Our Story
I'm an Ear Nose and Throat physician/surgeon and over the years used the joke ‘If only I could clone myself.’ Often this was in response to unacceptable wait times. Most patients would laugh with me but little did they know I was serious! “I’m not joking” would get another chuckle. But now technology has advanced and it it’s entirely possible to clone domain specialized knowledge! I’m really Not joking. Stay with me and read on for a bit. We learn by trial and error with numerous domain-specific examples that are not shared effectively through peer-reviewed papers. This knowledge is distributed across the world and is lost when a domain specialist retires or is no longer active in the field for any reason. This information is always multimodal in nature, and therefor sharing is a complex process. Multimodal inputs with Large Language Models now make it possible to process, store, and deliver domain-specific solutions once thought impossible. We can now store my knowledge, aggregate it with other medical providers, and then deliver a solution that can be cloned and replicated. Slow down and let that sink in for a bit.
The future of medicine lies in improving the way we share learned information. We are providing the beginnings of this amazing journey by reducing the limits of HIPAA compliance, having to keep a model up-to-date, dealing with server reliability, and most importantly, use and delivery, so now healthcare can iterate at an acceptable rate.
Sharing the details is the whole point. I in vision a platform where medical providers everywhere can share any small detail without burden of a peer review in a similar fashion as arXiv for other domains. Medicine will need a data sets and examples in a multimodal fashion as everything matters. For example diagnosing a chronic cough will likely include entire exam details from a directed history of present illness a physical exam of many systems and organs from from heart and lungs, oral exam, nasal exam with possible nasal endoscopy, abdominal exam, and that’s just a good starting point. This will require more than text files but audio, vision, touch, specialized measurements and lab values. The data will be immense but necessary.
It might be just making us aware of some learned knowledge that no one even thought to consider. I’m tired of waiting. Let’s do this together.