Dr. Bo Cao Harnesses the Power of Machine Learning In Bid to Predict and Treat Schizophrenia at an Early Stage

Dr. Bo Cao's self-chosen nickname is Cloud, reflecting his love for an animated character bearing

1 September 2018

Dr. Bo Cao's self-chosen nickname is Cloud, reflecting his love for an animated character bearing that name in the popular 1990s videogame, Final Fantasy VII.

"When you're a teenager you struggle with self-identity, and I related to this character," he says. "He discovers that he has been cloned and most of his childhood memories are made up, so he has to fight to find out who he is. It's about his journey."

Dr. Cao, who recently joined the Department of Psychiatry as a researcher and Assistant Professor, has been on a real-life journey of his own, one he began as a kid growing up in northeast China.

Even as a child he had a keen interest in human perception, and how the brain interprets the sights, sounds, sensations and complex organizational constructs of the world around us.

After completing a Bachelor's degree in Mathematics in 2005 he switched to Psychology, completing a Master's degree at Beijing's Peking University, with a focus on the area of vision science.

"Half the brain is actually connected to vision. We're affected a lot by how we see the world, and our perception is all from the brain," he says.

"When I looked at doing a PhD I considered Neuroscience. But I also wanted to use my expertise in Math and that's what drew me to the interdisciplinary field of Computational Neuroscience," he says, which employs mathematical tools, theories and other concepts to investigate brain function.

Dr. Cao earned his PhD in Computational Neuroscience at Boston University in 2013. After completing a one-year Postdoctoral Research Fellowship, he moved to Houston, where he spent the next four years on a Postdoctoral Research Fellowship at the University of Texas Health Science Center.

His primary focus: applying machine learning and statistical analysis to the study of the brain, using the genetic, cognitive, behavioural and Magnetic Resonance Imaging (MRI) data of patients with psychiatric disorders.

"Although theoretical work is very appealing, I wanted to do something more practical. I wanted to actually apply my expertise to some real-world problems. That's how I got into psychiatry," he explains.

"I still appreciate a lot of the progress that's happening in Neuroscience and I'm still reading a lot in that area. But psychiatry, mental disorders and neurological disorders are another way to investigate the brain. So that's one motivation. The other motivation is the practical part. There's a chance that you can actually directly - maybe more directly - help people."

In essence, that's what prompted Dr. Cao to lead a study - in collaboration with Dr. Xiang Yang Zhang, an Associate Professor in Psychiatry at the University of Texas Health Science Center in Houston - harnessing the power of machine learning algorithms as a tool for identifying patients with Schizophrenia.

"Repeated untreated psychotic episodes may be associated with irreversible alterations of the brain. Thus, it is crucial to identify Schizophrenia early and provide effective treatment," Dr. Cao explains, in a recent interview with MedicalResearch.com, an online publication.

"However, identifying biomarkers in Schizophrenia during the first (psychotic) episode without the confounding effects of medications has been challenging. Substantial but still limited progress has been made in leveraging these biomarkers to establish diagnosis and make individualized predictions of future treatment responses to antipsychotics."

By using machine learning algorithms to measure the connections between the brain's superior temporal cortex and other cortical regions, Dr. Cao and his team were able to identify first-episode drug-naïve (i.e., previously untreated) Schizophrenia patients with an accuracy of 78.6 per cent. They were also able to predict which individual patient could respond to a particular antipsychotic drug, Risperidone, with 82.5 per cent accuracy.

"The temporal cortex is related to auditory processing, and the altered functional connectivity in the superior temporal cortex may be related to hallucinations. So it's not surprising that the connections from this area to the rest of the cortex are critical to our understanding of mental illness," he explains.

The study, which involved 43 subjects, was partly funded by the NARSAD Young Investigator Grant of the New York-based Brain & Behaviour Research Foundation. Previously known as the National Alliance for Research on Schizophrenia & Depression, or NARSAD, it is a nonprofit organization that supports mental health research.

Dr. Zhang is one of Dr. Cao's NARSAD mentors.

Dr. Cao is hoping that his study methods and findings could be used as a step by other researchers toward achieving more accurate early-stage identification and improved treatment response predictability for previously untreated individuals with Schizophrenia.

"One component of our study was prognostic, but the other component that to me is equally if not more interesting, is the treatment prediction. Remember, we looked at one monotherapy. But in the real world you have multiple treatment choices, so you have to determine which one is the best in advance," he says.

"So this is just the beginning. The more helpful tool would be knowing which drug you should use for a new patient, instead of saying 'Okay, I only have this one drug, and let's see if this will tell me whether it will work or not.' So there is still a big gap there, and that's actually something for us to study in future."

Although he only arrived in Edmonton a few months ago, and has yet to explore other parts of Western Canada, Dr. Cao says he is enjoying his new role in the University of Alberta's Department of Psychiatry.

"If you look at how Psychiatry Departments are structured across Canada and the U.S., our Department is quite unique. In addition to encouraging collaborations between psychiatrists and basic scientists within the Department, we are also actively establishing collaborations with artificial intelligence and machine learning experts from across the university," he says.

"We are building a team of Computational Psychiatry and Digital Psychiatry experts, with a focus on data-driven methods and future data streaming in mental health. As a result we are creating a one-of-a-kind PhD program in Psychiatry," he notes.

"It can be challenging for PhDs to find a home within a medical department. But our Department, the Faculty of Medicine and Dentistry and the University of Alberta all want to expand into these new fields, including machine learning and artificial intelligence in mental health, so there is a lot of opportunity here. We are seeking talented colleagues to join our team as PhD students, Postdocs and Research Associates."