Can machine learning help us fight the global epidemic of depression?
How do we solve a problem like depression? On an individual level, one answer is personalized and effective treatment. But how do we work out what that treatment should be?
Introducing Aifred Health, the AI XPRIZE finalists using deep learning and neural networks to more accurately recommend treatments for depression patients. A team of 16 based mostly in Montreal, as well as the US and UK, Aifred are working with psychiatrists and computer scientists to develop a treatment system that relies less on trial and error and more on evidence and insights, to make smart predictions when it comes to care. Given that the World Health Organisation estimates that more than 264 million people of all ages suffer from depression globally, this technology has a big role to play in creating a happier, healthier society. Aifred believes it could also be expanded to help treat other mental health conditions beyond depression, too.
Below, David Benrimoh, Chief Science Officer at Aifred Health, tells us more about the team’s work.
Where is your team based and who is involved?
Most of us are based in Montreal, bar three team members in the US and 1 in the UK. We also work with collaborators in the U.S., U.K., and Israel. Our scientific advisory board is composed of two well-known psychiatrists (Drs. J. Karp and S. Parikh) and a respected computer scientist (Dr. K. Heller). The team overall is very interdisciplinary and brings together machine learning engineers, clinicians, healthcare leaders, and academics working towards a common goal.
Tell us more about that goal…
We are tackling the trial and error approach currently taken to treatment selection in mental health, starting with depression. We do this by addressing the challenge of matching the right patient to the right treatment for depression, in order to improve treatment success rates: getting patients better faster, and more patients better overall within a given patient population.
Walk us through your AI technology?
Our deep learning system uses clinical and demographic data from thousands of patients to learn patterns that help predict, quickly, at the point of care, and without expensive testing, the probability of a number of treatments helping an individual patient. This is then used by the patient and their clinician to support their decision making.
What made you want to enter the $5M IBM Watson AI XPRIZE?
It was actually the XPRIZE that inspired us to get our team together. A few of us had been working together on smaller-scale competitions, but the XPRIZE was a chance to do something big which could have a real impact on millions of people.
What's been your team's biggest challenge so far?
The biggest challenge so far has really been creating a clinical tool that delivers our technology seamlessly into the hands of clinicians and patients. It’s not a question of just making a state-of-the-art model and a clean app to house it; the AI needs to be built in a manner that is verified and validated and follows emerging health authority regulations; the application needs to be built in a very precise way to not interrupt the clinical workflow. In fact, doctors and patients seem to get the AI portion pretty quickly – the challenge is in presenting it in a way that makes it easy to use, and this is something we have spent a lot of time on getting just right.
Why is what your team is doing important now, and how do you see it scaling up in the future?
Over 300 million people in the world have depression at any one time, and only one-third of them will improve after their first treatment. What we are doing should help millions more people get better, faster; this approach can then be scaled to other areas of mental health for even greater impact.
How has the pandemic impacted your work?
Certainly, it's been a challenge not to be able to meet and work together in person, though already a lot of our work was online. There has been a tremendous opportunity for us however to help people because of the surge in depression related to the pandemic and its challenges. We have also been very glad to see how well our tool delivers support within a telemedicine/ virtual care environment.
Have you collaborated with IBM in any way? How has it impacted your work?
We have! IBM has provided very valuable help as we work on understanding the kinds of content patients are looking for, by using NLP on message boards. They have also helped us get access to high-quality health record data which will help us grow our model and fuel our innovation. This is a tremendous boost to our work today and going forward.
Overall, I think applying AI to not just predict but to model and understand these extremely complex conditions in mental health but also other fields, outside of mental health, is going to be really transformative, and that it will transform AI in return.
Learn more about the prize here