Major Depressive Disorder (MDD) afflicts 8 to 10 % of the world's population and is currently the 3rd largest cause of work place disability. Those with untreated or ineffectively treated MDD are over 20 times more likely to commit suicide than the general public. With a market size of about $17 billion, antidepressant medications rank 3rd among all classes of pharmaceuticals in the world and 2nd in the United States for number of prescriptions written. More than 10% of Americans and about 6% of Canadians are on these medications.
When correctly chosen, antidepressant medications can be very effective. However, according to one of the world's largest sequenced antidepressant treatment trials (the STAR*D trial), only 37% of patients will reach remission with the first drug chosen and only 31% of those still ill after the first medication trial will remit after the second . These support the common experience that, even applying current best practice methods, multiple drug trials are often necessary before good clinical effect can be achieved. This has major economic implications. The direct medical cost to remission is $3,600 when correct treatment is initiated from the beginning. However, these costs rise to an average of $16,000 if the initial treatment fails.
Our company has taken the unorthodox step of introducing engineering methods into the psychiatric domain. Our engineering colleagues have adapted a new mathematic method known as Machine Learning (ML) to address this issue. ML involves training a computer to recognize complex patterns in very large data sets in order to make intelligent decisions. Because ML technologies can examine and interpret thousands of variables, the discriminatory power of ML can exceed that of the human expert. At McMaster University we use ML to analyze clinical and laboratory data including electrical brain wave patterns (EEG data) to predict, in advance, whether a given person will response to a particular treatment.
Our pilot data show that we can predict, with 80-85 % accuracy, whether a depressed person will respond to the SSRI antidepressant sertraline, the antipsychotic drug clozapine or transcranial magnetic stimulation (a new physical treatment for MDD). We are currently extending our method to include cognitive behavior psychotherapy and the four antidepressant medications that, together, account for over 90% of the expenditures on antidepressant medications in the USA.
We have also adapted our ML algorithms to diagnose mental illness using the EEG. In pilot studies our algorithms are capable of differentiating, with 84 to 93% confidence, subjects with schizophrenia, major depressive disorder and the depressed phase of bipolar disorder from each other and from healthy volunteers.
The ML based digital psychiatric expert system that we have developed could assist the clinician to more accurately diagnose and treat mental illness. Our technology replaces the traditional impressionistic trial and error process with quantitative decision-making using data derived from the unique psycho-biological attributes of each patient. With this technology the health care professional could determine, from the outset, which of the numerous available treatments would be most effective for each patient. This would improve treatment efficacy, reduce personal suffering and dramatically reduce the cost of treating mental illness.
The patent applications describing our methodology have been allowed in the USA, and others are under review in Canada, Australia and the European Union.
. Warden D., et al., The STAR*D Project results: a comprehensive review of findings. Curr Psychiatry Rep, 2007; 9:449-459.
Dr. Gary Hasey, Chief Medical Officer
Dr. Jim Reilly, Chief Technical Officer
Bruno Maruzzo, Interim CEO
Dr. Hubert deBruin, VP Biomedical & Instrumentation
Dr. Ahmad Khodayari, VP Product Development
Dr. Duncan MacCrimmon, VP Electro Encephalography
What do we offer:
A Digital Psychiatric Expert System
Our machine learning based digital psychiatric expert system will offer the clinician a powerful new tool capable of assisting with the process of diagnosis and personalized treatment planning. Our technology would allow the health care professional to determine, from the outset, the specific type of treatment that would be most effective for an individual patient. This would reduce the time to recovery and dramatically reduce the impact of mental illness upon the individual and his or her family, the employer and the insurance provider.