Pediatric Moonshot Update

November 2024 Highlights

The Pediatric Moonshot mission is to reduce healthcare inequity, lower healthcare costs, and improve outcomes for children — locally and globally — by creating privacy-preserving, real-time applications based on access to data in all 1,000,000 healthcare machines in all 500 children’s hospitals in the world.

In this update: A privacy-focused, distributed AI cloud for healthcare, over 140 curated AI/deep learning imaging apps, the “Total Recall” LLM app, and talks at key conferences.

More than ever, AI applications are becoming a key component in reducing healthcare inequity, lowering costs and improving outcomes. Building accurate, non-biased AI applications in healthcare and life sciences requires large amounts of diverse training data. All the data we need is in the buildings — available real-time inside the clinical machines, research machines and PACS and EMR systems. Centralized architectures, which have powered mainstream AI solutions like ChatGPT, will not work in medicine. Here, the data sizes are much larger, the demands for privacy much higher and the need for real-time results much greater.

The Answer is a Distributed AI Cloud Infrastructure

By working closely with BevelCloud, we have engineered a privacy-preserving, real-time, distributed AI cloud infrastructure. Most of our focus recently has been on connecting with Build Partners to help complete the world’s first distributed AI Lab with 32 sites, and over 2,000 TB/year — roughly the equivalent of uploading 1 billion pictures on Facebook — of real-time imaging data as well as offline data from PACS and EMRs.

With the help of a fantastic team of CHOC summer interns, we have curated a list of AI/deep learning applications trained on medical images from ultrasound, CT, MRI, etc. and applied to specializations ranging from cardiology to oncology. All were trained on small amounts of data, and all are candidates for moving from the bench to the bedside using the Distributed AI Lab. Check out https://appcommons.bevelcloud.io to see the posted apps.

If you know of more examples, please let us know!

Total Recall Application - a Patient Digital Twin

While training AI/deep learning imaging applications requires breakthroughs in distributed learning with text data, the advent of LLMs has made a major leap. The BevelCloud architecture enables access to EMR data, so we created an application, right now only a pilot, which is the nearest thing to a digital twin of a patient. Training the LLM on the medical record of a specific patient is allowing us to create a simple application we call “Total Recall.” We’ve been demonstrating this based on training with medical record of a cancer patient, which is a document with over 4,000 pages.

See Total Recall digital twin in action on this YouTube video — And don’t miss the enthusiastic feedback from an oncology team after they have seen it in action!

Even More Support From Clinical Partners Worldwide

Thanks to Dr. Henrik Holmstrøm for the invitation to speak at the 52nd Nordic Pediatric Cardiology conference. You can find the YouTube talk at this link. It ends with a proposal to build a Distributed AI Lab in the Nordics

The Academic Nursing Leadership Conference in Washington DC was an opportunity to educate and inspire 400+ deans of nursing schools from around the US. The recorded talk is available on YouTube. Highlights include a short tutorial on AI, a description of the Pediatric Moonshot and one key, final question.

Join Us This Week in Seattle

Seattle Children’s Research Institute is hosting an event in Seattle this Friday, November 22, 2024. If you happen to be in Seattle, you’re invited to the Frontiers in Pediatric Research lecture series from 11:00am to 12:30pm. Just register at this link, I’ll be happy to meet you in person and talk about our mission.

If you’d like to learn more about any of the topics above, please reach out by replying to this email. If you support our mission, help spread the word by encouraging colleagues and friends to subscribe to our newsletter at pediatricmoonshot.com.

Stay tuned for more updates.

The Pediatric Moonshot Crew