Why Healthcare AI Fails Without Context

Keshavan Seshadri with host Chris Hutchins · 28 minutes

About this video

Keshavan Seshadri joins Chris Hutchins to explain why general-purpose intelligence is unsafe in healthcare without patient, task, operational, and institutional context. They examine confidence scoring, explainability, bias detection, guardrails, and when an AI system must defer to a human.

Topics covered

  • The four types of context a healthcare AI system needs
  • Why confidence scores and risk determine human deferral
  • How explainability, logging, and tool tracing support accountability
  • Why regulated AI needs continuous evaluation and guardrails

This conversation is educational and does not constitute medical, legal, or regulatory advice.

Keshavan Seshadri

Featured guest

Keshavan Seshadri

Senior ML Engineer

Machine-learning engineer and AI platform builder focused on context-aware systems, explainability, evaluation, and human-in-the-loop design.

LinkedIn Profile

More conversations like this

Stay in the signal

Subscribe for new conversations on healthcare AI, leadership, governance, and trust.

Subscribe on YouTube

Help others find the show

Loved this episode?

A 30-second Apple Podcasts review helps more healthcare AI leaders discover The Signal Room.

Leave a Review