Parth Shah

Biotech exec sharing views on #healthcare & #biopharma | Passionate about #leadership and advancing medicines to reach patients | Posts are my own | 🤝

[Parth Shah @ParthSharesBio]

Better use of digital and data in rare disease commercialization go hand in hand. However, applying judgment to these levers and creating an intentional strategy and org design to execute is the tough part (my brief take 👇). #pharma #biotech https://t.co/8IrcisUcRx

[The Assembly @InTheAssembly]

The Generate Platform is a design-build-test-learn loop combining computational AI with scalable biohardware.

It produces proprietary data and novel molecules for targets traditional discovery cannot reach.

Faster cycles. Higher success rates.

[ormoni bio @OrmoniBio]

We’re live. Ormoni Biosciences is industrializing peptide drug development: 30-day make-test-learn cycles, thousands of peptides per animal experiment, driving proprietary peptide PK/PD ML models. The result is a platform that generates DC-quality molecules in months.

[Tony Kulesa @kulesatony]

Tirzepatide was the 7,023rd peptide in Lilly’s GLP-1 series. In its first year, Ormoni Biosciences synthesized and assayed more than 250,000 unique peptides.

John Casey, Brad Pentelute, and the Ormoni team are bringing a step change to the development of new peptides.

There are hundreds of peptides that could treat disease, or enhance our health and longevity, but the challenge is making peptides into medicines: half-life, biodistribution, and uptake. Optimizing these properties requires many slow, expensive design-build-test cycles in animals.

Ormoni’s technology takes an idea to thousands of peptide candidates run in parallel in animal models in just 30 days. We will approach timelines to NHP and human PoC that have previously been unheard of in pharmaceuticals.

And with this collective dataset, we are training the most capable design and prediction AI models in the industry.

I met both founders John Casey and Brad Pentelute while at MIT, 13 years ago. John went on to co-found Inari Agriculture and Sail Biomedicines at Flagship Pioneering; I turned to him for advice many times on company building (and always with fingers crossed that one day we might work together). Brad welcomed me into his chemical biology class at MIT back in 2013, when my PhD advisor told me I should eat my vegetables and finally learn some chemistry. I loved it and was obsessed with his work for a decade; in the meantime Brad has become one of the world’s most renowned peptide chemists.

Pillar is proud to support the team.

[Dhruv Vasishtha @dvasishtha]

The strongest biopharma companies are going to need to own more AI-native preclinical R&D.

Large pharma isn't ignoring AI, but they are heavily partnering for next gen techbio. Lilly has TuneLab, NVIDIA, Isomorphic, Insilico, and OpenAI. Novartis has Isomorphic. Roche has Recursion. Sanofi has OpenAI / Formation Bio. Takeda has Nabla. Amgen has Generate. Pfizer has XtalPi and CytoReason. BMS has AI Proteins.

These are pointed at real discovery work: target identification, antibody and protein design, small molecule generation, binding prediction, disease modeling, and patient stratification. It's overwhelmingly buy vs. build tho.

The foundation model companies, however, are moving toward the upstream biology layer. Google has Isomorphic. OpenAI has GPT-Rosalind and protein engineering work with Retro Bio. Anthropic acquired Coefficient Bio. NVIDIA is building BioNeMo and drug discovery / autonomous lab infrastructure.

For these companies, life sciences may be one of the few markets large enough to help underwrite the next leg of valuation growth. That likely means they cannot stop at selling software, compute, or copilots. The more valuable position is generating targets, binders, proteins, molecules, and eventually licensable IP.

That creates an uncomfortable dynamic for biopharma: if AI-native platforms increasingly generate the upstream discovery substrate, does pharma remain the best owner of the discovery layer, or does it become the clinical development and commercialization partner for biology invented elsewhere?

IMO forward thinking pharma companies will need to in-house or acquire AI-native preclinical R&D teams the same way they historically treated medicinal chemistry, platform biology, and BD as core strategic functions for M&A.

[Parth Shah @ParthSharesBio]

@pnatarajanmd @Amgen Congrats and welcome!

[Parth Shah @ParthSharesBio]

@EricTopol @NEJM @skathire Great advance of science and I think for the right patient types certainly a viable option. This is a great one to watch to see patient acceptance of gene therapy at scale.

[Matthew Kirshner @MattyKirsh]

Great chart

[Zihai Li, MD, PhD @Zihai]

Super nicely done, Eric!! There is no question that the era of #ImmunoOncology has arrived. What remains up for debate in some circles is whether IO is simply one pillar of cancer medicine or a truly the foundation upon which the future of cancer therapy will be built.

