What rare disease AI teaches us about longitudinal health

I started working on Aether after watching my mother fall sick.

She was diagnosed with an exceptionally rare form of cancer only a few days before her death. All the reports were there. But no one ever had a complete picture of her health over time.

Recently, I read a new npj Digital Medicine paper on rare disease diagnosis using weakly supervised transformers. What struck me was not the model, but the premise: diagnoses are noisy, provisional, and only make sense longitudinally.

Rare diseases expose what breaks first when systems are episodic.

I wrote a longer essay connecting this research to longitudinal health records, health graphs, and EHR Lite design here:
https://myaether.live/blog/rare-disease-ai-longitudinal-health

Would genuinely appreciate thoughts from others building in healthcare, AI, or data systems that have to deal with messy real-world truth.