AI Model Detects Diabetes and Sorts Records Into Four Diagnostic Categories
Key Takeaways:
- Researchers have built a machine learning framework that first detects diabetes and then assigns positive records to one of four categories: prediabetes, type 1 diabetes, type 2 diabetes, or diabetes arising from pancreatic disease.
- XGBoost was the authors’ preferred classifier, though the paper reports inconsistent performance rankings across its own tables, with random forest outperforming it in one comparison.
- The framework is a proof-of-concept only. It has not been externally validated, its two stages were trained on separate datasets, and it is not ready for clinical use.
Why diabetes subtyping is a difficult problem
Diabetes is among the most common metabolic conditions worldwide, and its prevalence continues to rise. People living with diabetes typically experience raised blood glucose levels caused by insufficient insulin secretion, insulin resistance, or a combination of the two. Where hyperglycaemia persists, it can lead to serious complications affecting the eyes, heart, kidneys and nerves.
That burden has prompted interest in new strategies for detecting and classifying diabetes using clinical data that is already routinely available. If such tools were externally validated, they could in principle help clinicians identify individuals who warrant further diagnostic assessment, and support decisions about dietary, lifestyle or pharmacological management.
A study accepted for publication in Scientific Reports sets out one such approach: a machine learning (ML) model built on common clinical variables and a derived pancreatic-health index, designed both to detect diabetes and to classify it. The authors are explicit that clinical utility, patient outcomes and quality of life were not assessed.
About the study
The researchers presented an integrated, ML-based approach with two stages. Binary classification was used to determine diabetes status, and multiclass classification was then used to assign records to one of four dataset classes: prediabetes (PD), type 1 diabetes (T1D), type 2 diabetes (T2D), and diabetes from pancreatic disease, also known as pancreatogenic or type 3c diabetes (T3cD).
Two publicly available datasets were used. For the binary task, the team drew on the Pima Indians Diabetes Database, maintained by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), to separate records into diabetic and non-diabetic. For the multiclass task, they used a dataset from the Kaggle repository. The curated multiclass dataset comprised 21,539 samples, split 60% for training and 40% for testing.
How the models were built and tested
Inputs to the multiclass model included age, body mass index (BMI), waist circumference, cholesterol levels, blood glucose levels, insulin levels, and a derived pancreatic-health index.
Several ML algorithms were compared to identify the most effective approach for each classification task: logistic regression, decision trees, random forests, K-nearest neighbours (KNN), naive Bayes, and XGBoost.
To correct class imbalances, the team applied the Synthetic Minority Oversampling Technique (SMOTE) to the training data only. Hyperparameter tuning was then carried out to identify the best-performing parameter combinations, and the models were retrained using those selected parameters. Finally, the researchers ran a Local Interpretable Model-agnostic Explanations (LIME) analysis on their preferred classifier to examine how individual features contributed to model predictions.
What the models achieved
The authors selected XGBoost as their preferred classifier, although the paper’s reported performance rankings are not consistent across tables. Table 2 gives a value of 0.97 for every evaluation parameter, covering accuracy, precision, recall and F1 score. Table 7, however, reports an accuracy of 95.67% for XGBoost against 96.67% for random forest, with random forest also achieving a marginally higher macro-average ROC-AUC, a measure of how well a model discriminates across the four classes.
The researchers attributed XGBoost’s performance to its capacity to capture complex, non-linear associations between features. In the reported test results, XGBoost correctly classified all prediabetes and T1D records, though some confusion persisted between the T2D and T3cD groups. KNN was the least accurate of the algorithms tested.
This kind of discrepancy between a paper’s headline claim and its own supporting tables is exactly the sort of detail clinicians are increasingly expected to spot for themselves. CCH’s CPD-accredited short course AI Essentials for Primary Care covers structured appraisal of AI tools, including the SAFER Evaluation Framework and how to judge AI output against NHS standards.
Which features drove the predictions
General feature-importance analysis pointed to blood glucose levels, insulin and BMI as the most influential variables. The LIME analysis told a slightly different story, indicating that glucose dominated the model’s predictions, with age and cholesterol making secondary contributions in some classes. BMI, waist circumference, insulin and pancreatic health had comparatively lower influence in the LIME results.
