Using artificial intelligence ( AI ) to improve the lives of people living with mental illness.
Our digital psychiatric expert system is a powerful new tool for personalized treatment of serious mental illnesses.
We use artificial intelligence to analyze the brain electrical activity patterns* of an individual to more accurately establish diagnosis, determine optimal treatment and assess suicide risk.
*We measure brain activity using the electroencephalograph or EEG - a non-invasive test that can be done in a hospital lab or a medical office in 20 minutes or less.
Our Services
-
Diagnostic Assessment
Our AI software uses a single EEG to diagnose Major Depressive Disorder (MDD), MDD with Psychosis, Atypical MDD, Bipolar Disorder (BD) -Mania, BD-Depression, and Schizophrenia with over 90% accuracy.
-
Suicide Risk Assessment
Our AI software uses a single EEG to detect the presence of suicidal ideation with 75 - 90 % accuracy.
-
Treatment Response Prediction
Our AI software uses a single EEG to predict response to medication, psychotherapy and brain stimulation with 79-90% accuracy.
The Challenges We Face
While clinician training and experience are crucial determinants of treatment outcome, in the absence of accurate and objective biomarkers, current “best practice” is driven entirely by personal interview and observation. However, the symptoms of different psychiatric illnesses can show significant overlap, making accurate diagnosis difficult. For example, 40 % of those suffering from bipolar disorder (BD) depressed phase are misdiagnosed with major depressive disorder (MDD) [1]. The antidepressant medications which would be appropriate for MDD may induce mania or rapid cycling , mixed states and increase the risk of suicide in the person with BD.
To further compound the complexity, even those correctly diagnosed with MDD may have any one of several biological subtypes, each requiring a different form of treatment. As there is no accepted way to detect these biological subtypes, treatment choice becomes a matter of guess work or “serial trial and error”. This deficiency in our current “best practice” is illustrated by the large Star*D clinical trial, which demonstrated that only 37% of those treated with antidepressant medication will remit after treatment with the first antidepressant medication chosen [2].
Many of those who are afflicted with a depressive illness experience suicidal thinking. Suicide is the 2nd most common cause of death in those under 35. This may be partly because current methods of suicide risk assessment are highly inefficient. After reviewing the last 50 years of research on the subject, Franklin et al concluded that current suicide risk assessment methods are highly unreliable yielding predictive accuracy “slightly better than chance" [3].
The symptoms of different illnesses overlap making correct diagnosis challenging
Without relevant biomarkers medications are chosen by “serial trial and error” and the remission rate is only 37% with the first choice
Current suicide risk assessment accuracy may be only “slightly better than chance”
Our Technology in Action
Learn how our technology can easily be used in the clinic with an office-based portable EEG headset.
Podcast Interview
The Journal Of Clinical Neurophysiology interviews DME’s Dr. Maryam Ravan & Dr. Gary Hasey to learn how DME’s technology can discriminate between bipolar and major depressive disorder using AI.
Our Solution
In partnership with St Joseph’s Hospital and McMaster University Digital Medical Experts Inc. (DME) has developed artificial intelligence (AI) tools for enhancing psychiatric practice efficiency. Our interdisciplinary team, including experts from McMaster University, University of Toronto, and New York Institute of Technology, created a non-invasive and cost-effective method for assessing an individual's neuro-psycho-biological attributes. Trained using the data collected from hundreds of individuals, our AI algorithms analyze electrical brain wave activity measured using the electroencephalograph or EEG, to uncover hidden patterns, providing valuable insights for personalized treatment planning. DME's research indicates that our AI algorithms can offer clinical management advice with an accuracy matching or surpassing human psychiatric experts, particularly in diagnosing mental disorders, identifying suicidal ideation, and predicting treatment responses.
Our software can:
Differentiate persons with different forms of bipolar disorder, major depressive disorder, and schizophrenia from each other and from healthy individuals with 90% accuracy [4,5].
Detect suicidal ideation using questionnaire data with 75 % accuracy [6] and with up to 89 % accuracy using EEG analysis (soon to be published).
Calculate the probability of response to different antidepressant medications [7,8], the antipsychotic clozapine [9,10,15,16] cognitive behaviour therapy [unpublished] and transcranial magnetic stimulation [11] with up to 90% accuracy.
