Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.
To compare the prognostic performance of an artificial intelligence (AI) model with that of experienced clinicians in predicting tooth loss over a 10-year period. An AI model trained on structured clinical and radiographic data was compared with 12 periodontists and 11 general dentists (GDs), who independently assigned prognostic scores (0-10 scale) to 300 teeth with known 10-year outcomes.
Researchers tested a field-scale targeted sampling strategy inside a regional hybrid model that combines machine learning and geostatistics for soil organic carbon mapping. The method pairs a long-term satellite NDVI-based Productivity Index with regional prediction uncertainty to choose sampling locations, with new observations incorporated only via local residual kriging. Across 28 agricultural fields, the regional model alone averaged 0.24% SOC RMSE, while adding all field samples reached 0.17% and the four-sample targeted approach reached 0.18%.
Researchers applied unsupervised hierarchical clustering to 294 rheumatoid arthritis patients to integrate clinical, demographic, and genetic data related to Tumor Necrosis Factor inhibitor response. By publication date 2026-08-17 they reported distinct responder characteristics and identified three subgroups ranging from 73.5% response to 82.9% therapeutic failure.
In a study published August 12, 2026, researchers quantified CO2eq emissions for training ResNet-50, DenseNet-121 and EfficientNet-B0 on 128,907 chest radiographs for 20 epochs. They found validation loss minima at median epochs 2 to 4, meaning most emissions occurred after the best checkpoint, and compared retrospective selection, prospective early stopping, and fixed-epoch training on AUC and energy use.
Published September 8, 2026 as an opinion piece, the article argues that ambient AI tools that summarize patient accounts and generate clinical notes can introduce interpretive drift, a subtle change in meaning between what the patient said and what is recorded.
A 2026 conceptual review in Medical Education examined how artificial intelligence used to generate, score and interpret assessments affects validity. Using Kane's four inferences, the authors mapped threats such as prompt instability and domain shift and noted that AI assessment is advancing without formal scrutiny comparable to clinical AI.
On 2026-09-08, a peer-reviewed study reported development of an integrative clinical-molecular model for glioma malignancy grading using 400 patients from a single-center retrospective cohort. Using LASSO and machine learning, the authors built a Random Forest classifier based on seven predictors including age, KPS, tumor diameter, inflammatory ratios, IDH status and Ki-67, achieving AUC 0.864 in training and 0.820 in validation.
A systematic review covering January 2013 to June 2025 examined non-contact and AI-assisted neonatal heart rate monitoring versus conventional ECG. It found a progressive shift toward camera-based photoplethysmography, thermal imaging, and AI-enhanced multimodal systems that showed strong correlation with ECG, rapid acquisition, and better robustness to motion and lighting.
ace