Archive
1250 articles, sorted by publication time.
Researchers incorporated two often-omitted regions of the Plasmodium CSP protein into a nanoparticle vaccine design. Mouse experiments showed that they can elicit distinct antibody responses, but the existing R21 architecture remains the stronger benchmark for protection.
The research team produced 16 viable bacteriophages from nearly 300 artificially designed genomes. When used in combination, they even overcame bacterial resistance to the natural virus; this proof of concept opens a new path for antimicrobial treatment while bringing the safety governance of gene synthesis to the fore.
A perspective article in Nature Reviews Drug Discovery notes that improvements in model accuracy have yet to translate into clear clinical benefits; the next generation of evaluations should ask whether AI truly improves R&D decision-making.
Shape Therapeutics and the Rett Syndrome Research Trust are expanding their collaboration, using AI to select guide RNAs and harnessing cells’ own ADAR enzymes to correct MECP2 mutations. Mouse studies showed RNA editing in the brain and symptom improvement, but the human dose, safety, and eligible patient population remain to be clarified.
Raidium integrates lesion detection, segmentation, and longitudinal RECIST measurements into a single imaging workflow, but this deployment is limited to clinical trials and oncology research; performance evidence and FDA review remain two hurdles before it can enter routine clinical care.
The first patient with an advanced solid tumor has been dosed in the Phase 1 trial of ZI-MA4-1. The therapy seeks to combine precise cancer recognition with the advantages of off-the-shelf manufacturing, but at present only the trial’s initiation can be confirmed; no objective tumor response results are yet available.
Model-designed viruses can now infect bacteria in the laboratory and provide new genetic material for overcoming bacteriophage resistance; but they remain far from treating patients, and current DNA synthesis screening may not recognize highly novel sequences.
Researchers used a genomic model to design nearly 300 bacteriophages and successfully produced 16 viruses capable of infecting laboratory E. coli; the results demonstrate the potential of AI-generated living systems, but do not yet constitute a clinical therapy for antibiotic-resistant bacteria.
Whole-genome data from more than 24,000 people show that parental aging, intracytoplasmic sperm injection, ovarian stimulation, and embryo manipulation each correspond to distinct patterns of de novo mutations; however, these population-level associations cannot yet be translated into individual risk estimates or treatment recommendations.
Fifteen patients received local infusions of B7-H3-targeted CAR-T cells, with no dose-limiting toxicities; disease control was achieved in eight evaluable patients. However, the sample was extremely small and there was no control group, and the survival data still require validation in a phase 2 trial.
A novel dendritic cell vaccine connects CD4 T-cell memory left by the spike protein to tumor-recognizing CD8 T-cell responses; it has shown efficacy in melanoma and breast cancer models, but the real test will begin with an early-stage trial in patients with sarcoma.
The PSMAddition trial showed that adding lutetium-177–PSMA-617 earlier to dual hormonal therapy can delay radiographic disease progression; however, overall survival, long-term toxicity, and treatment costs remain unresolved.