OncoBench NHS: Real World Oncology Outcomes

Client: Simon Wan swym010@gmail.com

Clinical benchmarking using electronic health records is currently fragmented across systems and slowed by lagging public datasets. To address this, your task is to build a trade-show MVP for an on-premises oncology outcomes tool running on a static JSON file of synthetic patient histories. The core deliverable features interactive Kaplan-Meier curves and life tables, complemented by filters for comorbidity stratification, treatment patterns, and SACT data overlays. Designed for time-poor clinicians, the interface enables robust analytics suitable for coffee-break review, conferences, or grand rounds. Long-term commercial success relies heavily on strong data security and seamless interoperability across different EMR providers like Epic and Cerner. Future expansions could introduce local edge LLMs via WebAssembly or containers to parse unstructured clinic letters into structured JSON. Additional planned enhancements include a SMART on FHIR sandbox, federated analytics for cross-client benchmarks, and client-side covariate adjustment for balanced cohort modeling.