Our AI-based process accurately diagnoses illness, assesses suicide risk and predicts treatment response with a single 20 minute EEG
Predicting Treatment Response
The current “serial trial and error” method of choosing treatment is highly inefficient. In most trials of antidepressant medication (ADM), psychotherapy and brain stimulation such as repetitive transcranial magnetic stimulation fewer than 50% of those treated will show full remission of depression. Often several unsuccessful trials are required before an effective treatment is discovered. In contrast, DME’s technology can predict, for the individual depressed person whether a particular form of treatment will be effective or not, allowing the clinician to prescribe the best intervention from the outset. To date our algorithms have been trained to use pretreatment EEG to different antidepressant medications ADM [7,8], cognitive behavioural psychotherapy (CBT), repetitive transcranial magnetic stimulation (rTMS) and the antipsychotic clozapine [9,10,15,16] [11] with up to 79-90% accuracy. DMEs technology can even predict response to placebo [8].
Our published reports showing that AI analysis of EEG can be used to accurately predict response to ADM have been confirmed by independent research groups [12, 13]. In a recent survey of all published studies examining the use of EEG to predict ADM response [14] 4 of the 15 identified to be of the highest quality were written by our group.
Cost Savings
We estimate that the improved diagnosis and treatment selection delivered by our technology would reduce workplace disability by nearly 5 weeks and result in savings of $8000 in income replacement and medical costs for every person treated.
DME algorithms can predict response to different forms of treatment with 79-90% accuracy
Pre-treatment EEG be used to estimate how much improvement CBT will bring about.
Clozapine Treatment
The antipsychotic drug clozapine can be highly effective in persons with treatment-resistant schizophrenia, significantly improving recovery, reducing direct medical costs (22), hospitalizations (23), and mortality from self harm and other causes (24). However, a review of scientific literature concluded that clozapine is underused, in part because of concerns over the side effects (25). These can include dangerously low white blood cell count, weight gain, elevated lipids and increased blood sugar. DME has developed AI algorithms that use the pre-treatment EEG to predict, with 85 - 90% accuracy (10,15,16), whether clozapine will be effective in a particular person with schizophrenia. This test could provide information highly relevant to informed assessment of the advantages and disadvantages of clozapine treatment for a given individual.
Pharmaceutical Testing
Furthermore DME has also identified particular changes in EEG activity noted only in those who improve after treatment with clozapine has started (9). This analysis may be useful in screening new candidate drugs by pharmaceutical firms who seek to develop a new antipsychotic medication that possesses the effectiveness of clozapine without the side effects. DME’s placebo response prediction algorithms (8) may also be of utility in the efficient testing of new pharmaceuticals. The true biological effect of a new pharmaceutical can be more easily demonstrated in a smaller, and therefore less costly, clinical trial if placebo responders are identified.
Assessing Suicide Risk
Currently, clinicians using traditional methods can assess suicide risk with only about 50% accuracy. This may be because 70% of those who die by suicide will deny having suicidal ideation (SI).
As DME can detect SI using only EEG data those at greatest risk could be identified without need for verbal disclosure .
Detecting Suicide Ideation Using AI
In pilot studies our system can detect SI with 75-90% accuracy.
In a preliminary study of 60 persons with MDD, our AIAs detected the presence of suicidal ideation with 89% accuracy using only EEG data (unpublished).
Our AI algorithms (AIAs) may also detect covert SI using a written questionnaire. Using data collected from 800 military and police veterans our AIAs detected suicidal thinking with 75% accuracy using 10 simple questions, none of which asked about suicide [6]. The separation of those with and without SI is shown in the figure to the left .
How our System Works
When a patient presents at a physician’s office or emergency room with complaints of a psychiatric nature, the physician orders an EEG or does one on site with portable equipment. The EEG data are sent, via the cloud, to our analytic centre where they are preprocessed then analyzed by our automated algorithms. A medical report (see example to the left) containing diagnostic possibilities, suicide risk assessment and treatment options with response probabilities is electronically submitted to the physician within one hour of testing.
To date our EEG data have only been collected using standard laboratory EEG devices. However, in DME’s preliminary testing wireless, dry electrode portable EEG headsets that can be used in any office serve equally well.
Our suite of AI algorithms (AIA) have been trained using EEG and other data collected from over 2000 carefully diagnosed and treated psychiatrically ill patients and healthy volunteers. Using signal source localization techniques, brain connectivity measures (see figure 2 below) and other techniques our AIAs have discovered EEG patterns that can be used to make highly accurate psychiatric diagnoses, sensitively detect suicidal ideation and predict response to different types of treatment. With the appropriate consent, EEG, treatment outcome and other data from newly tested patients are continually being added to our database so that our AIAs can constantly improve as greater amounts of training data become available.
Our current training data set includes information on i) diagnosis, ii) personality, iii) social support, iv) symptom severity, v) physical and mental disability, vi) cognitive functioning, vii) suicide risk, viii) response to different forms of treatment.
