PORTER: Language-Grounded Event Representations for Portable Structured EHR Foundation Models
arXiv:2606.24102v1 Announce Type: new Abstract: Most electronic health record (EHR) foundation models encode clinical events as discrete event tokens from a fixed vocabulary and therefore cannot directly represent events containing unseen concepts or new combinations of concepts and attributes such as numeric values. This limits transfer across institutions and even across deployment pipelines within the same institution. We introduce PORTER, a language-grounded structured EHR foundation model that decouples event representation from this fixed vocabulary. PORTER represents events through thei
The proliferation of AI in healthcare necessitates more flexible and portable models for Electronic Health Records (EHRs) to overcome current limitations of fixed vocabularies and ensure broader applicability.
This development addresses a critical bottleneck in deploying AI across diverse healthcare institutions by enabling foundation models to handle previously unseen clinical concepts and attributes.
Clinical AI models can now represent events more dynamically and transfer effectively across varying healthcare systems, fostering better interoperability and broader adoption of AI in medicine.
- · Healthcare AI developers
- · Hospitals and healthcare systems
- · Patients (through improved AI applications)
- · Electronic Health Record (EHR) vendors
- · AI models reliant on fixed vocabulary
- · Fragmented healthcare data systems
Improved generalizability and interoperability of AI models in healthcare, particularly for EHR data.
Accelerated development and adoption of AI-powered diagnostic and treatment tools due to more robust data handling.
Potential for a unified, multimodal representation of patient health across different medical institutions and even national borders, enabling large-scale population health analyses.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL