Send us a text

In this episode, I’m joined by Dr. Hamid Tizhoosh, professor of biomedical informatics at the Mayo Clinic, to unravel what’s truly holding back AI in healthcare, especially pathology.

From the myths of general-purpose foundation models to the missing link of data availability, this conversation explores the technical and ethical realities of deploying AI that’s accurate, consistent, lean, fast, and robust.

📌 Topics We Cover

[00:01:00] The five essential qualities AI must meet to be usable
[00:04:00] Why foundation models often fail in histopathology
[00:08:00] What “graceful failure” looks like in AI for diagnostics
[00:13:00] The problem with data silos and missing clinical records
[00:22:00] Why specialization in AI models is non-negotiable
[00:34:00] The role of Retrieval Augmented Generation (RAG)
[00:43:00] How transformer models broke away from brain mimicry
[00:50:00] Academic dishonesty, publication pressure & bias
[01:04:00] Decentralized AI and why it won’t solve big problems
[01:12:00] Data diversity, disparity, and the realities of healthcare bias

🔍 If you’ve ever wondered why AI tools stall in real-world pathology labs, this episode breaks it down with honesty, clarity, and vision.

THIS EPISODE’S RESOURCES:

#DigitalPathology #AIinMedicine #ClinicalAI #PathologyInnovation #BiasInAI

Support the show

Become a Digital Pathology Trailblazer get the "Digital Pathology 101" FREE E-book and join us!

134: AI Trust Issues, Challenges, and Multimodal Insights in Pathology with Hamid R. Tizhoosh, PhD 01:39:23 Share

Digital Pathology Podcast

Shownotes Transcript

134: AI Trust Issues, Challenges, and Multimodal Insights in Pathology with Hamid R. Tizhoosh, PhD