Screening for major depressive disorder in a tertiary mental health centre using EarlyDetect: A machine learning-based pilot study

Mental illness is the leading cause of disability in the world. Diagnosing mental health syndromes have proved challenging for clinicians. This has led to delay in diagnosis, misdiagnosis or unfortunately in some a lack of diagnosis. Furthermore the majority of individuals who have mental disorders have more than one condition (co-morbid), which are essential to screen and diagnosis early to achieve symptom remission and achieve full functional recovery.
In the private sector, mental health claims have eclipsed physical injury leading to economic loss in the billions globally, and yet there is no widely available psychiatric screening tool for risk assessment . Currently in the Canadian climate of mental health the only options offered originate from an organizational management approach, or peer-to-peer and self-monitoring strategies. In our discussions with the private sector industry, the need and demand for a medical strategy to address issues of mental health would outstrip any conventional ability to meet the need.

Recognizing the need for an effective and convenient assessment tool to foster early detection of mental health disorders, improve functional outcomes and prevent mental illness, we developed EarlyDetect (ED). The ED program asks users to answer a series of clinical assessment questionnaires for depression, anxiety disorder, bipolar disorder, ADHD, and alcohol use disorder, using our proprietary Life History Questionnaire, and a functional assessment in the context of work/school, family life, and social life. Other assessment tools are under development, including PTSD screening, polysubstance abuse screening, and functional assessment focusing on improving work productivity and reducing absenteeism. This tool has flexibility to adapt to the specific needs of clients to screen for other mental health symptoms or disorders and measure functional and quality of life outcomes.

Our current research comparing ED to single-blind diagnoses made by a certified clinician shows superior performance from ED in diagnosing mental illnesses when compared to stand-alone assessment tools typically used to screen for these disorders. Through a collaboration with the Computational Psychiatry Group at the University of Alberta (https://comppsych.weebly.com/), we have meticulously developed a scientifically and statistically sound correspondence calculation that generates a hybrid (mathematically and clinically derived) risk assessment score, which our testing has shown to be significantly more accurate than existing stand-alone self-report assessments. Our machine-learning study for screening depression has recently been published in the Journal of Affective Disorders Reports (Liu et al., 2021a). More recently, we have published a user experience study (Liu et al., 2021b). Patients are able to complete in 10-12 minutes and are able to easily navigate through various sections of the app. Currently, we just received acceptance of our paper that further supports the effectiveness of the tool in screening for Bipolar Disoder (Liu et al, 2021, JADReports).

We are making continued research and development efforts to further develop the contents of the app, improve prediction algorithms, and user experience. Since the development of the iOS version, we developed a cloud-based version of ED on Microsoft Azure cloud, with cross-platform assessment and data management capability, and are working towards further validation of the impact of ED on patient's outcomes in primary health care clinics and in the private sector.

Our service targets the health care system assisting clinicians in facilitating diagnosis and co-morbid diagnosis and easing referral time, tracking documentation, and enhancing diagnostic accuracy and measuring functional and quality of life outcomes. We also target the private sector, where companies of all sizes and configurations can utilize mental health screening to facilitate employee wellness and risk management. Our customers will benefit from significantly reduced costs of absenteeism and presenteeism, an enhanced corporate culture of safety and wellness, and a positive reputation among customers and competitors for valuing and proactively addressing employee mental health.