Skip to main content

Talk Bio

Search This Blog

Clinical trial flaws being scaled, not solved by AI: report

on May 19, 2026

Get link
Facebook
X
Pinterest
Email
Other Apps

Phesi founder warns AI 'is scaling, not solving' clinical trial flaws in a recent analysis of over 600,000 clinical trial protocols, finding that fewer than one in three protocols are clearly linked to publicly documented patient data and outcomes.

Phesi president and founder Dr. Gen Li warns that AI may scale underlying weaknesses if the datasets used are not properly contextualized against patient populations and treatment outcomes.

The company’s analysis suggests a growing gap between protocol design and real‑world patient outcomes, increasing the risk that AI tools used to design or optimize future trials may perpetuate flawed assumptions rather than correct them.

Sources:

Clinical AI Has Boomed. A New Stanford-Harvard State of Clinical AI ...

Reporting Guidelines for Clinical Trial Reports for Interventions ...

Revolutionizing clinical trials: the role of AI in accelerating medical ...

Artificial Intelligence Applied to clinical trials: opportunities and ...

Can AI Help Solve the Issue of High Failure Rates in Clinical Trials?

Opportunities and Risks: Generative AI Use for Clinical Research

Phesi founder warns AI 'is scaling, not solving' clinical trial flaws

Best Practices in Writing a Clinical Trial Lay Summary with AI

AI's bold leap in clinical trials: rewriting the rules - YouTube

AI in clinical trials: Current status, challenges, and future ...

Methodological and Quality Flaws in the Use of Artificial Intelligence ...

AI Helps Scientists Correct Mistakes in Medical Studies | PSC

[PDF] Intelligent Clinical Trials: Using Generative AI to Fast-Track ...

Clinical Trials Data Quality Drug Development flaws patient outcomes Phesi protocol design scaling not solving

Get link
Facebook
X
Pinterest
Email
Other Apps

Popular Posts

KOL Bulletin: Brilaroxazine's Potential Differentiation on EPS Over Risperidone in Schizophrenia

A leading US KOL states that brilaroxazine shows placebo-like extrapyramidal symptoms (EPS) at week 4, potentially differentiating it from risperidone which cannot achieve this 1 . Reviva plans to initiate the RECOVER-2 Phase 3 trial for brilaroxazine in schizophrenia in H1 2026 following FDA recommendation for additional efficacy and safety data 2 . Brilaroxazine demonstrates broad-spectrum efficacy across schizophrenia symptom domains, including negative symptoms, with a well-tolerated safety profile in over 900 subjects 2 4 . Phase 3 RECOVER trial data shows low EPS and akathisia, mild weight gain (1.52 kg pooled), reductions in prolactin levels, and improvements in sexual function over 1 year 4 . New publication highlights speech latency as an objective vocal biomarker for brilaroxazine's effect on negative symptoms, reinforcing efficacy 3 5 . Sources: 1. https://firstwordhealthtech.com 2. https://www.biospace.com/press-releases/reviva-announces-regulatory-update-...

Xaira Therapeutics Launches X-Cell: First AI Virtual Cell Model Testing Scaling Laws on Massive Perturbation Data

Xaira Therapeutics announced X-Cell on March 17, 2026, a 4.9-billion-parameter diffusion language model, the largest virtual cell model to date, trained on X-Atlas/Pisces, the largest genome-wide CRISPRi Perturb-seq dataset with 25.6 million perturbed single-cell transcriptomes across seven diverse contexts. 1 2 X-Cell predicts cell responses to genetic perturbations, outperforming existing models by up to five-fold on metrics like Pearson delta, and generalizes zero-shot to unseen biological contexts such as iPSC-derived melanocyte progenitors and primary human CD4+ T cells. 1 2 The model follows power-law scaling dynamics similar to large language models, using a novel diffusion architecture that iteratively refines predictions. 2 Xaira, founded in 2024 with over $1 billion funding, plans to expand X-Atlas to primary cells, organoids, and in vivo perturbations, focusing on inflammatory/immunological diseases and antibody therapeutics. 1 3 A subset of the Pisces dataset and X-...

Ultragenyx Resubmits BLA for UX111 Gene Therapy with New Long-Term Data for Sanfilippo Syndrome Type A

Ultragenyx resubmitted its Biologics License Application (BLA) to the FDA for UX111, an AAV9 gene therapy for Sanfilippo syndrome type A (MPS IIIA), including long-term clinical data on neurologic, biomarker, and safety improvements. 1 2 3 4 5 The resubmission addresses a Complete Response Letter from July 2025 related to manufacturing (CMC) observations and adds durable biomarker reductions, such as CSF heparan sulfate, with favorable safety profile. 2 4 5 FDA's six-month priority review has begun; PDUFA action date expected in Q3 2026. 2 4 5 If approved, UX111 would be the first therapy for this fatal pediatric neurodegenerative disorder. 1 3 4 5 New long-term data announced February 3, 2026, shows substantial biomarker improvements and functional benefits vs. natural history, consistent across age and severity. 2 Sources: 1. https://simplywall.st/stocks/us/pharmaceuticals-biotech/nasdaq-rare/ultragenyx-pharmaceutical/news/ultragenyx-bla-resubmission-puts-ux111-and-ra...

Powered by Blogger

X-Data LLC