A web-based clinical risk prediction tool could help identify young adults with psychosis who are at high risk of developing cardiometabolic disorders years before they occur.

In validation analyses, the algorithms behind the Psychosis Metabolic Risk Calculator 2.0 (PsyMetRiC2) accurately predicted clinically significant weight gain within 1 year, metabolic syndrome within 6 years, and type 2 diabetes within 10 years among young patients with psychotic disorders, potentially enabling earlier preventive care.

“The tool is designed for use by health professionals, this may include medical or nursing staff and allied health professionals. It is designed to be quick and easy to use, requiring only routinely collected information in order to make predictions,” Benjamin Perry, PhD, associate clinical professor of psychiatry, University of Birmingham, Birmingham, England, who led development of the tool, told Medscape Medical News.

The study describing validation of the PsyMetRiC tool was published online on March 11 in The Lancet Psychiatry.

Focus on Young Adults

Perry noted that about two thirds of psychosis spectrum disorders first occur between the ages of 16 and 35 years.

“We focused on this younger early psychosis population given the firm evidence base that this group already show evidence of emerging cardiometabolic disorders in the earliest stages of illness, many years earlier than would typically be expected in the rest of the population,” Perry said.

“This is one reason why existing general population-based tools, which are developed to be accurate for middle to older age adults, are inaccurate and under-predict risk in young people with psychosis spectrum disorders,” he added.

First developed and validated in the UK in 2021, the PsyMetRiC psychosis prediction tool has since undergone further refinement and expansion. These advancements led to the development of enhanced algorithms, and the creation of a web application, resulting in PsyMetRiC2.

Perry’s group tested and validated PsyMetRiC2 using routine health record data from more than 25,000 young adults in the UK, aged 16-35 years, with psychosis, who were followed for over 20 years.

PsyMetRiC2 showed strong predictive performance. For metabolic syndrome, the algorithm achieved a C-statistic of 0.81 in external validation, indicating good discrimination between those who would, and would not, develop the condition.

For type 2 diabetes, predictive performance was similarly strong, with external validation showing a C-statistic of 0.81. The weight-gain model showed promising internal validation, with a C-statistic of 0.78. However, it has not yet been externally validated.

The PsyMetRiC web application has been certified by the UK’s Medicines & Healthcare products Regulatory Agency as a class 1 medical device.

Last year, Canadian researchers published the first validation of the tool in North American patients. Perry and his colleagues recently received funding to test it in a US population.

Perry noted that the PsyMetRiC tool is available for academic use anywhere in the world. Licenses are available via an express licensing portal. The website includes a downloadable risk communication guide for health professionals.

“We are hoping to increase vigilance about the physical health risks for young people with psychosis and expand the conversation between doctors and patients, so these risks are mitigated and premature mortality is reduced by earlier intervention and preventative measures,” Perry said in a news release.

Like most software, it will receive iterative updates over time, with improvements and updates shaped by new research, stakeholder input, and the results of upcoming qualitative, health economic, and impact evaluation studies.

Potential Game Changer?

Commenting on the research for Medscape Medical News, Murali Doraiswamy, MBBS, professor of psychiatry and medicine, Duke University School of Medicine, Durham, North Carolina, who wasn’t involved in the study, said cardiometabolic disorders in patients with psychosis are a “huge public health issue and if they can be prevented that would have a significant impact on both the person and society.”

PsyMetRiC is “exactly the type of tool we need to be proactive and predictive rather than reactive” and it “could be a game changer for sure but only if it’s very accurate,” said Doraiswamy.

“If it has a lot of false positives or negatives then clinicians will lose faith in it and it could cause harm unintentionally by depriving people of antipsychotic treatments,” Doraiswamy cautioned.

Another major issue with predictive tools is that they often don’t generalize well across different settings. It is therefore unclear how effectively this model will perform in the US, where diets and adherence to medical care may differ. Unlike the National Health Service, lower-income patients in the US do not always have access to a comparable safety net, Doraiswamy said.

He also noted that AI tools for predicting metabolic syndrome are already integrated into EPIC electronic health record systems, and it remains unclear how PsyMetRiC compares with these existing tools. He said that evaluating these differences should be a high research priority.

Funding for this research was provided by the National Institute for Health and Care Research (NICE). Perry reported being a NICE topic advisor for Severe and Enduring Mental Illness. Doraiswamy reported consulting for pharmaceutical and technology companies on other areas not related to psychosis.