UK researchers develop tool to flag obesity risk most likely to benefit from weight-loss meds
Obscore AI – Misryoum reports a UK-developed AI score ranks people by 10-year risk of multiple obesity-related diseases—aiming to guide NHS access to weight-loss jabs.
A new UK research team has built a risk-scoring tool designed to pinpoint who is most likely to develop obesity-related illnesses over the next decade.
The focus, Misryoum says, is not on creating a one-size-fits-all approach based only on BMI.. Instead. the work tackles a practical bottleneck: weight-loss medications and related interventions are limited. while NHS prescribing decisions have often relied on a mix of body-mass measures and whether someone already has certain obesity-linked conditions.
The tool. described by researchers as a more personalised way to allocate care. uses an AI technique known as interpretable machine learning.. In the study. published in Nature Medicine. the team trained the system on data from nearly 200. 000 people in the UK Biobank project—each with a BMI of 27 or higher. placing them in the overweight or obese range.
What the team built, Misryoum reports, is called Obscore.. It sorts people into five categories for each of 18 obesity-related complications, ranging from low to high 10-year risk.. For the model to work. researchers identified 20 health. lifestyle. and demographic features that can help predict those outcomes—signals that include age. sex. total cholesterol. and creatinine levels.. For conditions like gout. stroke. and type 2 diabetes. the score estimates how many people in each risk group would be expected to develop the complication over a ten-year period.
That “same BMI, different risk” message is central to why clinicians may care.. The study shows that two people with the same age. sex. and BMI can face meaningfully different risks across different obesity-related conditions.. Misryoum notes this matters because it challenges the assumption that BMI alone captures the biology and health trajectory driving disease.
The researchers also highlight a potentially important blind spot in current approaches.. For some conditions—type 2 diabetes is named among them—the people placed in the highest risk group were sometimes more often overweight than obese.. That suggests that an exclusive focus on BMI thresholds could miss individuals who have not reached the most severe weight category but still carry a high probability of developing certain diseases.
Beyond building the score. the team tested it for validity using the UK Biobank dataset and additional information from two independent health studies.. They further checked performance in a different way by applying an adapted version of Obscore to data from participants in a randomised control trial of tirzepatide. a weight-loss drug.. Misryoum reports that those predicted to be at higher risk for obesity-related conditions achieved a similar level of weight loss to others. an indication that the risk stratification did not simply identify people who respond differently but rather identifies those more likely to benefit from preventing future complications.
Still, translating a promising model into routine care is rarely straightforward.. Misryoum notes that one external expert cautioned that many obesity-related conditions are tightly connected. and that for some complications. there are already risk scores that may be simpler or more established.. There’s also the matter of feasibility: some of the metrics used in the new tool may not be routinely available within the NHS. which could complicate day-to-day implementation.
The deeper issue is how health systems decide who gets limited interventions.. When access depends heavily on BMI and a narrow set of criteria. resources can miss people whose risk is high for reasons not captured by weight alone.. A tool like Obscore. if further refined and validated. could make the allocation process more “holistic”—shifting from a single measure to a multi-factor view of risk across several outcomes.
For patients, that could mean earlier targeting of prevention rather than reacting after complications emerge.. For clinicians and policymakers, it could mean more defensible prescribing decisions that align treatment intensity with expected benefit.. Misryoum also sees a broader scientific direction here: interpretable AI models are increasingly being used to turn complex clinical data into actionable guidance. while the emphasis on interpretability suggests an effort to keep clinicians in control of how predictions are understood.
The next step is likely careful testing in real-world NHS settings. including whether the required measurements are accessible. whether clinicians can use the score without burden. and whether outcomes improve when Obscore guides decisions.. Misryoum’s bottom line: the tool represents a meaningful attempt to move obesity care toward precision risk prediction—one that could help ensure weight-loss therapies are aimed at the people most likely to avoid serious future disease.