--- dataset_info: features: - name: data dtype: string - name: criteria dtype: string - name: NCT_ID dtype: string - name: metadata struct: - name: brief_title dtype: string - name: conditions sequence: string - name: interventions sequence: string - name: location_countries sequence: string - name: nct_id dtype: string - name: official_title dtype: string - name: recruitment_information struct: - name: primary_completion_date(actual)(final_data_collection_date_for_primary_outcome_measure) dtype: string - name: study_completion_date(actual) dtype: string - name: study_start_date(actual) dtype: string - name: study_design struct: - name: allocation dtype: string - name: interventional_model dtype: string - name: masking dtype: string - name: phase sequence: string - name: primary_purpose dtype: string - name: study_type dtype: string - name: study_record_dates struct: - name: study_record_updates struct: - name: last_update_posted(estimated) dtype: string - name: last_updated_that_met_qc_criteria dtype: string - name: last_verified dtype: string - name: study_registration_dates struct: - name: first_posted(estimated) dtype: string - name: first_submitted dtype: string - name: first_submitted_that_met_qc_criteria dtype: string splits: - name: train num_bytes: 906076192 num_examples: 213877 - name: test num_bytes: 112448496 num_examples: 26735 - name: validation num_bytes: 112313748 num_examples: 26735 download_size: 555641648 dataset_size: 1130838436 configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* - split: validation path: data/validation-* --- # Dataset Card for EC-RAFT Raw ClinicalTrials.gov Dataset ## Dataset Summary This dataset provides a structured version of ClinicalTrials.gov data from , prepared for use in the EC-RAFT framework. It includes structured eligibility criteria (inclusion, exclusion, age, sex), trial descriptions, metadata, interventions, and study design fields. This dataset was used as the foundation for the paper: **EC-RAFT: Automated Generation of Clinical Trial Eligibility Criteria through Retrieval-Augmented Fine-Tuning** (*ACL 2025 Findings*). The EC-RAFT model and training pipeline are available at: 👉 [https://github.com/biodatlab/ec-raft](https://github.com/biodatlab/ec-raft) ## Dataset Structure ### Data Instances Each example is a single clinical trial record with structured metadata and eligibility criteria. ## Dataset Splits | Split | Type | Number of Trials | Total Inclusion Criteria | Total Exclusion Criteria | Mean Inclusion Criteria per Trial (± SD) | Mean Exclusion Criteria per Trial (± SD) | | -------------- | -------------- | ---------------- | ------------------------ | ------------------------ | ---------------------------------------- | ---------------------------------------- | | **Train** | Interventional | 168,429 | 838,948 | 1,256,242 | 4.98 ± 5.11 | 7.46 ± 7.05 | | | Observational | 45,448 | 137,234 | 156,298 | 3.02 ± 2.85 | 3.44 ± 3.68 | | **Validation** | Interventional | 20,928 | 103,910 | 154,896 | 4.97 ± 5.04 | 7.40 ± 7.00 | | | Observational | 5,807 | 17,531 | 20,470 | 3.02 ± 2.77 | 3.53 ± 3.64 | | **Test** | Interventional | 21,129 | 103,982 | 156,212 | 4.92 ± 5.01 | 7.39 ± 7.05 | | | Observational | 5,606 | 16,990 | 19,000 | 3.03 ± 2.99 | 3.39 ± 3.61 | ## Citation If you use this dataset, please cite: TBD ## Contributions * Nopporn Lekuthai (Mahidol University) * Nattawit Pewngam (Ravis Technology) * Supitcha Sokrai (Ravis Technology) * Titipat Achakulvisut (Mahidol University) ## Related Resources * [EC-RAFT Github](https://github.com/biodatlab/ec-raft)