Using AI to make the science behind Kabuki Syndrome more accessible

Kabuki Syndrome is a rare disease affecting 1 in 32,000 births. There is currently no cure, but there is a fair clip of research being conducted and published every month.

As a parent of a child with Kabuki, I have long struggled to read and comprehend the dense, jargon-filled academic papers that report on this research, despite my great curiosity.

Due to the rarity of the condition, there is little to no journalistic reporting on these kinds of academic papers. I will share a project I’m working on that is intended to bridge that gap with AI, using the latest LLMs to craft stories more accessible to the broader Kabuki community.

In the demo I will show how I use AI to break down the papers into their individual claims, then how I transform those claims, using multiple prompts and workflows, into easy-to-read plain english versions.

These plain-english versions live in a UI that has two goals. One is to be polished and reader-friendly, up to the standard of a national magazine, with subheaders, pull quotes, etc. Another is to provide an inline interactive lens onto the original text of the article, with individual sentences that can morph to show the claims and sentences in the original paper they are based on.