Blood glucose levels showed the strongest reported correlation with the class label, at a Pearson’s correlation of 0.86, while insulin showed a correlation of 0.59. These correlations should be treated with caution, since numerical values were assigned to what are, in fact, nominal disease classes.
Age, BMI and waist circumference showed moderate to strong intercorrelations, ranging from 0.63 to 0.68. Their respective correlations with the target were 0.41, 0.46 and 0.56. Pancreatic health showed a negligible negative correlation of -0.06.
In practice, the model primarily learned blood glucose-based decision patterns, which are consistent with the clinical diagnosis of diabetes. The authors interpreted these patterns as broadly concordant with diabetes pathophysiology and existing clinical knowledge, though that interpretation was not independently validated in clinical practice. On the evidence presented, the model is not ready for clinical use and would require considerably more evaluation before it could support, rather than replace, clinical judgement.
Conclusions and future directions
The study demonstrates an ML framework capable of classifying diabetes status and assigning positive cases to four labels: prediabetes, T1D, T2D and T3cD. The authors suggest the framework could eventually assist with diabetes screening and help guide further diagnostic investigation. Crucially, the study did not establish whether using the model improves care, treatment outcomes, quality of life, or the wider global burden of diabetes.
The varying influence of lipid, pancreatic and body measurements may reflect genuine subtype-related biological differences. Equally, it may be an artefact of dataset construction, class coding, correlated predictors, or the absence of clinically verified biomarkers. The authors recommend that these patterns be investigated in clinically characterised datasets before any mechanistic conclusions are drawn.
The framework should currently be regarded as a modular proof-of-concept, because its binary and multiclass stages were trained on separate datasets that may differ in population, variables and collection methods.
Future work, the authors suggest, should use a single training dataset containing both diabetes status and clinically adjudicated subtype labels, including verified pancreatic and autoimmune markers rather than derived variables. External validation in larger and more diverse clinical cohorts would be needed to improve generalisability. Ethical and data privacy concerns would also need to be addressed before any AI-based model of this type could be translated into clinical screening or decision-support settings.
CCH insight
Studies like this one arrive faster than most clinicians can appraise them, and the gap between a promising accuracy figure and a tool that is safe to use in practice is wide. Knowing how to interrogate that gap is now a core professional skill.
AI Essentials for Primary Care is a 100% online, CPD-accredited short course providing 3.5 CPD hours and a Certificate of Completion. It equips the whole primary care team to evaluate AI tools against NHS standards, recognise when AI output should be questioned, and apply the SAFER Evaluation Framework in day-to-day practice. No technical background is required.
Explore AI Essentials for Primary Care →
Read More
Novo Nordisk’s Oral Semaglutide Shows Cardiovascular Benefits Comparable to Wegovy Injection
Key Takeaways:
- Novo Nordisk’s new oral semaglutide 25 mg pill improved blood glucose control and reduced cardiovascular risk factors, matching the effects of its injectable counterpart, Wegovy.
- Data from the OASIS 4 trial showed significant weight loss and normalisation of blood glucose in people with prediabetes.
- The company expects U.S. regulatory approval for the first oral GLP-1 treatment for weight management by the end of 2025.
Oral GLP-1 pill shows comparable benefits to injection
Novo Nordisk has presented new findings suggesting that its experimental oral obesity medication delivers cardiovascular and metabolic benefits similar to those achieved with its blockbuster injectable, Wegovy. The results were shared at the ObesityWeek 2025 conference in Atlanta and strengthen the Danish company’s case for approval of the pill in the United States later this year.
The oral semaglutide 25 mg tablet, part of the company’s glucagon-like peptide-1 receptor agonist (GLP-1RA) portfolio, was shown to improve blood sugar regulation and reduce cardiovascular risk factors. These results could mark a milestone in obesity care, as the pill would become the first oral GLP-1 therapy approved for weight management.
OASIS 4 trial results
The data come from the OASIS 4 clinical trial, which compared oral semaglutide 25 mg with placebo in adults with overweight or obesity. After 64 weeks, 71.1% of participants with prediabetes who received the treatment achieved normal blood glucose levels, compared with 33.3% in the placebo group.
Participants who lost at least 15% of their body weight experienced additional health benefits, including reductions in blood pressure, inflammatory markers, and triglycerides. Overall, the trial demonstrated both significant weight loss and improvements in cardiometabolic health outcomes.
