Pediatric Moonshot October 2026 Update

Pediatric Moonshot

No one can accuse us of sending too many newsletters. Here is a quick update since July.

New Website

As part of our campaign to build the world’s first AI supercomputer for children’s medicine, we launched a completely new website on September 4. A gift of $100, once or monthly, helps bring this work to more children. Visit www.pediatricmoonshot.org and support us!

This AI supercomputer needs nodes to operate. If you have suggestions for who might be interested in sponsoring a specialized node in cardiology, oncology, neurology, or nephrology, let us know.

View full-size image

Disease-precise AI Agents: Next-Generation Computational Phenotyping

We prepared a test for 6 of our 176 kidney disease-precise agents on our privacy-preserving cloud infrastructure, the connecting tissue that will be used by the AI supercomputer, using patient data.

We just learned the term computational phenotyping. Disease-precise AI agents represent the ideal next step in this approach: they are continuous, not one-time; graded and explained, not binary; actionable; local and federated.

Getting More Children into Clinical Trials: Trial-precise AI Agents

Too few children with cancer ever enroll in a clinical trial. Many trials are canceled or delayed because recruiting is too difficult and expensive.

In August, we were invited to speak at the POETIC conference. We’re working with the leaders of POETIC, a 14-site clinical trial network, to deploy POETIC-AI, which will enable trial-precise AI agents to identify eligible children using computers, not human labor.

Disease-precise Next-Generation Learning Health System

We have been working with the National MS Society on building a next-generation learning health system.

The architect of an early general multiple sclerosis learning health system learned firsthand the challenges of bringing research and clinical data into a centralized architecture. We are co-authoring a white paper describing a next-generation learning health system based on privacy-preserving patient digital twins running in a distributed cloud infrastructure.

This new architecture applies to building learning health systems for each of the 618 cancers, cardiac diseases, kidney diseases, and neurological conditions.

An AI That Can Be Used For Good

In the news, AI can create deepfakes, eliminate jobs, and spread nonsense. But it can also help a doctor recognize a disease they may see once in their career. This technology isn’t good or bad on its own; what matters is what we point it at. We’re pointing it at children’s health. Read the full article on Medium.