OBSCORE AI Tool May Detect Hidden Obesity Risks Beyond BMI

OBSCORE, a new AI-based obesity risk tool, may predict future diabetes, heart disease, and other complications more accurately than BMI alone.

A new artificial intelligence-based tool called OBSCORE may help doctors identify which overweight or obese individuals are most likely to develop serious obesity-related diseases years before symptoms appear.
Developed using health data from nearly 200,000 adults in the UK Biobank, the tool predicts future risks of conditions such as type 2 diabetes, cardiovascular disease, chronic kidney disease, sleep apnea, and hypertension more accurately than body mass index (BMI) alone.

The findings, published in Nature Medicine, suggest that obesity risk is far more complex than body weight itself and that many high-risk individuals may currently be overlooked using standard BMI-based approaches (1^✔ ✔Trusted Source
Data-driven prioritization of high-risk individuals for weight loss interventions

Go to source).

How Does OBSCORE Predict Obesity-Related Disease Risk Beyond BMI?

The OBSCORE tool was developed by researchers using data from approximately 197,000 adults aged 40–69 years enrolled in the UK Biobank.

All participants had a BMI of at least 27 kg/m², meaning they were classified as overweight or obese.

The researchers tracked participants for nearly 10 years and analyzed the development of 18 obesity-related complications, including:

• Type 2 diabetes
• Cardiovascular disease
• Chronic kidney disease
• Sleep apnea
• Hypertension
• Gout
• Arthropathy
• Cardiovascular mortality

To build the tool, scientists evaluated more than 2,300 potential predictors ranging from demographics and blood biomarkers to medications, clinical history, lifestyle information, metabolomics, and genetic data.

Using a machine learning framework that combined LASSO feature selection and regularized Cox models, the team identified the 20 most informative clinical features that consistently predicted future disease risk. These features were then integrated into a single risk prediction model called OBSCORE.

The study found that general health information, existing medical conditions, medication history, and standard blood biomarkers produced the strongest predictive performance. By contrast, advanced genetic scores and metabolomics added relatively little extra predictive value for most complications.

OBSCORE performed especially well in predicting:

• Type 2 diabetes
• Chronic kidney disease
• Gout
• Sleep apnea

Researchers also externally validated the tool in independent European and South Asian population cohorts, suggesting the model may be useful across diverse populations.

Why Are Scientists Increasingly Questioning BMI Alone?

For decades, BMI has been the standard method used to classify overweight and obesity. However, scientists increasingly argue that BMI cannot fully reflect an individual’s metabolic health or actual disease risk.

A review published in Diagnostics explained that two people with identical BMI values may have completely different health profiles depending on factors such as visceral fat accumulation, inflammation, insulin resistance, and fat distribution (2^✔ ✔Trusted Source
Beyond BMI: Rethinking Obesity Metrics and Cardiovascular Risk in the Era of Precision Medicine

Go to source).

Researchers now recognize several obesity phenotypes, including:

• Metabolically healthy obesity, where a person has a high BMI but relatively healthy metabolic markers
• Metabolically unhealthy normal weight, where someone has a “normal” BMI but still faces high cardiometabolic risk

The review noted that BMI cannot distinguish between muscle and fat mass or determine where fat is stored in the body. Visceral fat surrounding internal organs appears to be far more harmful than overall body weight alone (2^✔ ✔Trusted Source
Beyond BMI: Rethinking Obesity Metrics and Cardiovascular Risk in the Era of Precision Medicine

Go to source

The study reinforced this limitation by showing that many individuals identified by OBSCORE as high-risk were classified only as overweight rather than obese. Researchers say this suggests BMI-based screening may fail to identify many vulnerable individuals early enough.

What Were the Most Important Findings from the OBSCORE Study?

One of the most important findings was that obesity-related disease risk varied substantially even among people with similar BMI levels.

OBSCORE distinguished high- and low-risk individuals within the same BMI category more effectively than BMI alone. Researchers say this could help doctors identify which patients are most likely to benefit from early intervention.

The study also found that:

• Blood biomarkers and routine clinical data were more useful predictors than genetics alone
• Existing diagnoses and medication history significantly improved risk prediction
• Polygenic risk scores contributed relatively little additional benefit
• Many individuals at the highest risk were overweight rather than severely obese

Researchers also tested OBSCORE using data from the SURMOUNT-1 trial involving tirzepatide, a GLP-1/GIP-based obesity drug. The study found that predicted risk estimates improved significantly after treatment across all risk groups, suggesting the tool may help monitor response to obesity therapies.

