Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.
As of November 2025, this peer-reviewed review examines agentic AI in newsrooms that can autonomously plan, decide, and generate content. Analyzing 46 sources from 2015-2025, it finds current evaluations emphasize technical accuracy and efficiency while neglecting trust, governance, and collaboration.
By August 2026, a multi-omics study combined single-cell and bulk RNA-seq from GEO with three machine learning algorithms to screen for kidney stone drivers, identifying GLS and LPIN2 as upregulated in fibroblasts and building a risk prediction nomogram.
Researchers used a Causal Forest causal machine learning model on a retrospective cohort of gastric cancer patients treated between 2007 and 2017 to estimate who benefits from adjuvant chemotherapy. The model identified lymph node ratio as the most dominant predictor of benefit, with a significant interaction at a 0.25 threshold in a propensity-matched Stage II-III cohort.
On August 22, 2026, researchers reported an integrative analysis combining single-cell RNA-sequencing with bulk transcriptomic and clinical data from TCGA-LIHC, GEO, and ICGC to find macrophage-associated biomarkers in hepatocellular carcinoma. They derived a 16-gene signature that stratified tumors by survival and clinicopathological features, with ridge regression showing AUC 0.988 in TCGA-LIHC and 0.895-0.924 externally, and XGBoost SHAP identifying LGALS1 as highly informative.
The paper examines how AI and extended reality enable creation of avatars and human digital twins from personal and biometric data that persist after death as Human Digital Remains. Using cross-disciplinary analysis and doctrinal review, it finds that current EU instruments do not extend protections to the deceased and identifies urgent legal and ethical gaps.
Researchers eye-tracked 100 cytotechnologists diagnosing 30 digital cytology images and then tracked 28 students before and after a 3-month training program. They found years of experience did not predict accuracy, while shorter fixation on the low-power field main object did, and students markedly improved time to first target fixation and reduced background attention after training.
By August 2026, peer-reviewed discussion in Giornale Italiano di Cardiologia described AI as entering hypertension care, able to give simple and well-documented answers to management questions for practicing physicians, while also being explored in research to identify secondary hypertension and predict future hypertension and complications such as heart failure.
In this cross-sectional study published August 1 2026, researchers asked GPT-4 via ChatGPT to answer 20 common psychosis psychoeducation questions sourced from a first-episode psychosis programme, then had two psychosis experts independently rate the answers on accuracy, clarity, inclusivity, completeness, clinical utility and overall quality.
ace