The MaML Podcast - Medicine & Machine Learning

Shanen Boettcher is a former general manager at Microsoft, product manager at Netscape, and now a PhD student at the University of Saint Andrews currently studying AI Ethics and Spirituality.

Shanen was recently featured in a New York Times Article titled, "Can Silicon Valley Find God?" He is a pioneer in this field and deftly explores two very disparate topics to deeper probe the pressing questions of our generation. 

In this interview we discuss how artificial intelligence can facilitate positive conversations about faith, the impact of faith/spirituality on health, as well as other interesting topics like the study of world religions and spiritual texts. We close with some advice for individuals looking to get involved in this sort of work. 

1:20 Introduction and Background

2:55 Why study world religions?

5:50 Looking back on the days at Microsoft

7:20 The research that led to a New York Times article

11:55 Should AI expose people to their existing religious beliefs or provide new perspectives from other religions.

17:00 Where does this research go from here?

20:10 Do interactions with AI have an impact on people’s religious beliefs?

27:18 Use of algorithms for answering existential questions

30:00 How can this research help people?

37:10 Preventing religious bias in AI systems

40:20 Will technology bring us closer or further from our spirituality?

49:55 What should AI say when you ask existential questions?

52:55 Does the source of the voice affect an individual's interpretation of the answer?

1:02:50 Advice for those in their 20’s

1:07:30 Do we need religion going forward?

1:13:45 A favorite spiritual text

Dr. James Zou of Stanford University is an inaugural Chan-Zuckerberg investigator and faculty director for the university-wide AI for Health program.  

Dr. Zou recently published a paper in Nature which is making waves in the clinical trial world because it is causing us to rethink how we set eligibility criteria for clinical trials. Using an ML approach, he shows that by changing such criteria, we can make trials both more inclusive, opening them up to way more patients, while at the same time safeguarding patient safety. 

We also talk about his various other research projects, which span the gamut from evaluating FDA approvals of AI algorithms, all the way to deeper mathematical concepts like data valuation. Dr Zou is an impressive titan in the AI and medicine  space. In this interview I really came to appreciate how broad his research spans, which I think is key to his many successful projects. We ultimately close with some good advice for people looking to get involved in this exciting and growing space. 

02:45 Introduction to the intersection of Medicine and AI

4:20 Life after Ph.D.

6:35 New Nature paper on AI and clinical trials

13:25 How did we approach this question?

14:15 Data Driven Approach - Trial Path Finder

16:59 The ethical implications of this approach

19:35 Why are minority populations excluded from research?

20:25 Using AI to include ineligible patients in clinical trials

23:50 Future for this project

27:00 Evaluation of FDA approvals for AI algorithms

32:23 Favorite Project Dr. Zou has worked on

35:01 Dr. Zou's favorite math concept in the machine learning space

39:10 Separating signal from noise

39:53 Dream research projects

41:00 Future of Ai % medicine in 10-20 years

43:15 The human and AI team

47:40 What advice would you give to your 20 year old self

Interviewer: David Wu

Producer: Aaron Schumacher & Alexander Jacobs

Art: Melanie Bussan

Follow us on Twitter: 

https://twitter.com/TheMaMLPodcast?s=20

In this episode we discuss a novel idea in the healthcare and AI space: using Swarm Learning and blockchain technology for decentralized and confidential machine learning on clinical data. This promising new framework for collaborative research improves both algorithm performance and preserves patient privacy. 

This idea has been pioneered by Dr. Joachim Schultze, who recently published an exciting new paper on the subject in Nature. Dr. Joachim Schultze is a professor of Genomics and Immunoregulation at the DZNE in Germany and the University of Bonn.

2:30 Introduction and Background

6:00 Studying Broadly as an Academic

9:10 Joachim Schultze's introduction to A.I. through work on Leukemia

14:45 Recent Nature Paper - Swarm Learning & The Blockchain

21:10 Federated Learning vs. Swarm Learning

23:00 Using the Blockchain and Smart Contracts to Secure Data Sets

25:00 External Threats to the Swarm

29:40 Reaching Agreement Before Inter-Institutional Swarm Learning

35:50 Utilizing Multiple Nodes to Answer a Clinical Question

39:18 Reducing Technology-Driven Noise and Decreasing Bias With the Swarm

44:50 Open Science, Open Insights, But is Open Data Absolutely Necessary?

