Researchers at Thomas Jefferson University have developed a groundbreaking automated machine learning (AutoML) model that can accurately differentiate between two common types of brain tumors using preoperative MRI scans, potentially improving surgical planning and patient outcomes.

The study, published in the December 2025 issue of Otolaryngology–Head and Neck Surgery, represents the first application of AutoML technology specifically trained to classify pituitary macroadenomas and parasellar meningiomas—two benign but challenging-to-distinguish brain tumors that require different treatment approaches.

Drug discovery is a high-risk, high-cost endeavor, with over 90% of candidates failing in clinical development. A significant proportion of these failures stem from unresolved biopharmaceutic challenges—poor solubility, limited permeability, transporter-mediated efflux, and extensive first-pass metabolism—which compromise a drug's absorption and bioavailability, rendering it ineffective in vivo. Traditional formulation and screening methods have proven inadequate for the modern pipeline of complex molecules. This review examines how the integration of artificial intelligence (AI)-driven predictive tools and innovative formulation technologies, such as 3D printing and biodegradable nanocarriers, is revolutionizing the approach to these age-old limitations, offering a pathway to more efficient and personalized drug development.

Researchers at the University of Kansas and University of Southern California led virtual support groups to measure dyadic alliance among participants, specifically what types of nonverbal behaviors notify that two people are making the connections that are vital to such groups. They also tested the feasibility of machine learning playing a role in such mental health settings, which could help address mental health professionals shortages, the authors write.

Decreasing Precision with layer Capacity trains deep neural networks with layer-wise shrinking precision, cutting cost by up to 44% and boosting accuracy by up to 0.68%

Researchers distill multi-teacher common knowledge + gradient alignment into one substitute model