WeiYa's Work Yard

A dog, who fell into the ocean of statistics, tries to write down his ideas and notes to save himself.

Medicine Meets AI

Posted on
Tags: Deep Learning, Radiology

Last two days, I attended the conference Medicine Meets AI 2019: East Meets West, which help me know more AI from the industrial and medical perspective.

The first day is about AI in radiology, which is a completely new word for me.

What is AI and Why is Medical Imaging at the Center of it?

In this talk, Michael D. KUO shared his experience with designing and running prospective AI clinical trials at MAIL@HKU.

The comparison between Machine and Human learning impressed me,

where I also learn the Dunning-Kruger Effect,

邓宁-克鲁格效应(英语:Dunning-Kruger effect),简称邓克效应或达克效应(DK effect),是一种认知偏差,能力欠缺的人有一种虚幻的自我优越感,错误地认为自己比真实情况更加优秀。

Testing and Validation of Medical AI

George SHIH shared the quote from the CEO of IBM to explain why everyone is talking about AI.

AI has the potential to find solutions to the world’s most unsolvable problems.

Ginni Rometty, IBM CEO

He emphasized that ML Validation is NOT equal to Clinical Validation, and gave two validation examples.

Validation 1

Validation 2 (Mixed Training)

Finally, he mentioned that we may need to “overfit” to perform well clinically.

It reminds me some similar idea in Stein’s paradox.

The original slides can be found here.

Tong Zhang’s Talk

AI Era

Modern AI Technology

AI Industrial Development

AI Hardware for Future AI Innovations


Published in categories Review