The technology DME has developed is described in issued patents in the USA, Canada, and Australia.
<1 hour
Diagnosing Mental Illness
Symptom overlap across different diagnostic categories makes accurate diagnosis challenging. Distinguishing bipolar disorder-depressed phase (BD-D) from major depressive disorder (MDD) is particularly difficult and may take years of observation. The graph to the right illustrates how our algorithms have been trained to distinguish persons with MDD from those with BD-D with 90 % accuracy using a single EEG [5] [26]. Our software can also differentiate several other diagnostic categories including MDD, MDD with Psychosis, Atypical MDD, BD-Mania, BD-D, and Schizophrenia from each other and from healthy volunteers with over 90% accuracy.
In the graph to the right, persons with Bipolar Disorder-depressed phase (BD-DE) are separated in hyperspace from those with Major Depressive Disorder (MDD)
90 % accuracy using a single EEG
Why Use AI in Psychiatry?
In a recent interview, Dr. Gary Hasey, Digital Medical Experts founder and Chief Medical Officer explains how machine learning, a subset of artificial intelligence, can be used to enhance a physician’s ability to make a more accurate diagnosis and determine optimal treatment of serious mental disorders.
Our Goals
-
1) Improve Lives
DME will improve the quality of life of people living with mental illness through enabling more effective treatment and more sensitive suicide risk assessment.
-
2) Accelerate Recovery
DME will improve diagnostic accuracy and treatment efficacy to accelerate recovery, reduce workplace disability and decrease health care costs.
-
3) Provide Objectivity
DME will reduce physician reliance on subjective symptom reporting by providing an accessible assessment of the neurobiological functioning of each individual patient.
-
4) Enable Remote Care
DME will introduce a digital "virtual psychiatrist" that will provide psychiatric management advice to remote and under-serviced areas with accuracy that duplicates or exceeds that of the human expert psychiatrist.
Consumer Product
DME offers a consumer product designed for use by persons other than physicians to identify at-risk individuals in the workplace, the military and in academic institutions where less than 50% of those with mental illnesses are recognized (click on “schools” “military” and “transportation” below for more information). For consumer market use the EEG would done on-site using commercially available portable equipment that would automatically the EEG signal to DME for analysis. The test subject’s EEG profile would be compared with those in our database of several hundred healthy as well as psychiatrically unwell persons. A report would be submitted to the institution or the individual.
The Consumer Report
The report would depict the test subject’s EEG profile (shown as the black bar in the bar chart to the right) along 4 color-coded reference dimensions, each dimension being defined by the EEG profile of healthy or psychiatrically unwell persons in our database. The green zone represents the range of EEG profiles seen in our healthy database. The red zone represents the range seen in our database of persons with various psychiatric conditions or those with suicidal ideation. The amber zone is intermediate between the two. If the tested individual’s profile is most similar to that of the green zone, then that individual is likely psychiatrically well. In contrast, a subject whose profile is most similar to that of the red zone may be suffering from a psychiatric illness and would be strongly advised to seek professional psychiatric assessment. This quantitative assessment could assist school guidance counselors, military health screeners, and employee health professionals to identify at-risk individuals. In the case of employees responsible for the safety of others e.g. airline pilots, train engineers or bus drivers, identification of at-risk individuals could have public safety implications.
Regulatory approval would not be required as this system does not produce a psychiatric diagnosis nor recommend specific treatment. It is merely a statistical estimate of brain activity “normality”.
-
To Identify & Prevent:
According to Centers for Disease Control “19.9% of U.S. public and private school students in grades 9-12 had seriously considered attempting suicide, and 9.0% had attempted suicide”. At the University of Toronto, safety barriers were installed after several student suicides. [17,18]
-
To Identify & Prevent:
In 2021, research found that 30,177 active duty personnel and veterans who served in the US military after 9/11 have died by suicide compared to 7,057 service members killed in combat in those same 20 years [19]. 71.6% who died by suicide previously denied having suicidal ideation [20].
-
To Identify & Prevent:
Andreas Lubitz, purposefully crashed the commercial airliner he was co-piloting into a mountain killing all on board. He had previously been treated for suicidal tendencies but kept this information from his employer. Instead, he reported for duty [21].
For the Investor
To date, the scientific work which DME is commercializing has been supported by several awards and non-dilutive scientific grants funded by Provincial, National and International granting agencies, by St Joseph’s Hospital, McMaster University, the Magstim Company, the CGX company (in kind donation) and private donations by individuals. The costs related to Intellectual property protection and other commercialization processes have been supported by the McMaster University Liaison Office and the cofounders.
