Aifred Health

Bringing personalized medicine and advanced decision support to mental health practitioners and patients with the power of AI.

1 in 9 people will suffer from depression, making it the most important cause of disability globally. However, mental healthcare often fails the patient. Psychiatric illnesses need to be treated on an individual basis, yet treatment selection still relies on a trial and error approach. This method often provokes side effects, and can be draining and ineffective - the longer someone is depressed, the greater the chance they remain depressed, relapse in the future, or commit suicide. Aifred Health is committed to solving this issue.

The Aifred Health solution is a clinical decision aid tool built with machine learning techniques that will help clinicians choose the most effective treatments for mental illnesses, starting with depression. Only 30% of patients with depression improve after the first treatment, and 60-70% improve after the fourth treatment. It can take months or years to find a treatment that works for the patient with minimal side effects. Using insights gained from individual patient characteristics, our application will reduce time to remission while minimizing side effects.

To achieve these goals, we created a model trained on large datasets which include all the factors identified as predictors of treatment outcomes in depression: information about socio-demographic factors, symptom profiles, initial symptom severity, treatment response profiles; we are also actively adding genetic, metabolic, endocrine, immunological, and neuroimaging data.

By equipping healthcare workers with a decision aid tool, better decisions can be made to get patients better, faster - especially in primary care. Due to resource limitations, most patients do not have access to psychiatric consultations. However, our app makes psychiatric knowledge available to all mental healthcare workers. Due to its affordability and accessibility as a desktop as well as a mobile app, Aifred Health will bring high quality expertise to resource-constrained healthcare environments around the world and provide support to frontline healthcare workers.

Our app will drastically increase treatment success rates and decrease remission time from 7 or more months to as little as 2 months by 2020. We also predict a reduction of cost from $7K to $2K/patient. Within our first 12 months, we expect to serve 20-30K patients with depression and in 3-5 years, over 200K patients.

We will launch the first version of our product, Aifred Lite, this fall. We will then undergo an ease-of-use study and an open-label trial. This will improve our tool, provide us with clinician feedback and allow us to test our model’s safety and effectiveness. By 2020, we will launch our product, Aifred Premium, loaded with our full AI model. Following this step, we will replicate our model to other mental illnesses such as OCD, anxiety, psychosis, and addiction.

Due to the immense burden of mental illnesses on individuals and on the healthcare system, we firmly believe that the way patients are treated must change. Our solution will change the field of psychiatry as we know it - not only improving treatment selection and outcomes for patients around the globe but also bringing new insights into patient data to push treatment research forward.

Aifred Health is the first company to apply deep learning to treatment decision support in mental healthcare. Our app eliminates the trial and error process that is long, painful, and regretfully today’s norm in mental health treatment. Instead, our tool uses insights from AI models to assist doctors by predicting which treatment is most suited to each patient based on an individual’s predictors of treatment outcome. Our decision support will not only be useful to psychiatrists - it will also equip GPs and nurse practitioners to successfully treat mental illnesses, increasing the accessibility of effective treatments.

The deep learning framework we created, Vulcan, is integral to our solution: it predicts which treatment has the highest likelihood of success for each patient, optimizing quality of care. The simplicity of the app format easily integrates the existing workflow. Equipping frontline workers such as GPs and nurse practitioners with tools to treat in a more informed and efficient way allows for a greater coordination of care, and for more patients to be seen and successfully treated. Finally, because our app has a mobile version, it will be accessible in developing countries, where computers and tablets are less common.

We expect to release Aifred Lite in 3 months, an early product that will allow us to collect data to further fuel our model. In 6 months, we will begin clinical trials for Aifred Premium in a healthcare system with over 300K patients, the results of which will improve and validate our model. User feedback will optimize the UX. Simultaneously, we aim to publish papers to inform and educate clinicians and researchers on the important predictors of treatment outcome in mental illnesses and the functioning of our technology. These activities will help support training and development of frontline workers.

