The Gift From Yesterday Matthew Lungren, MD MPH Assistant - - PowerPoint PPT Presentation

the gift from yesterday
SMART_READER_LITE
LIVE PREVIEW

The Gift From Yesterday Matthew Lungren, MD MPH Assistant - - PowerPoint PPT Presentation

The Gift From Yesterday Matthew Lungren, MD MPH Assistant Professor, Radiology, Stanford University School of Medicine #FDAPDSsymp | #AIinHealth COMPUTER SCIENTIST TASKS ARE SIMILAR TO DOCTOR TASKS No acute intracranial Saggital knee


slide-1
SLIDE 1

The Gift From Yesterday

Matthew Lungren, MD MPH Assistant Professor, Radiology, Stanford University School of Medicine

#FDAPDSsymp | #AIinHealth

slide-2
SLIDE 2

COMPUTER SCIENTIST TASKS ARE SIMILAR TO DOCTOR TASKS

slide-3
SLIDE 3
slide-4
SLIDE 4
slide-5
SLIDE 5
slide-6
SLIDE 6
slide-7
SLIDE 7

“No acute airspace disease” “No acute intracranial abnormality” “Saggital knee MRI with acute ACL tear” “Heterogenously dense. BIRADS 2” “PET CT metabolically active GE junction mass” “Mild cerebellar atrophy”

slide-8
SLIDE 8

OPEN SOURCE DATA SETS ARE THE PRIMARY CATALYST FOR MACHINE LEARNING ADVANCEMENTS

slide-9
SLIDE 9
slide-10
SLIDE 10

Negascout Planning Algorithm Developed 1983 1991 1997 The Extended Book of Chess Games Dataset Released Deep Blue Defeats Kasparov 14 years 6 years

slide-11
SLIDE 11
slide-12
SLIDE 12

Mixture of Experts Algorithm Developed 1991 2009 2011 8.6 Million documents from Wikipedia and Project Gutenberg Released to Public Watson becomes world Jeopardy Champion 20 years 2 years

slide-13
SLIDE 13
slide-14
SLIDE 14

Convolutional Neural Network Created 1989 2009 2015 1.5 Million classified images with 1,000 object categories released to public Microsoft Research surpasses human recognition performance 26 years 6 years

slide-15
SLIDE 15

MEDICAL DATA ARE A PUBLIC GOOD AND SHOULD BE OPEN SOURCED TO BENEFIT ALL

slide-16
SLIDE 16
slide-17
SLIDE 17

Not divided or consumed and easily replicated “Ownership” of data is an imprecise concept

  • Rights to control access
  • Rights to share of profit

Value of data and information is relative

  • Exclusive access to data may give relative advantage
  • Advantage negated when others have access to the same data

Credit: David Larson MD MBA

DATA are not Traditional Form of Property

slide-18
SLIDE 18
slide-19
SLIDE 19

but… Medical Data are PEOPLE

slide-20
SLIDE 20

Our position: Once clinical data have been used to provide care, the primary purpose for acquiring the data is fulfilled. For secondary use: Clinical data should be treated as a form of public good used for the benefit of all thorough open source research and education NOT for sale or exclusive partner access

Credit: David Larson MD MBA

slide-21
SLIDE 21

Rajpurkar& Irvin et al., PLOS Medicine, 2018 Rajpurkar& Irvin et al., MIDL,201Bien & Rajpurkaret al., PLOS Medicine,2018Park & Chute & Rajpurkaret al., JAMA NetworkOpen. 2019

slide-22
SLIDE 22

7,200+ teams worldwide competing in medical imaging challenges hosted at Stanford Benchmark Performance with Consensus Expert Ground Truth

slide-23
SLIDE 23

Pesapane F. et al. Eur Radiol Exp. 2018

Publications on AI in Radiology

slide-24
SLIDE 24

Data user agreement and registration Manual PHI review of all images Curate and label data

Process for Open Source Medical Imaging Data

slide-25
SLIDE 25

ALTERNATIVE STRATEGIES TO OPEN SOURCE MEDICAL IMAGING DATA ARE COMPLIMENTARY NOT REPLACEMENTS

slide-26
SLIDE 26

Detecting and Simulating Artifacts in GAN Fake Images (Extended Version) Xu Zhang, Svebor Karaman, and Shih-Fu Chang

slide-27
SLIDE 27
slide-28
SLIDE 28
slide-29
SLIDE 29

Responsible ethically sound processes exist to make medical data available Publishing without sharing data is just marketing NOT science Alternative approaches to solve data availability are complimentary Open source medical data and benchmark competitions will accelerate innovation

slide-30
SLIDE 30

Matthew P Lungren MD MPH

Assistant Professor of Radiology Associate Director Center for Artificial Intelligence in Medicine & Imaging Stanford University School of Medicine mlungren@stanford.edu

@mattlungrenMD

Thank You

#FDAPDSsymp | #AIinHealth