DME is actively seeking investment from the private sector. We believe that our technology -
has been scientifically validated
fills an important gap in the current practise of psychiatry
serves a very large international market
is easily scalable
News & Highlights
-
![]()
DME scientist Maryam Ravan discusses importance of brain research in AI in 2022 Lifewire article
-
![]()
DME featured in IEEE Spectrum article
-
![]()
DME wins First Prize in 2018 Universal Innovation Competition
-
![]()
DME Wins Toronto 2019 Science Innovation Award
Dr. Gary Hasey
Founder, Chief Medical Officer
-
DME Founder. Associate Professor, Department of Psychiatry and Behavioural Neurosciences, School Of Biomedical Engineering, Department of Electrical and Computer Engineering, McMaster University. Director of the ECT and rTMS Program St Joseph’s Hospital.
Dr Hasey MD, FRCP(C), MSc: Dr Hasey is a practicing psychiatrist whose clinical and research work in Mood Disorders allowed him to identify the “pain points” in current psychiatric practice. Dr Hasey is the author or co-author of 4 patents and numerous psychiatric papers and book chapters. In 1997 he opened Canada’s first therapeutic transcranial magnetic stimulation (TMS) clinic at McMaster University where he treats patients with Mood Disorder and PTSD using brain imaging guidance to enhance the effectiveness of TMS. Partnering with electrical and computer engineers his group developed artificial intelligence algorithms to improve the management of psychiatric disease. In 2014 he received the RO Jones Award from the Canadian Psychiatric Association for the paper entitled: Building a Virtual Psychiatrist: Using Digital Technology to Assist With Diagnosis and Treatment Planning”.
Bruno Maruzzo, MASc, MBA
CEO
-
CEO. Bruno is an Electrical/Biomedical engineer with an MBA. He has many years’ experience with small/medium sized companies in technology, medical devices, in vitro diagnostics and biotech, in various capacities including general management, business development, corporate development, fund raising and technical areas. He also worked in the venture arm of a major insurance company where he sourced, negotiated and made investments in technology, healthcare and biotech. At the University Health Network in the tech transfer and commercialization offices he was involved in licensing out technology and start-up formation. He currently sits on the boards of 2 public and one private company. Being on both sides of the table in transactions and negotiations at different times has given him great insights into company formation, financing and management.
Dr. Jim Reilly, PhD
Founder, Machine Learning Scientist
-
DME Founder. Professor, Department of Electrical and Computer Engineering, School of Biomedical Engineering, McMaster University.
Dr Reilly BASc, MEng, PhD, is an electrical engineer and internationally renowned researcher whose ground-breaking work is in the area of artificial intelligence (AI) applications and brain imaging including Electroencephalography (EEG). He has published widely in the area of signal processing at large and has co-authored pioneering papers on machine learning analysis of the EEG for the treatment of major depression and schizophrenia. He is a member of the IEEE MLSP Technical Affairs Committee and has enjoyed extensive collaboration and funding from industry, through which he has obtained 9 patents.
Dr. Hubert DeBruin, PhD
Founder, BioMed Engineer
-
DME Founder. Professor Emeritus, Department of Electrical & Computer Engineering, School of Biomedical Engineering, McMaster University.
Dr DeBruin P.Eng. PhD, Developed McMaster's original Electrical and Biomedical Engineering program. His research interests include the development of new computer-based systems and techniques for recording and analyzing physiological signals in the basic research and clinical medical laboratories. His work has involved basic and clinical studies of nerve and muscle stimulation in animals and humans as well as basic and clinical studies in trans-cranial magnetic stimulation (rTMS) in humans. He is particularly interested in EEG signal processing and machine learning in neuropsychiatry.
Dr. Ahmad Khodayari, PhD
Founder, Machine Learning Scientist
-
DME Founder. Staff Data Scientist, Walmart Global Tech.Dr Khodayari-Rostamabad PhD, Former PhD student under the supervision of Drs Reilly and Hasey conducted our earliest machine learning analyses of EEG to predict response to clozapine, SSRI antidepressant and repetitive transcranial magnetic stimulation and to determine diagnosis. Dr Khodayari employs machine learning, deep learning, data analytics, and statistics to solve problems in fraud detection, trust and safety, risk Assessment and eCommerce in the corporate sector. He maintains an active interest in DME and it’s technology.
Dr. Duncan MacCrimmon
Founder, EEG Expert
-
DME Founder. Associate Clinical Professor, Department of Psychiatry and Behavioural Neurosciences, Department McMaster University (retired).
Dr MacCrimmon, MD, FRCP(C), is a psychiatrist and researcher expert in electrophysiology with a special interest in the application of EEG technology in the context of the treatment of schizophrenia with clozapine. Dr MacCrimmon was a pioneer in the use of quantitative EEG analysis in neuropsychiatry.