By 2020, we will have tested and begun distributing Aifred Premium to replace Aifred Lite - which should have served greater than 100,000 patients by that point. Depression affects 322 million people, so there is a concrete need for frontline workers to select better treatments that will drive adoption. We will also grow our tool by expanding to different psychiatric conditions and countries, thereby impacting more people. This transition of psychiatric treatment paradigms is particularly needed in resource constrained environments, cutting costs while improving care quality so that more patients can be successfully treated by frontline workers.

We will focusing on securing partnerships with forward thinking and technologically innovative healthcare providers, who will help us access data, clinicians, and patients. Our high quality product will not only retain clients by increasing their work efficiency and treatment success rates but by also continually improving and expanding to other conditions. Our product is a downloadable app available through a SaaS model. It will be available on desktop as well as mobile interfaces, the latter allowing for easier access within resource-limited regions (we will be working on an SMS-based version).

We are currently in the R&D phase, however we have already signed a contract with a major Canadian telemedicine company that serves many major employers, and are in discussion with several other companies. Our client will soon have access to a treatment selection algorithm that will sort their patients into the appropriate stream of care, faster. The population we plan on serving includes clinicians and large healthcare organizations. While using our app, will see a decrease in time to remission (7 months to 2 months, on average) and the number of medical visits (13 to 4 visits, on average).  

We serve all non-pediatric patients with depression and their clinicians. Within our first 12 months, we expect to serve 20-30K patients with depression and in 3-5 years, over 200K patients. Our AI model is efficient and precise, enabling a decrease in remission time from 7 months to as little as 2 months by 2020. Non-AI techniques can reduce treatment time by up to 50%, whereas our AI model aims to push this to 70-75%. Our app will reduce strain on limited resources, providing support to frontline workers while increasing treatment accessibility.

Our interdisciplinary team uses the collective expertise of doctors and allied healthcare professionals, neuroscientists, software engineers, computer scientists, and other scientists from various disciplines. Due to the combined skills of our highly trained machine learning and research teams, we obtained accuracies between 70-80% when predicting remission to several different treatments. Our experienced business team also provides our company with a great potential for growth: we have already partnered with universities across the USA, Canada and Brazil and a major Montreal hospital.

Our app is subscription based following a Software as a Service (SaaS) model. We will be targeting healthcare providers (health management organizations, telehealth companies, hospital networks and public healthcare systems), clinicians reimbursed by healthcare insurers and individual clinicians paying out-of-pocket. The cost of our app will be paid off by seeing only two psychiatric patients, making it worthwhile for physicians to purchase a license, especially with the treatment benefits they will observe; larger organizations will see cost savings and increased value that will make purchasing a subscription profitable. We will invest a significant portion of our profits towards growth and expansion beyond North America and Europe. This will allow frontline workers operating in more resource-constrained environments to benefit from our app. We also expect to replicate our model in other mental health conditions such as, but not limited to, anxiety, psychosis, OCD, and addiction which will increase the scope of the app and will suit the needs of a larger number of clients. Most importantly, all of our models will continuously improve and be updated with new results and treatments, as well as with region-specific insights from the data, which will make our SaaS a true service.

As a new startup, we are constantly seeking to improve our product. Entering the Solve community would allow us to obtain precious feedback about our tool as well as high-quality mentorship from experts in our field. This collaboration would allow us to better tailor our AI-powered decision aid to the workflow of frontline workers such as GPs, psychiatrists and nurse practitioners. The valuable advice we would receive would help us further our mission of enhancing frontline care in mental health, improving the quality of life of both patients and healthcare professionals.

Although we are confident in our innovative product and ideas, we recognize that many startups are unsuccessful because they fail to break through the market. Solve can support our product by connecting us with individuals who are experienced in launching products. This way, we can successfully market our product and establish ourselves as a profitable company. Solve will allow us to enter a community of highly skilled and resourceful people, who can mentor us and provide us with valuable advice which will allow us to refine our action plan, grow our company’s exposure, reach investors and advance our solution.