Summary: A revolutionary “digital twin” AI model has discovered that psychological and social factors are far more powerful predictors of Type 2 diabetes than previously realized. The study analyzed 17 years of data from nearly 20,000 UK adults.
Unlike standard medical tools that rely on blood sugar or BMI, this AI focuses on the “human” side of health. It found that loneliness, insomnia, and poor mental health each raise diabetes risk by an estimated 35 percentage points. When all three factors are present, the risk skyrockets by 78 percentage points, proving that mental well-being is just as critical as diet in preventing the disease.
Key Facts
The Power of Three: While diet is a known factor, the combination of loneliness, sleep disruption, and mental health struggles was a more accurate predictor of diabetes than food choices alone.The Stress Connection: Researchers believe these factors trigger a “slow-motion” health crisis by keeping stress hormones high, causing chronic inflammation and breaking the body’s ability to regulate insulin.Dietary “Stress Eating”: The model identified a strong link between high stress and “pro-inflammatory” diets—specifically high salt, sugary cereals, and processed meats.Ethnic Disparities: The AI confirmed significant disparities, showing that South Asian, African, and Caribbean participants face a markedly higher risk than White participants, highlighting the need for culturally targeted prevention.Cost-Effective Screening: Because this model uses lifestyle data instead of expensive blood tests or wearable devices, it could be used to identify high-risk individuals in underserved or low-income communities.
Source: Anglia Ruskin University
A new study using an advanced “digital twin” artificial intelligence model has found that factors such as loneliness, insomnia and poor mental health substantially raise a person’s future risk of developing type 2 diabetes.
The research, led by Anglia Ruskin University (ARU) in collaboration with Cranfield University, the University of Portsmouth, and Intelligent Omics Ltd, and published in Frontiers in Digital Health, used lifestyle and health data from 19,774 UK adults in the UK Biobank, tracked for up to 17 years.
Digital twin systems allow us to move past over-simplified models like BMI to see the complex emotional factors behind diabetes. Credit: Neuroscience News
Unlike traditional prediction tools, the new model focuses entirely on behavioural, lifestyle and psychosocial information rather than blood tests or wearable devices.
The digital twin model system, developed by ARU, simulated how changes in people’s day‑to‑day lives could alter long-term diabetes risk. It found that loneliness, insomnia and poor mental health were each associated with an estimated 35‑percentage‑point rise in risk, under AI‑modelled assumptions.
When all three of these factors occurred together, the model predicted a 78‑percentage‑point increase in absolute risk and is a more accurate predictor of type 2 diabetes risk than diet alone, the study found.
Researchers note these effects are likely linked to the body’s response to long-term stress, which raises stress hormones, triggers inflammation and disrupts how the body manages blood sugar.
The study also uncovered strong links between stress-related factors and dietary habits, including higher consumption of salt, sugary cereals and processed meats, which are all associated with increased risk of developing type 2 diabetes.
Even small dietary shifts reinforced risk levels, the model suggested. It also suggested cheese may have protective qualities, but this reduced in significance when mental health issues were present.
The digital twin model system also highlighted significant ethnic disparities, with South Asian, African and Caribbean participants showing markedly higher estimated risk than White participants, echoing long‑established NHS and Public Health England findings.
Because the model does not rely on medical tests, researchers say it could help health services identify high‑risk individuals earlier and design affordable, targeted prevention programmes.
Type 2 diabetes affects more than 500 million people and remains one of the world’s most pressing public health challenges, driven largely by preventable factors. It differs from type 1 diabetes, which is an autoimmune condition not linked to lifestyle.
Healthcare professionals have historically struggled to predict who will develop type 2 diabetes early enough to intervene effectively.
Co-author Professor Barbara Pierscionek, Deputy Dean for Research and Innovation in the Faculty of Health, Medicine and Social Care at Anglia Ruskin University (ARU), said: “Type 2 diabetes is a rising global health concern which we know is heavily influenced by lifestyle. However, current risk prediction models rely on BMI, age and blood pressure, which over-simplify this disease and overlook the more complex interconnected behavioural and emotional factors that precede and shape the onset of the condition.
“Digital twin model systems replicate an individual’s health profiles, enabling us to test ‘what-if’ scenarios and tailor care to individual needs. However, most of these existing models rely on real-time data from wearable devices, which can be a barrier for settings lacking in technical infrastructure or underserved communities that struggle with costs.
“Digital Twin model systems present a viable cost-effective way of diagnosis, testing and treatment for a number of conditions.”
Dr Mahreen Kiran, lead author and postgraduate researcher at ARU, said: “This study shows the importance of including behavioural and psychosocial variables such as loneliness, sleep disruption and mental health history within health datasets used for risk prediction.
