BioASQ Participants Area
BioASQ - Task ELCardioCC: Clinical Coding of Greek Cardiology Discharge Letters
This challenge focuses on advancing methods for automatic clinical coding in cardiology, encouraging innovative solutions from researchers, clinicians, and data scientists.
The training set contains 3,000 discharge letters. Of these, 1,453 documents are annotated at both the document level and the mention level, making them suitable for NER and entity linking approaches. The remaining documents are annotated only at the document level.
The goal of the system is to produce document-level annotations. In addition to the training set, the label set is also provided. The training data is stored in JSONL format, where each line represents a single JSON object.
Document-level annotations are structured as a list of lists. For each inner list, the model is required to identify at least one of the included codes. Identifying more than one code from the same list does not yield additional points. However, if the model fails to identify any code from an inner list, it receives no credit for that list.
The training data can be found here . More info about the task https://elcardiocc.web.auth.gr/
