Computational analysis pipeline accurately identifies schizophrenia and bipolar disorder in tissue cultures

Diagnosing neuropsychiatric disorders like schizophrenia (SCZ) and bipolar disorder (BD) can be difficult, as diagnostic criteria can differ even between doctors. In previous research, author Annie Kathuria tested lab-grown tissue cultures — known as organoids — for electrical signals related to psychiatric diseases, and in new research, Cheng et al. used that data to create a computational model to identify SCZ and BD.

“The beauty of an induced pluripotent stem cell is that it carries a person’s genetics,” Kathuria said. “So, if there are cells taken from a schizophrenic patient, we would expect the organoid to show [the same biological] characteristics.”

Because there are no clear genetic markers for diseases, electrical signals from multi-electrode arrays (MEA), which are like electroencephalograms for organoids, can be processed and used for identifying SCZ and BD.

“What our MEA machines can do is record the electrical activity of an organoid placed on top of them, and we can also stimulate the organoid using electrical pulses,” Kathuria said.

Their computational analysis pipeline was created with a support vector machine classifier — a machine learning algorithm — optimized for high-dimensional data. The pipeline was able to differentiate BD and SCZ from control organoids with 83.3% accuracy, and that number improved to 91.6% in the signals from the organoids that experienced electrical stimulation.

Beyond increasing the organoid number by expanding the donor cohort, Kathuria imagines their research will be used for drug development. By creating organoids from a patient’s stem cells, researchers will be able to test a drug’s efficacy on an organoid rather than a person.

Source: “Machine learning–enabled detection of electrophysiological signatures in iPSC-derived models of schizophrenia and bipolar disorder,” by Kai Cheng, Autumn Williams, Anannya Kshirsagar, Sai Kulkarni, Rakesh Karmacharya, Deok-Ho Kim, Sridevi V. Sarma, and Annie Kathuria, APL Bioengineering (2025). The article can be accessed at https://doi.org/10.1063/5.0250559 .

This paper is part of the Bioengineering of the Brain Collection, learn more here .