Aifred Health Files for IPO: Leveraging Biological Data and AI to Personalize Depression Treatment

Globally, approximately 300 million people suffer from depression. Although many effective treatments for depression are currently available, there is no reliable method to help physicians determine the optimal therapy for each individual patient.

For years, many patients have battled depression through a process of “trial and error,” enduring the side effects of medications and the anguish of despair. “We are committed to identifying better treatments for patients with depression,” said Robert Fratila, Co-Founder and Chief Technology Officer of Aifred Health, a Montreal-based startup.

The company is leveraging GPU-accelerated deep learning to predict optimal treatments based on patient symptoms, demographic information, and specific medical test results. Currently, it ranks among the top ten contenders in the ongoing $5 million IBM Watson AI XPRIZE competition for its achievements.

Individuals with depression not only experience persistent feelings of dejection, but in the most severe cases, depression can lead to suicide. During depressive episodes, persistent sadness or hopelessness can affect many aspects of life, including interest in daily activities, sleep, appetite, and even concentration.

To treat patients, psychiatrists may choose from dozens of antidepressants or various types of psychotherapy. For patients with severe conditions, doctors may even opt for brain stimulation techniques. According to Fratila, physicians select treatment methods based on their experience and medical guidelines, but there are no objective decision-making criteria.

Aifred Health is dedicated to integrating more technology into treatment methods to help physicians tailor treatment plans for each patient. “Our R&D philosophy is to identify the right treatment for patients as early as possible, enabling them to recover more quickly. This will also reduce the medical costs associated with depression,” pointed out Fratila.

For most diseases, doctors can use medical tests such as MRI (magnetic resonance imaging), X-rays, or blood tests to develop treatment plans and monitor patient responses. Although no such diagnostic tests currently exist for depression, a growing body of research suggests that neuroimaging, genetics, and other biological factors may help clinicians select the optimal therapy.

Researchers at Aifred Health have integrated research data with patient demographics, symptoms, and medical histories to develop deep learning-based software that assists physicians in creating personalized treatment plans. Using their own cuDNN-accelerated deep learning framework and NVIDIA GPUs on IBM Cloud, the researchers trained their neural networks on data from the U.S. National Institute of Mental Health and other sources.

Furthermore, the company is one of the 14 teams participating in the SMART Mental Health Prediction Challenge, a competition designed to identify the best predictive models for treatment response in anxiety and depression.

Fratila stated, “With Tesla GPUs, we can spend less time on model training, thereby allowing us more time to consider how to improve network performance.”

Aifred Health was founded by five students from McGill University in Canada and is currently collaborating with medical experts and other data scientists from five universities.

After additional training with more data, the company plans to collaborate with physicians on a series of trials to test its algorithm against standard guidelines for prescribed therapies. The trials will explore the performance and safety of Aifred’s software, as well as its ease of use by clinicians.

The software’s design will include checking for drug interactions, health risks, and side effects, as well as assessing how frequently patients need to adjust their therapies. “This software is not intended to replace doctors, but rather serves as a data-driven tool to deliver better healthcare,” said Fratila.

Source: NVIDIA's Official WeChat Account