NextFin news, on November 25, 2025, a collaborative research team led by Dr. Elena Ghotbi and senior author Dr. Shadpour Demehri at Johns Hopkins University School of Medicine unveiled a pioneering AI-driven discovery: the identification of the first biological marker of chronic stress through routine medical imaging. Using a deep learning artificial intelligence model applied to thousands of existing chest computed tomography (CT) scans, the researchers quantified adrenal gland volume — a physiologically relevant proxy for long-term stress exposure. Their findings were presented at the Radiological Society of North America (RSNA) annual meeting in Chicago.
The study analyzed data from 2,842 participants enrolled in the Multi-Ethnic Study of Atherosclerosis (MESA), whose chest CT scans were complemented by validation measures including salivary cortisol rhythms, stress-related questionnaires, body mass index (BMI), blood pressure, heart rate, glucose levels, and allostatic load indices — the cumulative physiological burden imposed by chronic stress. By segmenting and calculating adrenal gland volume into an Adrenal Volume Index (AVI), the AI model demonstrated statistically significant correlations with validated biochemical and psychosocial stress indicators. Notably, participants exhibiting higher AVI had elevated cortisol levels, greater allostatic load, and reported higher perceived stress.
Furthermore, the study linked higher AVI values to structural changes in cardiovascular health: specifically, an increased left ventricular mass index and a markedly elevated risk of heart failure and mortality over a follow-up period of up to ten years. Each incremental 1 cm³/m² increase in adrenal volume was associated with clinically meaningful cardiac outcomes, establishing adrenal volume as a predictive biomarker of chronic stress’s physiological sequelae beyond subjective patient reports or inconvenient serum testing.
Until now, assessment of chronic stress effects relied on imperfect tools such as questionnaires, transient cortisol measurements, or surrogate inflammatory markers, which are limited by variability and practicality. This novel AI-enabled technique leverages widely available and routinely performed chest CT data, offering an objective, scalable, and non-invasive metric embedded seamlessly into existing clinical workflows without additional radiation or testing burdens—a critical innovation amid rising mental health and stress-related disease burdens globally.
The implications are multi-fold: medically, this biomarker could enhance cardiovascular risk stratification by integrating stress as a measurable factor, enabling clinicians to identify high-risk individuals earlier and tailor preventive or therapeutic interventions accordingly. Economically, the integration of AI models that utilize existing imaging data exemplifies cost-effective innovation by harnessing underutilized information embedded in medical scans to reduce downstream costs of complications associated with chronic stress, such as heart failure and depression.
This development also signals a turning point in mental health diagnostics, combining radiological expertise with data science to operationalize the elusive concept of chronic stress’s cumulative biological load. By quantifying adrenal volume changes that physiologically manifest from prolonged stress activation of the hypothalamic-pituitary-adrenal (HPA) axis, the model serves as a biological barometer — a notion reinforced by epidemiological expertise from co-author Professor Teresa Seeman of UCLA. Her commentary highlights this as a significant advancement in translating decades of psychosocial stress research into a tangible clinical tool that predicts hard outcomes.
Looking forward, the adoption of AI-based stress biomarkers in routine care awaits peer-reviewed publication and broader validation across diverse populations and imaging modalities. However, the technology’s scalability is promising given millions of chest CT scans performed annually in the U.S. alone. Widespread use could catalyze cross-disciplinary research intersecting radiology, endocrinology, psychiatry, and cardiology to better understand chronic stress’s multifactorial role in disease. It can also inform public health strategies aiming to mitigate societal stressors amplifying chronic disease risk.
President Donald Trump’s administration, inaugurated earlier this year, has emphasized advancing health technologies and mental health initiatives. Such innovations align with federal priorities to modernize healthcare infrastructure and improve outcomes through AI-driven diagnostics. The translation of this biomarker into clinical practice will depend on regulatory assessments, insurance reimbursement policies, and integration into electronic health record systems for real-time decision support.
This breakthrough underscores AI’s transformative potential in medicine, illuminating hidden pathophysiological markers previously undetectable by human analysis alone. It exemplifies the convergence of advanced imaging analytics, deep learning algorithms, and clinical epidemiology to tackle complex health challenges shaped by psychosocial factors often overlooked by conventional diagnostics.
According to Euronews, this technological advancement promises earlier, more accurate intervention opportunities for chronic stress-related conditions, potentially reducing the high global economic burden of cardiovascular diseases and mental health disorders, and shifting healthcare paradigms towards precision preventive medicine that holistically accounts for lifestyle and stress physiology.
As research and clinical translation progress, continuous monitoring of AI model performance, ethical considerations regarding data privacy, and equitable access across healthcare systems remain paramount to maximize benefits and mitigate risks, ensuring this discovery drives sustainable improvements in public health and patient care.
Explore more exclusive insights at nextfin.ai.