IBM and The University of Alberta use machine learning to diagnose schizophrenia

IBM and the University of Alberta have used artificial intelligence to diagnose schizophrenia, with a 74% success rate. Data was utilised from MRI brain scans, acquired from patients diagnosed with schizophrenia or schizoaffective disorders, and a machine learning algorithm was applied to develop a model for diagnosis.

IBM have recently published an article on their website, identifying trailblazing research into the field of artificial intelligence and mental health. The article entitled – IBM and University of Alberta Publish New Data on Machine Learning Algorithms to Help Predict Schizophrenia – highlights the collaborative research they have been undertaking with The University of Alberta. The findings of the research have been published in Nature’s partner journal – Schizophrenia (IBM, 2017)

The research, which utilised retrospective analysis, successfully used machine learning and artificial intelligence algorithms to predict the likelihood of schizophrenic symptoms, with 74% accuracy. Not only were the algorithms capable of identifying schizophrenia, but they were capable of pinpointing the specific symptoms experienced by patients, with significant correlation. This groundbreaking neuroimaging research could be utilised, by scientists and physicians, to more accurately identify biomarkers associated with the development of schizophrenia (IBM, 2017).

The researchers used a data set of functional Magnetic Resonance Imaging scans (fMRI), from patients with schizophrenia or schizoaffective disorders. To identify the likelihood of schizophrenia, or similar disorders, the research analysed the blood flow to particular parts of the brain. The data set included the brain scans of 95 patients and the researchers employed machine learning techniques, to identify specific areas of the brain, connected to schizophrenia and schizoaffective disorders, which developed a model for diagnosis (IBM, 2017).

As a result of the research the machine learning algorithm could decipher those with schizophrenia, from the control group, with 74% accuracy. Dr Serdar Dursun, a Professor of Psychiatry & Neuroscience at the University of Alberta said:

“This unique, innovative multidisciplinary approach opens new insights and advances our understanding of the neurobiology of schizophrenia, which may help to improve the treatment and management of the disease,” (IBM, 2017).

This paves the way for further research into the field of machine learning, artificial intelligence and neuroimaging. Not only could this type of diagnostic tool be used to identify schizophrenia and schizoaffective disorders, but it could also be applied to other mental health areas. The article identified that bi-polar disorder, post traumatic stress disorders, and other psychiatric conditions, could be diagnosed with the utilisation of machine learning algorithms. As Mental health appears to feature more regularly within our societies, and governments are set to tackle these concerns, this type of innovation will undoubtedly become more sought after (IBM, 2017).