Dr. Maryam Ravan, PhD
Machine Learning Scientist
-
DME Scientist. Assistant Professor New York Institute of Technology (NYIT).
Maryam Ravan is currently an Assistant Professor in the Department of Electrical and Computer Engineering, New York Institute of Technology (NYIT). Prior to that, she was a Senior Research Scientist in the Department of Research and Development at LivaNova PLC (formerly Cyberonics Inc.), a post-doctoral fellow at University of Toronto in collaboration with Raytheon Canada Ltd and Defense Research and Development Canada (DRDC), and a post-doctoral fellow at McMaster University in collaboration with Mood Disorders Program in Centre for Mountain Health Services, St. Joseph Hospital. She has authored/co-authored over 90 journal and conference papers, one book chapter, and a book. She was a recipient of numerous awards including NSF, DND/NSERC Research Partnership Grant, MITACS Internship, NSERC Engage Grant. She is a senior member of IEEE. Her research interests include signal processing and machine learning with applications in biomedical signal analysis (Electroencephalography (EEG), Electromyography (EMG), and Electrocardiography (ECG)), microwave imaging, microwave sensing, and wearable technology. She will co-direct and supervise graduate students who will develop and refine machine learning (ML) algorithms to diagnose illness, predict responses to specific psychiatric treatments, and detect suicidal ideation.
Dr Kevin Ming PhD
VP of Business Development
-
Kevin received his B.Sc. in Electrical Engineering from Queen’s University, M.A.Sc. in Electrical Engineering from University of British Columbia, and Ph.D. in Biomedical Engineering from the University of Toronto, where he developed and patented a smartphone-based point-of-care infectious disease diagnostic device. He then joined a government-funded regional innovation centre to work with and advise over 200 technology start-ups, as its Executive Director. Kevin then went on to join a publicly-listed enterprise software company, where he developed go-to-market strategies and oversaw various corporate strategic initiatives as its Director of Growth. He is currently the R&D Chief of Staff at a medical device start-up, as well as serving as the VP of Business Development at Digital Medical Experts.
Dr. Sinisa Colic, PhD
Machine Learning Scientist
-
DME Scientist. Assistant Professor, University of Toronto.
Dr. Colic PhD, Completed his PhD at the University of Toronto in the area of personalized treatment options for epilepsy using advanced signal processing techniques and machine learning. After that Sinisa was a post-doctoral fellow supervised by Dr Reilly and Hasey at McMaster University where he worked with medical imaging data for the diagnosis and treatment of mood disorders. To date Dr. Colic has contributed to a number of refereed publications, conference proceedings and presentations in the field of biomedical engineering. He was awarded best paper at the 35th Annual Conference of the IEEE EMBC in Osaka, Japan for his work on cross-frequency coupling for characterizing seizure-like events. In 2018 he was awarded the Digitech Innovation Prize in Paris, France for the commercial work on developing an EEG-based system for the management of major depressive disorder. Dr Colic has taught several courses at University of Toronto covering a broad range of topics in mechatronics and machine learning. Formerly post doctoral fellow under Dr Hasey and Reilly's supervision he will coordinate the data acquisition using portable EEG devices, design the cloud-based transfer protocols, and support the development of a ML pipeline for standardization and deployment of ML algorithms for real-time detection and diagnosis of psychiatric disorders.
Dr. Nick Kates
Expert in Community Psychiatry
-
DME Advisor. Professor, Department of Psychiatry & Behavioural Neurosciences, McMaster University.
Dr Kates MD, RCPSC Specialist, also is the former chair of the Department of Psychiatry and Behavioral Neurosciences at McMaster University: As a psychiatrist and clinical educator Dr Kates's special interest is the interface between primary care and psychiatry. His expertise in optimizing the quality of psychiatric management in the community setting will help us to better understand the psychiatric needs and “pain points” in the primary care setting. This knowledge will assist our group to develop a technology and user interface that are useful and easily integrated into community practice.
Our Team
Joseph Agostino, LLb
Corporate Lawyer, Advisory Board
-
DME Advisor. Joseph Agostino, was formerly General Counsel at Hydro One Inc.
Joseph Agostino, LLb. Hydro One Inc. is Ontario’s largest electricity transmission and distribution service provider. As the leader of their legal department Mr Agostino has extensive expertise in corporate law, corporate governance and mergers and acquisitions. His knowledge of corporate law including IP protection, employment contracts and legal partnerships will have great value to an emerging company like our own and his experience with larger corporations will be useful as DME grows.