The researchers believe this approach could eventually support more personalized obesity care, rather than treating all individuals with similar BMI levels the same way.

Why Could This Matter for India’s Growing Obesity and Diabetes Burden?

The findings could have major implications for India, where obesity, diabetes, and metabolic diseases are rising rapidly across both urban and rural populations.

A review published in BMC Nutrition found that overweight and obesity among Indian adolescents have increased sharply over recent decades. The review analyzed 93 studies and reported obesity prevalence ranging from 0.3% to 24.6%, while overweight prevalence ranged from 1.25% to 35.8% among adolescents (3^✔ ✔Trusted Source
Prevalence and associated risk factors of overweight and obesity among adolescent population of India: a scoping review

Go to source).

Researchers identified several major contributors behind this rise, including:

• Physical inactivity
• High-calorie diets
• Increased screen time
• Urban lifestyles
• Junk food consumption
• Reduced sleep duration

The review also noted that India is undergoing a major nutritional and lifestyle transition, with increasing adoption of sedentary behaviors and processed-food consumption patterns (3^✔ ✔Trusted Source
Prevalence and associated risk factors of overweight and obesity among adolescent population of India: a scoping review

Go to source

).

At the same time, diabetes rates continue to rise across the country. A study published in Frontiers in Endocrinology reported that the incidence of diabetes in India increased from 162.74 to 264.53 per 100,000 population between 1990 and 2021. The study projected that diabetes prevalence and disability-adjusted life years (DALYs) will continue increasing through 2031 (4^✔ ✔Trusted Source
The rising burden of diabetes and state-wise variations in India: insights from the Global Burden of Disease Study 1990–2021 and projections to 2031

Go to source).

Southern and western Indian states such as Tamil Nadu and Goa showed some of the highest diabetes burdens, while experts warned that obesity, physical inactivity, unhealthy diets, and high BMI remain major contributors to India’s growing metabolic disease crisis.

Researchers say tools such as OBSCORE could eventually help identify high-risk individuals earlier and improve prioritization of preventive care and obesity treatment in countries facing rapidly rising metabolic disease burdens.

Could Artificial Intelligence Change the Future of Obesity Care?

Artificial intelligence and machine learning are increasingly being explored as tools for obesity prediction, prevention, and long-term disease management.

A review published in Medicina explained that AI systems can analyze large volumes of health data from electronic medical records, wearable devices, imaging scans, and lifestyle information to more precisely identify obesity-related risks than traditional methods (5^✔ ✔Trusted Source
The Role of Artificial Intelligence in Obesity Risk Prediction and Management: Approaches, Insights, and Recommendations

Go to source).

Researchers believe AI-driven systems may eventually help doctors:

• Predict obesity complications earlier
• Personalize treatment plans
• Improve long-term patient monitoring
• Identify high-risk patients for targeted interventions
• Support lifestyle coaching and preventive care

However, the OBSCORE researchers caution that the tool is not yet ready for routine clinical use. The current study mainly involved middle-aged and older adults from the UK Biobank, which may not fully represent broader populations worldwide. Predictive performance was also weaker for certain conditions, such as gastroesophageal reflux disease (GERD) and arthropathy.

Still, scientists believe the study marks an important step toward more personalized obesity care. Rather than relying only on BMI, future obesity management may increasingly focus on identifying which individuals are truly at the highest risk of developing serious long-term complications.

References:

1. Data-driven prioritization of high-risk individuals for weight loss interventions – (https://www.nature.com/articles/s41591-026-04353-2)
2. Beyond BMI: Rethinking Obesity Metrics and Cardiovascular Risk in the Era of Precision Medicine – (https://pmc.ncbi.nlm.nih.gov/articles/PMC12691488/)
3. Prevalence and associated risk factors of overweight and obesity among adolescent population of India: a scoping review – (https://pmc.ncbi.nlm.nih.gov/articles/PMC12139068/)
4. The rising burden of diabetes and state-wise variations in India: insights from the Global Burden of Disease Study 1990–2021 and projections to 2031 – (https://pmc.ncbi.nlm.nih.gov/articles/PMC12104079/)
5. The Role of Artificial Intelligence in Obesity Risk Prediction and Management: Approaches, Insights, and Recommendations – (https://pmc.ncbi.nlm.nih.gov/articles/PMC11857386/)

Source-Medindia