Schizophrenia and bipolar disorder share many characteristics. Both are severe mental illnesses that have a major impact on quality of life. Both are associated with autonomic nervous system dysfunction and can be assessed through cardiac activity analysis. (In fact, there is a continuing debate as to whether two diseases that are classified as clear based on the patient's symptoms actually represent a continuum.) Both may challenge a diagnosis. And early detection of both diseases is extremely important for better outcomes.
Diagnosis at the progenitor stage – leading to a gradual decrease in emotional, behavioral and cognitive function after the appearance of early warning signs and symptoms, but before the full onset of schizophrenia – is rare. Later diagnosis often results in hospitalizations in psychiatric units, increased patient costs, and increased burden on the health care system.
Psychiatry is one of the few areas of medicine that does not have the advantages of laboratory blood tests, including genetic profiling to measure biomarkers for diagnosis and disease monitoring. Most serious mental disorders are diagnosed based on clinical interviews with trained physicians who interpret subjective and different sets of symptoms reported by patients.
The diagnostic approaches explained and tested in recent research will break that type. This study was published on September 3, 2025 PLOS Computational Biologyused heart rate data may not meet the strict definition of biomarkers, but is a significant departure of mental state. Based on the results of two groups of 30 patients each, one member diagnosed by a psychiatrist with schizophrenia or bipolar disorder (data were analyzed collectively) and another control – classified each disease with an accuracy of approximately 80%.
“While it is not an alternative to clinical diagnosis, cardiac monitoring combined with artificial intelligence can detect signs of mental illness and lead to faster interventions,” the authors wrote, adding that their research could ultimately lead to the development of easy-to-use tools that can support physicians and improve patient care. ”
Corresponding authors Kamil Ksiąjek, Ph.D. , a postdoctoral researcher in the Machine Learning Research Group at Jagiellonian University in Kraków, Poland, and affiliations and colleagues from around Eastern Europe have published highly technical papers.
We explain the development and evaluation of an automated classification method that uses deep learning techniques to analyze RR intervals, i.e., the period of time between the peaks of two consecutive “R” waves in the electrical measurement diagram (ECG).
Low cost of devices, researchers write about the high reliability and the relatively short period of time they need to wear to capture the required data, and it will become useful beyond the clinical setting. In homes, rural areas and low-income countries, there are not enough psychiatrists to diagnose mental disorders based on physician interviews.
It can also be used for screening for mental disorders on a population basis. The treatment and progression of the disease can be monitored via telehealth. Unlike many advanced medical diagnostic techniques, data requires contextual interpretation through software or trained physicians, but highly specialized personnel are not required in the field.
This study compared multiple machine learning models used to analyze heartbeat data with the results from which each result was published. Diagnostic accuracy is based on data differences between the schizophrenia/bipolar group and control.
The author has given one long term potentially important caution. Several studies have shown that antipsychotics have a significant impact on the variability of heart rates that research relied on, with different drugs having different effects. However, the researchers were unable to ethically stop the subject's medication.
To assess the possible effects of antipsychotic drug use on heartbeat intervals used in studies, researchers are known to have the greatest effect of those values in a subset of patients receiving quetiapine (seroquel) on other schizophrenia/bipolar groups. No significant differences were found, suggesting a small effect on antipsychotic medication measurements.
“Our research shows that wearable devices may be a cost-effective diagnostic tool for both inpatients and outpatients,” the authors conclude.
