Ontology-Agnostic EHR Tokenizer MEDIATOR Boosts Prediction and Transfer Accuracy
Start here
A discrete tokenizer called MEDIATOR maps clinical concept text from electronic health records into a shared vector-quantized codebook, without manual mappings or a fixed vocabulary, making it ontology-agnostic. Evaluated on 20 MIMIC-III tasks, it raises mean AUPRC by 4.0 points over the strongest prior tokenizer. In zero-shot transfer to 14 eICU tasks, where it can use all events, the gain reaches 8.6 points. Controlled experiments attribute in-domain accuracy mainly to the text encoder, while the discretization objectives drive robustness in transfer settings. The method stands as a validated approach for cross-ontology EHR modeling; open questions include performance beyond the tested datasets and tasks.
Written by a model from the reports · updated 4 hours ago
Coverage timeline
Follow the coverage to see the different angles.
- arxiv.orgDiscrete Neural Interlingua for Ontology-Agnostic EHR Modeling
MEDIATOR is an ontology-agnostic discrete tokenizer that maps arbitrary clinical concept text into a shared vector-quantized codebook, without manual mappings or a fixed vocabulary. It raises mean AUPRC over the strongest prior tokenizer by 4.0 points on 20 MIMIC-III tasks, and by 8.6 points in zero-shot transfer to 14 eICU tasks while using all events. Control experiments show the text encoder drives in-domain accuracy, while discretization objectives drive transfer robustness.
Coverage and heat for this story
Reports so far 1·Heat now 9·Peak heat 10 (Oct 21 20:00)·Reports, 24h +1
Two panels, same clock (pointer stays in sync): top is cumulative editorial reports — only rises; last point = All reports; slope = new reports that hour. Bottom is Live-board heatIndex — 7-day participants with a 24-hour half-life, so it can fall when talk cools.