Research finds neural firing patterns as biomarkers for schizophrenia and bipolar disorder — Evidence Review

Scientists at Johns Hopkins University have used lab-grown brain organoids to identify distinct neural activity patterns linked to schizophrenia and bipolar disorder, potentially offering new tools for diagnosis and treatment. These findings align with previous research showing both shared and unique neural signatures across psychiatric conditions.

• Multiple studies support the presence of overlapping but distinct neural circuit and molecular abnormalities in schizophrenia and bipolar disorder, with shared disruptions in connectivity and gray matter, yet more severe changes in schizophrenia 1 4 10.
• The use of patient-derived organoids to model disease-specific neural activity builds on prior work demonstrating the promise of organoids in uncovering cellular and developmental differences in psychiatric disorders 5, and in other fields such as cancer and personalized medicine 6.
• The integration of machine learning to decode disease-specific electrophysiological signatures provides a novel approach that complements existing neuroimaging and genetic research, potentially enabling more personalized and objective diagnostic and treatment strategies 2 8 9.

Study Overview and Key Findings

Understanding the biological mechanisms underlying psychiatric disorders such as schizophrenia and bipolar disorder remains a considerable challenge, partly due to the lack of clear molecular or structural markers. Traditionally, diagnosis and treatment rely on clinical assessment, which can be subjective and often involves lengthy trial-and-error with medications. This new study is significant because it leverages advances in stem cell technology and machine learning to probe disease-specific brain activity in vitro, using organoids derived from patient cells as a model system.

Property	Value
Study Year	2025
Organization	Johns Hopkins University
Journal Name	APL Bioengineering
Authors	Kai Cheng, Autumn Williams, Anannya Kshirsagar, Sai Kulkarni, Rakesh Karmacharya, Deok-Ho Kim, Sridevi V. Sarma, Annie Kathuria
Population	Patients with schizophrenia, bipolar disorder, and healthy volunteers
Sample Size	n=12
Methods	In Vitro Study
Outcome	Neural firing patterns, biomarkers for schizophrenia and bipolar disorder
Results	Accuracy of identifying organoid origin increased from 83% to 92%.

This study recruited blood and skin samples from individuals with schizophrenia, bipolar disorder, and healthy controls, reprogramming these into stem cells to grow brain organoids—miniature, simplified versions of brain tissue. By measuring and analyzing the electrical activity of these organoids with multi-electrode arrays and machine learning, the researchers identified disorder-specific neural firing patterns. These patterns served as biomarkers that enabled accurate classification of the organoid's origin. The ability to distinguish between schizophrenia, bipolar disorder, and healthy-derived organoids reached 92% accuracy after stimulation, suggesting a potential pathway for more objective diagnosis and drug testing.

Literature Review: Related Studies

To contextualize these findings, we searched the Consensus database—housing over 200 million research papers—using the following queries:

1. neural mechanisms schizophrenia bipolar disorder
2. organoid origin identification accuracy
3. mental health disorders neurobiological differences

Below, we group key findings from the literature into major topics:

Topic	Key Findings
How do neural circuit and molecular differences distinguish schizophrenia and bipolar disorder?	- Both disorders share some neurobiological features, such as white matter deficits and changes in cortical gray matter, but schizophrenia typically involves more pronounced and widespread changes, particularly in the insula and thalamus 1 4 10. - Decreased parvalbumin-expressing neurons in the thalamic reticular nucleus and disruptions in excitation-inhibition balance are implicated in both conditions, potentially affecting cognitive and emotional processing 3 5.
What is the potential of organoids and machine learning for personalized medicine in neuropsychiatric disorders?	- Organoids derived from patient tissue retain key molecular and genetic characteristics, and have shown utility in cancer for predicting drug responses and validating biomarkers 6. - Emerging machine learning approaches can accurately analyze organoid behavior, enabling high-throughput, objective identification and tracking of disease-specific patterns, which could accelerate personalized diagnosis and treatment 8 9.
How do connectivity and network function differ across psychiatric disorders?	- Both schizophrenia and bipolar disorder show disruptions in large-scale brain network connectivity, particularly in networks involved in cognitive and emotional regulation, with schizophrenia often exhibiting more severe or widespread deficits 2 4 10 14. - Functional imaging and meta-analyses reveal both shared and distinct network disruptions, supporting the idea of both common and disorder-specific neural substrates 1 2 10 14.
Are there shared molecular and genetic underpinnings across psychiatric conditions?	- Transcriptomic and genetic analyses reveal overlapping gene expression patterns and polygenic risk factors across disorders, suggesting common molecular pathways underlying multiple psychiatric illnesses 11. - These shared molecular features may explain similarities in clinical presentation and challenge strict diagnostic boundaries, motivating the search for more biologically based classification systems 11 12.

How do neural circuit and molecular differences distinguish schizophrenia and bipolar disorder?

