On this special episode of Tell Me Where IT Hurts, host Dr. Jay Anders welcomes David Lareau, CEO of Medicomp Systems. Medicomp Systems is the sponsor of the Tell Me Where IT Hurts podcast.
Together with Dr. Jay Anders, Lareau discusses the growing role for artificial intelligence (AI) in healthcare technology, including how AI documentation summarization is improving but not yet perfect. Lareau also addresses Dr. Anders’ concerns that many ambient listening technologies driven by AI and large language models (LLMs) remain only about 80% accurate.
Lareau explains that shortcomings in AI and LLMs can be minimized by linking these technologies to a clinical knowledge graph and by filtering results through internal evidence-based links in Medicomp’s clinical intelligence engine to flag inconsistencies for healthcare providers to check.
“One of the things causing people to turn to Medicomp is that they need to trust the information they get.”
David Lareau, CEO, Medicomp Systems
Issues around “dirty data” remain, Lareau says, especially the need to clean incoming patient data that varies from disparate coders and coding systems, or with no coding or structure at all. If someone has asthma, you might have 8 to 10 different ICD-10 codes depending on who coded it, for example, making it more challenging to synthesize data and summarize patient history. But using AI, a synonymy engine can help find anything that sounds like that diagnosis.
As for the one thing he would change in healthcare, Lareau would like to see systems open up more and allow outside utilities and vendors to add on capabilities not currently supported within their systems.
Show Notes
Among the topics covered:
- The importance of building trust, and addressing hallucinations and omissions, in ambient listening and LLM outputs
- Converting text to structured clinical data so it is usable for patient care and downstream applications
- Augmenting ambient documentation so clinicians can act on the information
- How intelligent chart analysis can filter and summarize patient data to present clinically-targeted details to the provider
- The need for tools to clean up clinical data so it is trusted and usable now that interoperability frameworks have data flowing through the pipes
- Solutions for ensuring acceptable operating costs when using LLMs to summarize records on an ongoing basis
- The differences between implementing healthcare IT solutions in the U.S. versus other parts of the world
- “If you could change one thing…”
- And more…
Show Links
- David Lareau on LinkedIn
- Medicomp Systems on LinkedIn
- Dr. Jay Anders on LinkedIn
- Tell Me Where IT Hurts on Healthcare NOW Radio