatment clinic, which is world-renowned for its rich clinical datasets. We possess a unique database comprising more than 9,000 clinician-annotated clinical hours. Working closely with our psychologists who are championing the center is a team from our Department of Computer Science.
Depression, moreover, is just the beginning. The Department of Psychology has treasure troves full of data on many conditions, and it is gathering more data every day. Once this method has been proven with depression, it can be applied to numerous other psychological and psychiatric conditions.
With its wealth of data and expertise in clinical psychology and computer science, Bar-Ilan is positioned to be an international leader in mental health technology.
At present, the team of the Depression Meter Impact Center is developing the tools needed to feed the huge data trove into an artificial neural network.
Prof. Eva Gilboa-Schechtman
Prof. Eva Gilboa-Schechtman is a clinical psychologist (who also has a BSc in Computer Science). Research conducted in her laboratory encompasses behaviour, hormones, brain imaging, physiological factors, and subjective experience. She has published extensively in the fields of psychopathology and psychotherapy and is a recipient of multiple grants, among them from the National Institute of Mental Health (US).
Dr. Dana Atzil-Slonim
Dr. Dana Atzil-Slonim is a clinical psychologist who studies process and mechanisms of change that underlie gains in psychotherapy. The research conducted in her lab advances the integration between research and practice in clinical psychology. Dr. Azil-Slonim specializes in tracing the process of psychotherapy with advanced machine learning techniques.
Dr. Yossi Keshet
Dr. Yossi Keshet spearheads the data-processing dimension of the project. Dr. Yossi Keshet is a Machine Learning expert. During his service in the Israel Defense Forces, he developed software that listened to conversations and identified keywords—a crucial piece of technology for gathering intelligence.
For this work, he won Israel’s highest prize for contributing to our nation’s security. Dr. Keshet has used his expertise in Machine Learning to develop software for numerous medical diagnostic purposes and to extract valuable information about the identities of telephone callers, including age, gender, ethnicity, height (within 1 inch of actual height), weight (within 10 pounds of actual weight), body-mass index, accent, and various environmental factors.