The current study's findings of distinct electrophysiological signatures for schizophrenia and bipolar disorder in brain organoids are consistent with previous literature showing both shared and unique neural changes in these conditions. Research using neuroimaging and postmortem analyses has identified widespread gray matter reductions and white matter deficits in schizophrenia, with somewhat milder or more localized changes in bipolar disorder 1 4. Additionally, disruptions in specific neuronal subtypes and excitation-inhibition imbalances have been implicated in both conditions 3 5.

• Schizophrenia is often characterized by more severe and extensive structural brain changes than bipolar disorder, especially in regions such as the insula and thalamus 1 4.
• Both conditions display decreased parvalbumin-expressing interneurons in the thalamic reticular nucleus, which may contribute to cognitive and emotional symptoms 3.
• Imbalances in excitatory and inhibitory neuron development, observed in organoid models, support the hypothesis that altered cortical circuitry underlies psychosis 5.
• These findings underscore the value of organoid models in recapitulating disease-relevant neurodevelopmental processes and neural activity patterns 5.

What is the potential of organoids and machine learning for personalized medicine in neuropsychiatric disorders?

The use of patient-derived organoids combined with machine learning analysis offers a promising avenue for developing personalized diagnostic and therapeutic approaches in psychiatry. While such methods are well-established in oncology for stratifying treatments and validating biomarkers 6, their application in neuropsychiatry is more recent. Advances in deep learning have enabled accurate detection and analysis of organoid behavior, facilitating high-throughput and objective measurements 8 9.

• Organoids derived from patients retain relevant disease features, allowing for in vitro modeling of individual response to drugs and interventions 6.
• Machine learning algorithms can analyze complex electrophysiological or morphological data, improving the accuracy and scalability of disease classification 8 9.
• These technologies may reduce reliance on trial-and-error in psychiatric medication management by enabling pre-clinical testing of drug responses in patient-specific models 6.
• The current study's high classification accuracy (up to 92%) demonstrates the feasibility of this approach for neuropsychiatric disorders.

How do connectivity and network function differ across psychiatric disorders?

Both schizophrenia and bipolar disorder exhibit disruptions in large-scale brain network connectivity, affecting cognitive and emotional processing. Studies using resting-state and task-based functional imaging have shown that these disorders impact similar networks, though schizophrenia tends to have more pronounced or extensive disruptions 2 4 10 14. The present study's identification of disorder-specific neural firing patterns in organoids suggests that electrophysiological signatures observed in vitro may mirror in vivo network dysfunctions.

• Disconnections between cognitive and emotional brain networks are common to both disorders, with schizophrenia often showing greater connectivity deficits 2 4.
• Functional connectome studies reveal shared patterns of network disruption across different psychiatric diagnoses, with certain features scaling with illness severity 10 14.
• While there is overlap in affected networks, some connectivity alterations—such as those in the default mode network—may be more specific to psychotic illness 14.
• These findings support the development of functional and electrophysiological biomarkers for more precise diagnosis and monitoring.

Are there shared molecular and genetic underpinnings across psychiatric conditions?

Genomic and transcriptomic meta-analyses indicate that psychiatric disorders share substantial molecular and genetic overlap, challenging the boundaries between traditional diagnostic categories. Studies have found that many risk genes and gene expression changes are not unique to a single disorder but are instead common across several conditions 11. These insights motivate the movement toward biologically based nosologies, such as the Research Domain Criteria (RDoC) framework 12.

• Shared gene expression modules and polygenic risk factors underlie multiple major psychiatric disorders, including schizophrenia and bipolar disorder 11.
• This genetic and molecular overlap may explain similarities in symptoms and complicate differential diagnosis 11.
• The search for disorder-specific biomarkers remains important for clinical differentiation and targeted intervention.
• The current study's identification of unique electrophysiological signatures in organoids complements genetic and transcriptomic data, advancing the search for actionable biomarkers.

Future Research Questions

While the new study provides important insights, further research is needed to validate these findings in larger, more diverse populations and to translate them into clinical practice. Key questions remain about the reproducibility of organoid-based biomarkers, their correlation with in vivo disease states, and their utility in guiding treatment decisions.

Research Question	Relevance
How do organoid-based electrophysiological biomarkers correlate with in vivo neural activity in psychiatric patients?	Establishing a direct relationship between organoid-derived and in vivo brain activity would validate the translational relevance of these models for diagnosis and treatment 5.
Can organoid models predict individual responses to psychiatric medications?	If organoids can reliably forecast drug efficacy or resistance, this could reduce trial-and-error prescribing and accelerate personalized treatment strategies 6 8 9.
What are the long-term effects of abnormal neural firing patterns identified in organoids on brain development and function?	Understanding the developmental trajectory and functional consequences of these patterns could clarify disease mechanisms and inform early intervention strategies 3 5.
How generalizable are organoid-based biomarkers across diverse populations and psychiatric diagnoses?	Larger, more heterogeneous samples are needed to determine if these biomarkers are robust across different genetic backgrounds, ages, and environmental exposures 1 4 10 11.
Can combining organoid data with genetic and neuroimaging findings improve psychiatric diagnosis and treatment?	Integrating multiple data types may enhance diagnostic accuracy and provide a more comprehensive understanding of disease mechanisms 1 10 11.