AI Deployment Digest

The Family Heart Foundation's machine learning model flags atherosclerotic patients most likely to have elevated lipoprotein(a) using routine EMR...

Wayve's appointment of ex-Waymo CFO Elisa de Martel highlights the startup's pivot from development to financing, commercial deals, and scaling for...

Fortell's founder was driven by a simple observation: people resist hearing aids the way they embrace glasses, highlighting that consistent use hinges...

Novo Nordisk is partnering with Anthropic to embed Claude Science directly into drug discovery workflows for chronic diseases.

• The companies will...

Google's MCP server rollout lets compatible AI agents control Nest and Matter devices while accessing event history and camera summaries via natural...

AI now generates thousands of protein binders in minutes, but wet-lab testing remains the choke point. Pairing cell-free synthesis with direct SPR on...

GE HealthCare's CareIntellect for Operations launches with Duke Health and The Queen’s Health Systems as first sites, shifting focus from prediction...

AI/ML adoption in life sciences outpaces existing legal frameworks, creating interconnected risks that must be addressed before moving from...

The shift of AI agents from demos to real workflows highlights that reliable execution layers and human-in-loop interfaces matter more than raw model...

Two perspectives converge on a key limit for diagnostic AI: earlier or more sensitive detection does not automatically improve patient results.

-...

Leaked iOS code reveals Siri can delegate tasks to Claude or fully swap in models like GPT-5.6, enabling actions such as creating reminders or...

• Deterministic first: Routine tasks like payroll run via APIs; AI agents only handle what software cannot
• Managerial AI: Agents review demo...

Apitegromab (ISEMBYLD) approval for spinal muscular atrophy blocks easy obesity use despite strong early tirzepatide data.

• Off-label access faces...

An optimized hybrid framework using ResNet50V2, Xception, and ensembles reached 100% accuracy on an internal same-source test set for multiclass...

Can biopharma AI platforms explain why programs fail and map competitive biology, not just list records?

DrugBank's platform tackled: "Who's got a...

Agent-MIRA retrieves phenotypically similar patient cases from imaging databases while surfacing treatment history, survival outcomes, and trial links...

Stanford Medicine's TranscriptFormer offers two new AI models of the biological cell, published in Nature, that open fresh horizons for cellular research.

AI triage can filter high-volume genomic candidates before scientists review them, yet functional validation in disease models remains the decisive...

Plasma p-tau217 leads single-biomarker screening, yet A/T/N panels with XGBoost and multimodal fusion reach AUC 0.94. SHAP interpretability and the Step-Care framework support scalable clinical translation.

AI speeds formulation timelines via Bayesian optimization, yet effective development still demands early integration of chemical, physical,...