Healthtech Pigeon What does "Explainable AI" in healthcare truly mean?
Feb 22, 2026
Hansen Tsui, product lead at Sanome focused on clinical AI for infection risk and clinician-centered design. He discusses co-design with frontline clinicians, integrating AI into EPRs to avoid extra workload, a TRL6 pilot with NHS trusts, explainability via SHAP dashboards, continuous monitoring and bias audits, and strategies to reduce alarm fatigue and improve real-world adoption.
AI Snips
Chapters
Books
Transcript
Episode notes
Early Prediction Of Hospital Acquired Infections
- Sanome predicts hospital-acquired infections three to seven days before onset.
- Hansen describes pairing that predictor with a patient explainability dashboard to show clinicians how the score was derived.
CoDesign By Shadowing Clinicians Daily
- Do co-design with clinicians by shadowing their day and aligning on what success looks like.
- Hansen stresses building empathy across the full clinical day to design tools that fit workflows rather than one-off solutions.
Build Continuous Clinician Feedback Programs
- Create continuous clinician feedback loops to overcome time-poor participation.
- Hansen explains Frontline to the Future offers practical experience, networking and education to engage junior staff who are hardest to reach.