The primary OASIS 4 results, published in September in the New England Journal of Medicine, reported an average weight loss of 16.6% among participants taking the oral semaglutide.
Comparable outcomes with Wegovy injection
An indirect comparison between OASIS 4 and Novo Nordisk’s earlier STEP 1 trial, which evaluated injectable semaglutide (Wegovy), found the two formulations delivered comparable outcomes in weight reduction and improvements across key cardiometabolic markers.
These findings suggest that people who prefer not to use injectables could soon have an equally effective oral alternative. As demand for obesity pharmacotherapy continues to rise, an oral formulation may further expand access and adherence to GLP-1 treatments.
Regulatory outlook and market plans
The U.S. Food and Drug Administration (FDA) accepted Novo Nordisk’s application for oral Wegovy in May and is expected to deliver a decision by the end of the fourth quarter of 2025. The company has stated that, if approved, it intends to launch the product shortly thereafter.
Despite a recent dip in share price and slower sales growth, Novo Nordisk’s prospects have been buoyed by positive trial outcomes and an improved pricing arrangement under Medicare. The company is also undergoing leadership changes, including a new Chief Executive Officer and a restructured board, amid efforts to stabilise growth.
Novo Nordisk has indicated that, once approved, the pill will be made available through telehealth platforms such as Ro and WeightWatchers, with a potential subscription model offering discounted pricing. Additionally, Hims & Hers Health recently confirmed it is in discussions with Novo to provide both injectable and oral forms of Wegovy through its digital platform.
A step forward in accessible obesity care
If approved, Novo Nordisk’s oral semaglutide could redefine accessibility and adherence in obesity care. The convenience of a pill that matches the efficacy of an injectable treatment offers a compelling new option for people managing obesity and related cardiometabolic risks.
By broadening the range of treatment modalities within the GLP-1 class, Novo Nordisk continues to shape the evolving landscape of obesity pharmacotherapy — a field that is rapidly transforming the management of metabolic health worldwide.
CCH insights
This news about oral semaglutide is very welcome, but shouldn’t come as a surprise. Oral semaglutide is the exact same compound as injectable semaglutide, and as long as the dose administered orally is sufficient to deliver a similar blood concentration as the injectable form, then the effects should be very similar. It’s the same drug, just a cheaper and easier, but less efficient, route of administration.
Read More
Strawberries May Support Glucose Control and Reduce Inflammation in Prediabetes, Study Finds
Key Takeaways:
- Daily intake of freeze-dried strawberries for 12 weeks improved fasting glucose and reduced inflammation in adults with prediabetes.
- Antioxidant biomarkers including superoxide dismutase, glutathione, and β-carotene significantly increased during strawberry supplementation.
- Findings suggest strawberries could serve as a practical, food-based intervention to prevent progression to type 2 diabetes.
Strawberries and prediabetes: A promising link
A new randomised controlled trial published in Antioxidants has found that consuming freeze-dried strawberries (FDS) daily may help reduce fasting blood glucose and vascular inflammation in adults with prediabetes. The research also showed marked improvements in antioxidant status, highlighting the potential of strawberries as a simple dietary intervention for metabolic health.
The authors concluded that “strawberries may represent a practical dietary intervention that improves fasting glucose and strengthens antioxidant defence in adults with prediabetes.”
Understanding prediabetes and oxidative stress
Prediabetes represents a critical stage between normal glucose metabolism and type 2 diabetes mellitus (T2DM). It is characterised by mildly elevated blood glucose levels, which contribute to oxidative stress and low-grade inflammation.
High glucose levels increase reactive oxygen species (ROS), impairing insulin function and damaging pancreatic β-cells. Proinflammatory cytokines such as tumour necrosis factor-alpha (TNF-α) further aggravate insulin resistance by interfering with glucose uptake and triggering inflammatory pathways. Over time, these effects contribute to vascular dysfunction and atherosclerosis through increased endothelial adhesion molecules and reduced antioxidant enzyme activity.
Dietary antioxidants – including vitamins, polyphenols, and carotenoids – can neutralise oxidative stress. However, studies using supplements have produced inconsistent results due to differences in absorption and bioavailability. Evidence from clinical trials and meta-analyses indicates that plant-based antioxidants can improve total antioxidant capacity and glycaemic outcomes in people with prediabetes or T2DM.
