Investigators across Europe leverage PRAEVAorta2 AI-driven segmentation on pre- and post-EVAR CT angiograms, combining imaging and clinical variables in deep learning models to forecast postoperative outcomes and optimize surveillance strategies for aortic aneurysm patients.

Key points

• Automated segmentation and morphometric measurement of aneurysms using CE-marked PRAEVAorta2 on CT angiography
• Integration of clinical, procedural, and imaging features into deep convolutional neural networks for postoperative risk stratification
• Multicenter retrospective cohort of 500 EVAR patients with 70/30 training-testing split to develop and validate predictive models

Why it matters: This protocol establishes AI-enabled precision surveillance and risk stratification post-EVAR, potentially reducing complications and personalizing vascular care.

Researchers at the Second Affiliated Hospital of Army Medical University develop a CatBoost model enhanced by active learning to predict Philadelphia chromosome-positive acute lymphoblastic leukemia using routine clinical and laboratory parameters, with feature selection via BorutaShap and interpretability via SHAP.

Key points

• Ten routine clinical and laboratory features—age, neutrophil and monocyte counts, liver enzymes, among others—are selected via BorutaShap.
• CatBoost model integrated with an active learning algorithm achieves validation AUC of 0.797 and external AUC of 0.794 for Ph+ALL prediction.
• SHAP analysis identifies age, monocyte count, γ-glutamyl transferase, neutrophil count, and ALT as critical drivers of model output.

Why it matters: This interpretable ML approach enables early, low-cost detection of Ph+ALL in settings lacking genetic testing, improving diagnostic access and guiding timely treatment choices.

A research team at NOVA University Lisbon performs a comprehensive scoping review of supervised ML frameworks—including XGBoost, Random Forest, and LASSO—leveraging electronic health record datasets to predict 30- and 90-day heart failure hospitalisation and readmission risks, emphasizing ensemble methods and the current lack of economic impact assessments.

Key points

• Ensemble algorithms (XGBoost, CATBOOST) achieved top predictive performance with mean AUC up to 0.88 for unspecified-period heart failure risk.
• EHR-derived datasets across 13 countries provided clinical, demographic, and utilization variables for 30- and 90-day risk modelling.
• No reviewed studies included economic evaluations, indicating a critical gap for assessing cost-effectiveness before clinical deployment.

Why it matters: This synthesis underscores ensemble ML's potential to refine heart failure risk stratification and highlights gaps in cost-effectiveness evaluations crucial for clinical adoption.

Researchers at Amsterdam University Medical Centres deploy AI to analyse local field potentials recorded by Medtronic’s Percept PC deep brain stimulation system. By correlating spectral features from implanted electrodes with smartwatch kinematics and clinical ratings, they aim to generate patient‐specific neuronal fingerprints to optimize stimulation for Parkinson’s disease in real‐world settings.

Key points

• Longitudinal multimodal dataset of 100 Parkinson’s patients with sensing‐enabled STN DBS.
• AI algorithms correlate LFP spectral power and volatility with wearable kinematic metrics and UPDRS scores.
• Patient‐specific neuronal fingerprints drive development of adaptive, responsive DBS programming.

Why it matters: This AI‐driven approach represents a shift toward personalized, responsive brain stimulation, potentially improving efficacy and reducing side effects compared to continuous DBS.

A cross-disciplinary team from Sichuan University's NICUs employs a machine learning pipeline to classify neonatal intestinal diseases using bowel sound recordings captured by a digital stethoscope. They preprocess audio with filters, extract time–frequency features such as MFCCs, and train a transformer-based model combined with a Random Forest to detect conditions like NEC, FPIAP, and obstruction, aiming to supplement subjective clinical assessment with objective, automated diagnostics.

Key points

• Collected neonatal bowel sounds via 3M Littmann 3200 digital stethoscope with 2-minute recordings from six abdominal regions, filtered to exclude noise exceeding 30%.
• Extracted acoustic features—zero-crossing rate, spectral centroid, chroma, MFCCs—after pre-emphasis, framing, and Hamming windowing, forming a multidimensional feature vector.
• Trained a Random Forest for disease detection and a transformer-based network for multi-class classification (NEC, FPIAP, volvulus, obstruction), validated via tenfold cross-validation and external cohorts with high AUC.

Why it matters: An AI-based bowel sound diagnostic tool offers rapid, noninvasive neonatal intestinal disease screening, potentially reducing delays and improving outcomes compared with subjective auscultation.

A team from Jordan University of Science and Technology and Al-Najah National University conducted a bibliometric analysis of AI applications in early detection and risk assessment of noncommunicable diseases. They retrieved publications from Scopus (2000-2024) and used VOSviewer for network mapping to highlight research hotspots and collaboration trends.

Key points

• Scopus query (2000-2024) yields 1,745 publications on AI in early NCD detection, totaling 37,194 citations.
• Annual publication and citation counts exhibit exponential growth, peaking in recent years.
• Core journals include Scientific Reports and IEEE Access; top institutions are Harvard Medical School and China’s Ministry of Education.
• Leading countries are China, USA, India, UK, and Saudi Arabia, with strong USA–India collaboration.
• VOSviewer mapping highlights hotspots like machine learning, deep learning, CNNs, and disease-specific studies in Alzheimer’s and diabetes.

Why it matters: This study offers a panoramic view of AI's growing influence on early NCD detection and risk evaluation, guiding researchers and policymakers toward emerging trends and collaboration opportunities. By mapping key journals, institutions, and hotspots, it informs resource allocation and fosters data-driven strategies to advance proactive disease management.

A scoping review in BMJ Open examines factors influencing clinician AI adoption. It highlights performance expectancy and facilitating conditions as key drivers across various care settings. For instance, improved workflow integration and targeted training can boost AI acceptance in clinical practice.

A recent BMJ Open trial introduces the iRITUX protocol—an AI-driven approach to tailor rituximab dosing in membranous nephropathy. Researchers, including Teisseyre and Destere, conducted the study across 13 French hospitals. By predicting underdosing risks early, the protocol refines treatment, improving clinical remission. This work highlights AI’s potential to customize therapies in chronic kidney disorders.

The BMJ Open study by Li et al. presents a machine learning framework for predicting lymph node metastasis in gastric cancer. By integrating clinical features like tumor size and T category, the model achieved promising accuracy in cross-validation, suggesting its potential as a personalized risk assessment tool in oncology.

A recent BMJ Open study examined various machine learning models, including XGBoost and logistic regression, to predict type 2 diabetes using lifestyle and body measurements. By highlighting factors such as age and waist circumference, the research points to promising non-invasive early risk assessments for diabetes, paving the way for improved preventive strategies.

A recent BMJ Open study applied seven machine learning models to Tanzanian survey data, revealing that the random forest classifier achieved 95% accuracy in predicting zero-dose children. The research illustrates how statistical tools can identify key factors, like maternal unemployment and low education, to drive public health interventions.

Researchers led by Raghunathan and Morris present a scoping review protocol investigating AI in allied health. The study examines how AI supports disciplines such as physiotherapy and occupational therapy, addressing benefits like enhanced patient safety. For example, improved diagnostics are highlighted. Published in BMJ Open (2025), this review underscores transformative potential and challenges in integrating digital technologies in healthcare.

A systematic review by BMJ Open examined 52 ML studies in rheumatoid arthritis, revealing that 42 studies ignored sex bias issues. This omission, despite skewed data, underlines a gap in addressing fairness in healthcare. It’s an important cue for professionals to revisit bias mitigation in clinical research for more reliable outcomes.