[Mike Rea @ideapharma]

Credit to IDEA, where it’s due…

As part of the Pharmaceutical Innovation and Invention Index, released at the @statnews West event

@matthewherper @ADeAngelis_bio

[Healthcare AI Guy @HealthcareAIGuy]

The AI in drug discovery market map

[Jason Kelly @jrkelly]

Lilly is only biopharma so far leaning in on consumer biotech applications. I'm hopeful the rest of the industry will wake up soon. Huge market and opportunity to create products that people love.

[NVIDIA Healthcare @NVIDIAHealth]

(1/2)
🚨 Data scarcity is the #1 blocker in medical imaging AI.

We built the open-source fix.

NV-Generate-CTMR synthesizes realistic 3D CT & MRI volumes at scale - with paired segmentation masks - so you can train more robust models without touching real patient data.

[Parth Shah @ParthSharesBio]

@dbittermanmd @PearlF Congrats !

[Alex Karnal @alex_karnal]

We may be underestimating how deeply connected human biology really is.

A large new study published in @Lancet Psychiatry found associations between GLP-1 use and lower risks of depression, anxiety, substance use disorders, and other psychiatric outcomes across nearly 95,000 people.

We are still early in understanding what is driving this. Some of it is likely indirect: better metabolic health, weight loss, lower inflammation, improved sleep, mobility, and the psychological impact of finally feeling healthier again.

But there is also growing evidence that GLP-1s interact more directly with the brain’s reward and impulse pathways in ways that extend beyond appetite regulation.

For decades, medicine has treated metabolic health, cardiovascular health, and mental health as separate systems. Biology may not work that way.

The same pathways tied to obesity and insulin resistance may also influence addiction, compulsive behavior, mood, and cognitive health.

There is still a lot we need to learn, but the potential impact of these medicines keeps getting bigger - especially as they grow increasingly more affordable.

#Healthstack #Healthspan #Longevity #PatientBack

https://t.co/yHcwOD1TzX

[Biosimilars Council @BiosimsCouncil]

There is a lot of misinformation floating around when it comes to biosimilar medicines and interchangeability.

Here's the bottom line: Interchangeable biologics ARE NOT superior versions of biosimilars.

Unfortunately, the statutory distinction between biosimilars and interchangeable biosimilars continues to generate confusion and misinformation about the safety of biosimilar medicines. @US_FDA has consistently affirmed that there is no scientific difference between biosimilars and interchangeable biologics; and recently recommended that Congress remove this unnecessary distinction. The Biosimilar Red Tape Elimination Act is consistent with FDA’s science-based recommendation and represents an important step toward building confidence and streamlining patient access to biosimilar medicines.

Learn more: https://t.co/HOBMt1BXnG
@AccessibleMeds

[Sanat Mishra @sanatmishra7]

Everyone is racing to build medical AI agents.

Almost nobody is stress-testing the judges grading them.

We tested frontier models as judges on @OpenAI HealthBench answers.

Then we hit the answers with adversarial perturbations:

Delete a negation.
Tamper with clinical values.
Reverse the conclusion.
That’s it.
Most LLM judges missed the mistakes.

@claudeai Opus 4.7 missed an obvious medical lie 83% of the time.

@Gemini 3.1 Pro Preview was the best model we tested and still missed 43%.

Why this matters:

These judges are becoming reward models, safety filters, eval graders, and training signals for healthcare agents.

If the judge can’t catch the lie, the model trained on that judge learns the blind spot.

The failure does not stay in the eval.

It propagates.

First post in our @getbiostack x VibeOps Research series on where LLM-as-judge breaks in medical AI.

Full piece:
https://t.co/koc7HvRqzz

Thumbnail artwork inspired by H. Matisse.

[Samuel Hume @DrSamuelBHume]

Clinical trial abundance: progress

Almost 5,000 trials for innovative drugs were started last year — more than twice as many as 10 years ago

[Parth Shah @ParthSharesBio]

@angelosgeo @Medicenna1 Congrats @angelosgeo

[Parth Shah @ParthSharesBio]

@MichaelAlbertMD A great mission ! Congrats

[Google for Health @GoogleForHealth]

New from @GoogleDeepMind: Announcing AI co-clinician, a research initiative exploring the potential for real-time multimodal AI as an assistive component of the care team. https://t.co/qQzc5fu7O3