“These factors are often overlooked, yet they provide meaningful signals about future disease risk. Incorporating them into digital twin models and other AI based approaches can support more accurate and equitable prevention strategies.”
Dr Nasreen Anjum, of the University of Portsmouth, said: “A key strength of this work is the use of transparent modelling and causal simulation techniques that help explain how behavioural factors interact over time. This improves confidence in how AI tools can support decision making in preventive healthcare.”
Key Questions Answered:Q: How can being “lonely” physically change my blood sugar?
A: Loneliness isn’t just a feeling; it’s a physiological stressor. When the body feels socially isolated, it enters a “threat state,” pumping out cortisol. Over years, high cortisol levels tell the liver to release extra glucose for energy and make your cells less responsive to insulin, eventually leading to Type 2 diabetes.
Q: What is a “Digital Twin” in healthcare?
A: It’s a virtual “clone” of your health profile. The AI takes your specific data—how much you sleep, your stress levels, your ethnicity, and your habits—and runs “what-if” simulations. It allows doctors to see your future health trajectory and test which lifestyle changes would lower your risk most effectively without needing to wait years for real-world results.
Q: The study mentions cheese might be protective. Should I eat more cheese?
A: The model found a protective link, but with a major catch: that protection largely disappeared in people with poor mental health. This suggests that the biological benefits of certain foods can be overridden by the physical damage caused by chronic stress and depression.
Editorial Notes:This article was edited by a Neuroscience News editor.Journal paper reviewed in full.Additional context added by our staff.About this AI and diabetes research news
Author: Jamie Forsyth
Source: Anglia Ruskin University
Contact: Jamie Forsyth – Anglia Ruskin University
Image: The image is credited to Neuroscience News
Original Research: Open access.
“A digital twin framework for predicting and simulating type 2 diabetes onset using retrospective lifestyle data” by Mahreen Kiran, Ying Xie, Graham Ball, Rudolph Schutte, Nasreen Anjum, and Barbara Pierscionek. Frontiers in Digital Health
DOI:10.3389/fdgth.2026.1710829
Abstract
A digital twin framework for predicting and simulating type 2 diabetes onset using retrospective lifestyle data
Introduction:
Type 2 Diabetes Mellitus (T2DM) is a rising global health concern, heavily influenced by modifiable lifestyle and psychosocial factors. However, most predictive tools focus on biomedical markers and rely on real-time data from wearables or electronic health records, limiting their scalability in resource-constrained settings.
This study presents a novel digital twin (DT) framework that uses retrospective lifestyle, behavioral, and psychosocial data to forecast T2DM onset and simulate the estimated effects of preventive interventions.
Methods:
Data were drawn from 19,774 participants in the UK Biobank cohort, followed for up to 17 years. A penalized Cox proportional hazards model was employed to estimate individual time-to-event risk trajectories based on 90 candidate predictors.
Predictors were selected through univariate screening, multicollinearity assessment, and variance filtering, yielding a final model with 14 significant variables. Causal inference techniques, including directed acyclic graphs (DAGs) and counterfactual simulations, were used to explore intervention effects on disease progression.
Results:
The model demonstrated strong predictive performance (C-index = 0.90, SD = 0.004). Psychosocial stressors such as loneliness, insomnia, and poor mental health emerged as strong independent predictors and were associated with estimated increases in absolute T2DM risk of approximately 35 percentage points individually and nearly 78 percentage points when combined, under the modeled assumptions.
These effects were partly reinforced through diet, with high intake of processed meat, salt, and sugary cereals acting as risk amplifiers within the modeled causal pathways. Cheese intake was protective overall, but its estimated benefit was attenuated under psychosocial stress, where reduced consumption produced a small, directionally harmful mediation effect.
Counterfactual simulations suggested that improvements in psychosocial conditions could reduce estimated T2DM risk by approximately 11.6 percentage points within the modeled cohort, with protective dietary patterns such as cheese consumption re-emerging as psychosocial stress was alleviated.
The model also revealed pronounced ethnic disparities, with South Asian, African, and Caribbean participants exhibiting significantly higher estimated risk than White counterparts within this cohort. These findings highlight the potential of integrated, stress-informed prevention strategies that address both psychosocial and dietary pathways.
Conclusion:
This study introduces a transparent, simulation-enabled DT framework for estimating T2DM risk and exploring behavioral intervention scenarios without reliance on real-time data streams. It enables interpretable, personalized prevention planning and supports exploration of scalable deployment in public health, particularly in underserved or low-infrastructure environments. The integration of psychosocial and lifestyle data represents an important step toward more equitable and behaviorally informed digital health solutions.