Berries, particularly strawberries, are rich in polyphenols such as anthocyanins and ellagic acid, which are known to enhance antioxidant enzyme activity and improve insulin sensitivity. Previous studies using FDS have already shown benefits for inflammation and oxidative stress in metabolic disorders, providing a foundation for this new research.
Study design and methodology
Researchers at the University of Nevada, Las Vegas, conducted a 28-week randomised controlled crossover trial involving 25 adults who met the American Diabetes Association’s diagnostic criteria for prediabetes.
Each participant completed two 12-week phases: one with daily FDS intake and another control period without strawberries, separated by a four-week washout. Participants were randomly assigned to begin with either the FDS or control phase.
During the intervention, participants consumed 32 grams of FDS powder per day – equivalent to roughly 2.5 servings of fresh strawberries – containing dietary fibre, polyphenols, flavonols, and anthocyanins. They were instructed to maintain their usual diet and physical activity throughout the trial.
Compliance was carefully monitored using dietary logs, returned powder packets, and plasma ellagic acid levels. Blood samples were collected at baseline, 12, 16, and 28 weeks to measure fasting glucose, antioxidant enzyme activity, total antioxidant capacity, and vascular adhesion molecules using standardised assays. Carotenoid levels were analysed via high-performance liquid chromatography (HPLC).
A mixed-model analysis of variance (ANOVA) was used to evaluate treatment effects while accounting for treatment period, randomisation order, age, sex, fasting glucose, and baseline values. Power analysis confirmed the study was adequately powered to detect meaningful changes in metabolic and antioxidant markers.
Improvements in antioxidant and metabolic markers
Results showed high adherence rates, with more than 85% compliance confirmed by elevated plasma ellagic acid during the FDS phase.
Compared with the control period, strawberry supplementation produced significant improvements in several antioxidant biomarkers, including superoxide dismutase, glutathione (GSH), total antioxidant capacity (AC), and β-carotene. No significant changes were observed in catalase, glutathione reductase, glutathione peroxidase, or α-carotene.
Fasting blood glucose levels also decreased significantly during the FDS period, indicating better glycaemic control. Moreover, markers of vascular inflammation – particularly intercellular adhesion molecule (ICAM) and vascular cell adhesion molecule (VCAM) – were notably reduced. Levels of P-selectin and E-selectin remained unchanged.
Correlation analyses revealed modest inverse relationships between ICAM and GSH, AC, and β-carotene, and between VCAM and AC, suggesting that stronger antioxidant status was associated with reduced vascular inflammation.
Only minor side effects were reported, such as mild gastrointestinal discomfort and headaches.
Implications and limitations
The findings suggest that incorporating strawberries into the diet could help strengthen antioxidant defences, lower inflammation, and improve fasting glucose regulation in people with prediabetes. These benefits may be linked to polyphenols enhancing glutathione synthesis and superoxide dismutase activity, alongside carotenoids and anthocyanins reducing oxidative stress and endothelial dysfunction.
The study’s strengths include its randomised crossover design, objective biomarker measurements, and the use of a realistic dietary dose of strawberries. However, the relatively small and predominantly female sample, the absence of a placebo control drink, lack of participant blinding, and single-site recruitment limit the generalisability of results.
The study was funded by the California Strawberry Commission.
A food-based approach to diabetes prevention
In summary, consuming a daily portion of freeze-dried strawberries for 12 weeks led to measurable improvements in antioxidant capacity, fasting glucose, and vascular inflammation among adults with prediabetes.
While further research in larger, more diverse populations is needed, these results point to strawberries as a simple, accessible dietary strategy that could help prevent the progression from prediabetes to type 2 diabetes in everyday clinical and public health settings.
CCH insights
It is great to have research that shows health benefits from eating strawberries, because nearly everyone loves strawberries and there aren’t many foods that are extremely popular and good for us. However, the amount of freeze-dried strawberries consumed in this study would set you back about £20 per week – not a huge amount, but during a cost-of-living crisis might not be feasible for many people. This study was, unsurprisingly, funded by the California Strawberry Commission, and it begs the question whether eating other berries or certain other foods might not have a similar effect – but credit to the strawberry industry for making the effort to do the research.
Read More