Developing Innovative Systems Through Responsible AI In Healthcare
summary
An AI system predicts surgical complication risk for hip and knee patients using Azure Machine Learning trained on a Northumbria dataset with 220 features, built as an MVP then improved with explainable models and responsible AI dashboards. A retrospective audit shows fewer than 1 in 20 patients have serious complications and reveals bias against socially deprived postcodes, so the pipeline enforces transparency, clinician human-in-the-loop decisions, patient opt-out, and plans for MLops and MHRA software-as-a-medical-device registration.
AI-written summary · about
chapters (11)
- 0:00Introduction to panel
- 0:42Patient risk prediction in surgery
- 3:11Project overview and AI in healthcare
- 4:57Team roles and project setup
- 7:25Why AI for surgical risk prediction
- 10:37Turning vision into reality
- 13:28Addressing bias in healthcare AI
- 15:54Ensuring transparency and human agency
- 18:40Model development insights
- 20:49Key learnings and challenges
- 23:17Closing and next steps
related talks
38mResponsible AI In Practice
20mResponsible AI in Health
36mResponsible AI
23mEngineering AI-Powered Predictive Healthcare Systems
22mLeveraging AI/ML for Transformative Software Solutions in Healthcare
06mTransitioning AI from the lab to real-life applications
11mDesigning Governed AI Pipelines for Healthcare Impact
13mThe Role of AI in Transforming Dialysis Healthcare