Current psychiatric diagnostic systems are based in long-standing observations and study in the field, careful consideration, and statistical verification over many decades. In the United States, we have the DSM-5, the most current edition of the Diagnostic and Statistical Manual of Mental Disorders in mental health. Abroad, the ICD-11 (International Classification of Diseases) is used for medicine more generally, with sections for psychiatry. With some exceptions, such as the noninclusion of complex posttraumatic stress disorder from the DSM-5 and the identification of burnout as a disease in the ICD-11, they largely overlap. To a significant extent, the biology is not well understood, and the current “polythetic”1 system is what is used within psychiatry.
Both are the result of categorical classifications based on more than 100 years of clinical work, and speak to clusters of symptom presentation that have been organized into discrete but highly overlapping criteria. By and large, the cause of the disorder and the biology of how psychiatric disorders work, how they wax and wane, resolve or progress—is not addressed.
A Medical Understanding of Psychiatric Conditions
As technology, information theory and computational tools, and neuroscience progress, we are finally approaching a time when our understanding of mind, brain, and body puts us within striking distance of having a true medical understanding of psychiatric conditions, their causes, and how those putative causes can drive better prevention and treatment.2 This is not to say that all the causes of mental illness are brain-based—far from it, as the rule is that most conditions are multifactorial, with an array of complex genetic, environmental, and dynamic factors that evolve and interact over time at play.
Nevertheless, understanding what is happening within the brain is a critical piece of the psychiatric diagnostic puzzle. Over the years, there have been different approaches to finding “biotypes” in psychiatry—for example, subtypes of depression,3 or using complex analysis to understand different mental typologies based on large data sets.4
Having a clear, empirically based model, for example, should allow for the development of critically needed biomarkers to test for different psychiatric conditions and to guide treatment planning and response. As it stands, there are no true tests for conditions like bipolar disorder, which often goes untreated due to delays in diagnosis, if it appears to be major depression (unipolar), a personality disorder, or attention-deficit/hyperactivity disorder. This is a direct consequence of having overlapping symptoms for a range of mental illnesses, making definitive diagnosis difficult and highly subjective, even with the most rigorous tools.
Recent Research on a Biologically Based Classification Model
Recently, Lett, Vaidya, and Jia, et al. (2025) presented their “Framework for Brain-Derived Dimensions of Psychopathology” in the journal JAMA Psychiatry. As they discuss, psychiatry needs a better understanding of neuroscientific biological mechanisms—brain structure, function, and connectivity, or network models. An empirically based organizing system, or “nosology,” to frame how we work with complex psychiatric problems, is overdue.
Their research combined current clinical measures and a range of brain-imaging strategies to derive a biologically based classification model with six cross-diagnostic psychopathology domains. They used data from population-based and cross-disorder studies (respectively, IMAGEN and STRATIFY/ESTRA) in which neuroimaging was done with participants when they were 14, 19, and 23 years old, and included psychological evaluation when they were 16. They included 1,003 participants in the study, and using SGCCA (sparse generalized canonical correlation analysis), examined the relationships among MRI data on structure, function, and connectivity with respect to clinical evaluation using standard diagnostic approaches.
Analysis derived six components of interest, which were statistically the most core canonical psychopathological elements connecting neuroimaging findings with clinical assessment in the Development and Well-Being Assessment (DAWBA) ratings sections and Alcohol Use Disorders Identification Test (AUDIT) Questionnaire. The brain imaging findings were linked to tasks that subject participants had completed while being scanned, including the stop-signal task, monetary incentive delay task, and emotional face task.
Six Psychopathological Dimensions
- Excitability and impulsivity
- Depressive mood and distress
- Emotional and behavioral dysregulation
- Stress pathology
- Eating pathology
- Social fear and avoidance
Correlations With Neuroimaging and Functional Measures
Each of the above psychopathological dimensions showed significant unique patterns when correlated with known diagnoses and associated neuroimaging patterns. For example, excitability and impulsivity were strongly related to bipolar disorder symptoms in DAWBA and were associated with brain imaging findings in key executive function areas in the frontoparietal network, including the dorsolateral prefrontal cortex, anterior cingulate, and inferior parietal cortex. Depressive mood and distress scores were correlated with a different pattern of findings on tasks and imaging studies, and so on for all the derived core domains.
Implications
This is not a full alternative diagnostic model, but it is an important step toward bridging existing clinical models with neurobiological data using a rich array of structural, functional, and connectivity-based brain imaging data. In principle, using this kind of analysis, it could be possible to use brain imaging in the clinical setting to scan a person’s brain while both at rest and performing specific tasks, and generate a report based on the six psychopathological profiles of what their unique brain activity indicates, and how that relates to conventional clinical diagnoses.
Looking deeper into the crystal ball, we could imagine relatively simple brain imaging combined with additional testing, either psychological/social evaluations or more biologically based lab testing for markers of inflammation or other systemic factors that may contribute to mental illness. As artificial intelligence becomes more sophisticated, the ability to work with large data sets will only improve—add to clinical evaluations the capacity to make sense of massive oceans of information, online habits and social media and communications, and real-world behaviors based on geotracking and wearables—as they evolve over time.
Regardless of what the future holds, the current research is significant for being a proof of concept that this kind of analysis can provide a rigorous biologically based framework, though limited by the techniques used, pool of participants, and types of clinical information analyzed. These six core psychopathological dimensions need verification and model refinement, but future iterations could be used to look back on current diagnostic models, ground them in neurobiology, and provide a framework that could potentially be used to update how we understand mental illness. Whether that will happen in a complex field lacking consensus is another matter.