KOL Pulse

AI in Oncology — KOL Digest

KOL DigestMonday, September 7, 2026 All editions →

This edition

A randomized trial finds AI trial notifications alone don't move enrollment, Astra drafts five R01s in an hour — and OpenEvidence ships its own models

In this edition: Roupen Odabashian, MD on the randomized trial where AI-triggered trial notifications did not change enrollment — and on launching an AI agent that tracks approvals · Benjamin Herzberg, MD on the slowest step in AI drug development · Jason Locasale on Astra drafting five NIH R01 proposals in about an hour · Jame Abraham, MD on OpenEvidence's new model family · Katy Beckermann, MD on chart-finding time · the Oncology Brothers on bias, privacy, and liability · Arturo Loaiza-Bonilla, MD on verifying AI alignment · Hidehito Horinouchi, MD on the IASLC AI Taskforce surveys · NEJM AI on agentic research workflows · Plus the thread: Masahiro Torasawa, MD, PhD on the WCLC26 abstract database he built with Astra, and Guilherme Correia, MD's response.

Trial enrollmentAgentic AIClinical workflow

Roupen Odabashian@RoupenMD on X

Oncologist; host, Delta HealthTech Innovators podcast; founder, MeDucation AI (maker of OncoPulse).

Sep 5, 2026

No oncologist can read every FDA approval and every phase 3 readout. We all pretend otherwise.

The honest math: dozens of label changes a year, hundreds of practice-relevant trials, and a clinic schedule that does not move. So you skim an abstract on your phone between patients and hope you did not miss the one that changes what you do on Monday.

Today we are launching OncoPulse from @MeDucationai , and it is free.

You create an account, pick the topics you care about, and set up an AI agent that tracks them for you. Mine emails me at 8am every day with what moved in GI oncology. That is it. No feed to scroll, no newsletter you never open.

Oncologists, hematologists, NPs, PAs, fellows, students. If you are responsible for knowing what changed, this is built for you.

Set up your agent here: meducationai.com/tutor/signup ↗

Every oncologist should have an agent doing this. Tell me what your agent should be watching.

View post on X ↗

Sep 4, 2026

A randomized trial notified oncologists whenever a patient's tumor genomics matched an open clinical trial. 20,707 patients with genomically characterized solid tumors. The notifications worked. Enrollment did not move.

Same trial accrual in the AI arm as in the control arm.

Every oncology AI pitch deck I have seen in the last two years assumes the bottleneck is identification. Find the eligible patient, surface the trial, and enrollment follows. This trial says otherwise. The AI found the patients. The patients still did not enroll.

If you have ever tried to actually enroll someone, this is not surprising. The bottleneck is the 90 minutes of coordinator time, the insurance call, the biopsy slot that is three weeks out, the patient who lives two hours from the site and cannot take Thursdays off. None of that is an information problem. All of it is a logistics problem wearing an information problem's clothes.

The uncomfortable read: a lot of clinical AI is being built to solve the easy half of a hard problem, then measured on whether it solved the easy half.

Which makes me wonder what the right endpoint even is. Not "did we identify more eligible patients," but "did anyone get a drug they otherwise would not have gotten." Almost nobody reports the second one.

PubMed header of the randomized trial 'Clinical Trial Notifications Triggered by Artificial Intelligence-Detected Cancer Progression' (Mazor et al., PMID 40257799).

View post on X ↗

Benjamin Herzberg@bherzbergmd on X

Thoracic oncologist; associate director of Phase 1 research, Columbia.

Sep 4, 2026

View post on X ↗

Jason LocasaleAcademic scientist@LocasaleLab on X

Academic scientist and professor.

Sep 6, 2026

I asked Astra to look through my published science, identify new directions, and draft five NIH R01 proposals -- the multiyear, $1M-$2M grants that support much of academic biomedical science.

In about one hour, using roughly $20 in compute, it produced entirely reasonable proposals. They were well grounded in my work and structured to appeal to study section reviewers. They looked very similar to what I would write if my goal were to get another grant.

Astra reproduced that logic well. It turned existing work into questions, hypotheses, and experiments that would look familiar and defensible to a study section.

The output was immediately recognizable as the work of an experienced academic scientist. If AGI means being able to produce the work of highly trained professionals, this is a clear concrete example. It produced something academic scientists spend a great deal of time doing.

Academic medical centers build careers, laboratories, and their finances around this process. Scientists spend years writing these proposals as the next fundable extension of their work.

AI-drafted NIH R01 Specific Aims page titled 'Predictive Principles of Metabolic Control Across Nutrient Environments' — three aims on metabolic control in breast cancer models, drafted by Astra from Locasale's published work.

