Archive
1250 articles, sorted by publication time.
Five AI agents with clearly defined roles worked together—from trial protocols and medical-record fields to causal analysis—to reconstruct 20 historical studies. The results demonstrate the potential to support study design while also highlighting the limits that prevent real-world data from replacing randomized trials.
Researchers identified a candidate small molecule capable of simultaneously targeting SMO and c-MET that showed activity in a three-dimensional model comprising human cancer cells and fibroblasts; however, the findings remain at the non-peer-reviewed in vitro stage.
Northwestern University will spend $20 million over four years to build the DREAM platform, allowing researchers to remotely specify desired protein functions and then use AI and robots to iteratively design, manufacture, and test them; however, the production target of 300,000 proteins remains a plan, not a demonstrated result.
Generative AI helped create D16-M1P2 from 2,023 candidate structures. The molecule can both direct cells to clear PKMYT1 and inhibit its residual activity; it showed oral activity in mouse tumor models, but a long road of validation remains before human trials.
The research team altered the splice acceptor of HTT exon 13, enabling the protein to bypass a cleavage site prone to generating toxic fragments. Brain atrophy and multiple functional deficits were subsequently reduced in mice, but challenges involving brain delivery and long-term safety still stand between the therapy and human trials.
Relation will use automated experiments to generate large-scale human cell perturbation data, then feed it to the MORGAN model to identify disease targets; what the deal is really betting on is whether AI can derive new leads from verifiable biological responses.
Researchers reanalyzed old trial data from more than 300 patients with progressive supranuclear palsy using a revised scale and found that davunetide may slow deterioration in women; however, this post hoc subgroup finding cannot overturn the original trial’s negative conclusion.
This multimodal model learned cancer trajectories from the laboratory tests, vital signs, and chest X-rays of millions of people, initially demonstrating potential across early warning, staging, and prognosis; however, population bias and its retrospective design still leave it short of clinical validation.
For patients with advanced disease who cannot receive PD-1/PD-L1 immunotherapy, this TROP2 antibody-drug conjugate extended median overall survival by five months; however, its approved scope, toxicity management, and reimbursement in individual countries still define its real-world impact.
SAR445399, designed to simultaneously block three inflammatory signals, was intended to assess its potential to clear mucus plugs using lung imaging; shortly after the trial began, it was paused for strategic business considerations, leaving both efficacy and safety unanswered.
In an exploratory trial of 61 adults, both single and repeated nasal dosing improved anxiety scores, but neither outperformed placebo in the primary analyses; the signal in the severe subgroup awaits confirmation in a larger study.
The deal brings an inherited macular disorder with no disease-modifying therapy into Tarsus’s pipeline; the real test is a global Phase 3 trial requiring two years of follow-up, with its primary assessment not expected to be completed until 2029.