{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/1e7e89542c6b4f28b34ccbb46d944d36\" frameborder=\"0\" width=\"1108\" height=\"831\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":831,"width":1108,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":831,"thumbnail_width":1108,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/1e7e89542c6b4f28b34ccbb46d944d36-4925e49dfa12bae7.gif","duration":288.173,"title":"Building a Traceable Patient Radiology Timeline","description":"This Loom explains a solution for automatically tracking changes in radiology findings across a patient’s history. It highlights that reports often use different wording for the same lesion, making manual comparison of growth like 3.2 cm to 3.8 cm difficult, and notes the gap in existing systems. The approach uses a hybrid extraction pipeline with handcrafted high-precision patterns and an analytical layer to map findings to standardized medical concepts, producing subject cards with certainty scores and evidence from source IDs. Clinician approval is required before facts enter an append-only, ontology-grounded fact graph, enabling traceable clinical summaries over time. The team is also working on adding OCR and voice transcription for doctor notes, guided by feedback and dataset support from Memorial Center in India."}