Second AI-drafted Specific Aims page, 'Metabolic and Epigenetic Mechanisms of Cellular Memory of Nutrient Exposure' — three aims on methionine exposure and H3K4 methylation. — Two of the five AI-drafted Specific Aims pages attached to the post; the rest are on X.

View post on X ↗

Free Access · KOL Pulse Intelligence

Track oncology’s top KOLs in your specialty

Create a free account, pick the tumor types you cover, and go beyond the tweet — the intelligence pharma teams use to map influence and prepare for engagement:

✓Enhanced KOL profiles
✓Pharma influence & potential advisory-board patterns
✓2025 Open Payments financial analysis
✓Social-media sentiment & trend signals

Get free access — pick your specialty

Choose the tumor types you follow. No cost, unsubscribe anytime.

Jame Abraham, MD, FACP@jamecancerdoc on X

Professor and chair, Hematology & Medical Oncology, Cleveland Clinic; deputy editor, The ASCO Post.

Sep 4, 2026

View post on X ↗

Katy Beckermann@katy_beckermann on X

GU medical oncologist — clinical trials and drug development.

Sep 3, 2026

View post on X ↗

Oncology Brothers@OncBrothers on X

Community oncologists Rohit Gosain, MD (Roswell Park) and Rahul Gosain, MD (Wilmot Cancer Institute).

Sep 2, 2026

View post on X ↗

Arturo LoAIza-Bonilla, MD MSEd@DrArturoAI on X

Hematologist-oncologist; network hem/onc chief, St. Luke's; co-founder, Massive Bio.

Sep 6, 2026

View post on X ↗

Hidehito HORINOUCHI@HHorinouchi on X

Medical oncologist, National Cancer Center Japan.

Sep 4, 2026

🔥IASLC #AI Artificial Intelligence Taskforce Survey
🆙 @IASLC #AI Taskforce
☑️Leads: @mihaela_aldea @SeastedtMD Dr. Muhammad Rafiqul Islam
🎯Parallel, mutually complementary surveys for physicians and patients
🎯Addressing global challenges in AI accessibility
▶️Patient: iaslc.co1.qualtrics.com/jfe/form/SV_eQ… ↗
▶️Physician: iaslc.co1.qualtrics.com/jfe/form/SV_cw… ↗
#LCSM @OncoAlert @Larvol @EGFRResisters @ALKPositiveinc @ros1cancer @KRASKickers @Exon20Group @RETRenegades @metcrusaders
*Image Created by @HHorinouchi with Gemini

IASLC Artificial Intelligence Taskforce graphic 'Two Perspectives. One Important Conversation.' with QR codes for the parallel patient and physician surveys on AI in lung cancer care.

View post on X ↗

NEJM AIPublication@NEJM_AI on X

Journal on medical artificial intelligence from NEJM Group.

Sep 1, 2026

NEJM AI Perspective Figure 1 — worked example of an agentic research workflow from clinical question to auditable output, noting capability may remain constrained to well-resourced settings.

View post on X ↗

Discussion

An oncologist's WCLC26 abstract database — built with Astra

Masahiro TORASAWA, MD. PhD.@M_Torasawa

🫁 Ready for #WCLC26! 🇰🇷
I’ve put together a Notion database:
📚 210 Plenary, Oral & Mini Oral presentations
📝 181 summaries of abstracts released so far 🏷️ Biomarker, stage & treatment tags

https://t.co/rHMSF0NSgz

View post on X ↗

Guilherme S. C. Correia, MD@guicorreiamd

This database is amazing! Thank you!
Helped me during ASCO, and now for WCLC!
👏🙏🏻

View post on X ↗

Masahiro TORASAWA, MD. PhD.@M_Torasawa

@guicorreiamd Thank you! I used GPT-6 Astra for most of the work this time. 🤖
It’s made things so much easier than when I used to do everything manually! 😂

View post on X ↗

Guilherme S. C. Correia, MD@guicorreiamd

@M_Torasawa Amazing! Hopefully we will get a chance to connect at WCLC. I would love to talk about lung cancer and AI, and some collaboration ideas

View post on X ↗

Masahiro TORASAWA, MD. PhD.@M_Torasawa

@guicorreiamd Absolutely!
I’d love to connect at WCLC and chat with you!

View post on X ↗

Compiled and reviewed by the KOL Pulse research team, led by Brian Shields, Founder, KOL Pulse. Quotes are verbatim from their authors’ public posts on X. Last updated September 7, 2026.