| language: | |
| - hi | |
| - en | |
| tags: | |
| - medical | |
| - india | |
| - text2text-generation | |
| - reportraahat | |
| datasets: | |
| - ekacare/clinical_note_generation_dataset | |
| # ReportRaahat Medical Simplifier | |
| Fine-tuned `google/flan-t5-small` on EkaCare's Indian clinical note generation dataset. | |
| **Task**: Takes a raw medical finding (test name, value, status) and outputs a | |
| layman Hindi or English explanation a patient can understand. | |
| ## Usage | |
| ```python | |
| from transformers import pipeline | |
| pipe = pipeline('text2text-generation', model='CaffeinatedCoding/reportraahat-simplifier') | |
| result = pipe('simplify medical finding: Hemoglobin 9.2 g/dL LOW') | |
| print(result[0]['generated_text']) | |
| ``` | |
| ## Part of ReportRaahat | |
| Built for HackerzStreet 4.0 @ Manipal University Jaipur. | |
| Paired with a FAISS RAG layer over NidaanKosha 100K Indian lab readings. | |