46:46 The Necessity of an Interprofessional Team to Complete This Project

48:30 Next Steps For This Project

54:10 Central Maintenance For The Swarm

55:10 Future of A.I. in Medicine

59:40 What Advice Would You Give To Your 20-Year Old Self

Interviewer: David Wu

Producer: Aaron Schumacher & Alexander Jacobs

Art: Saurin Kantesaria @saorange314 - Instagram

Professor Glenn Cohen is a James A. Attwood and Leslie Williams Professor of Law at Harvard University. Professor Cohen is one of the world’s leading experts on the intersection of bioethics and the law and is the author of more than 150 articles appearing in such places as New England Journal of Medicine, JAMA, The American Journal of Bioethics, The New York Times, and The Washington Post. He also leads the Project on Precision Medicine, Artificial Intelligence, and the Law, which is part of the larger Centre for Advanced Studies in Biomedical Innovation Law.

In this interview, we discuss a variety of legal and ethical topics like data privacy, liability and medical errors, and AI use disclosure in patient settings. Professor Cohen provides many examples of how AI is changing the face of our society from driverless cars to Target knowing us better than our own family members! He also makes a few great literature and media recommendations: "Exhalation" by Ted Chiang, "The Paper Menagerie" by Ken Liu, "The Three-Body Problem" by Liu Cixin, and of course, the Netflix original, "Black Mirror."

P.S. Follow professor Cohen on Twitter (@CohenProf) for more nuggets of wisdom on legal and ethical issues in artificial intelligence (and in many other healthcare sectors)!

1:30 Professor Cohen's Journey

3:17 Project on Precision Medicine (PMAIL)

5:46 "Case-based" approach

8:57 Who takes the blame?

11:20 Driverless cars and healthcare

12:33 Medical errors

13:08 Big data, HIPPA

16:30 Where are we going?

18:40 Bias in AI + Healthcare

20:00 Advice to your past self!

22:30 Vital interprofessional collaboration

Interviewer: Madeline Ahern

Producer: Melanie Bussan

Art: Saurin Kantesaria @saorange314 - Instagram

Dr. Vivian Lee, MD, Ph.D., MBA is currently President of Health Platforms at Verily, an Alphabet Company. Dr. Vivian Lee is also the author of the latest book “The Long Fix,” a book about solving America’s healthcare crisis.

Dr. Lee has accomplished much in her diverse career. She received a doctorate in medical engineering from Oxford as a Rhodes scholar, her MD from Harvard Medical School, was valedictorian at NYU Stern School of Business, authored over 200 peer-reviewed research publications, as well as a cardiovascular MRI textbook, former CEO of the University of Utah Health and dean of their medical school and, is now the President of health platforms at Verily Life sciences, an Alphabet company.

In this interview, we talk about her journey to Verily today and her thoughts on how healthcare has been changed for the better by new technologies like Digital Health Platforms, an example being Onduo for blood glucose management in diabetics.

We also talk about how medicine has changed from the days she started medical school to the future landscape that current medical students face today, one that is much more integrated with payers, tech, politics, and employers. We hope that this interview inspires you as it did to us to try and tackle all of healthcare’s problems with renewed vigor. Thank you and enjoy!

P.S. Please check out Dr. Vivian Lee’s latest book “The Long Fix” and review it on Amazon/GoodReads!!

2:50 Dr. Vivian Lee’s journey

8:20 Transition to Radiology

12:40 Transition to Univ. of Utah

15:50 “What does your job at Verily entail?”

17:10 Onduo - an example of Health Platforms in action

23:50 Verily and COVID testing

27:40 “The Long Fix” and the Co-Production of Health

37:00 MedSchool now vs. MedSchool then

44:00 Verily and how it affects the future of medicine

46:00 David’s misattributed Luddite fears

50:00 What advice would you give your younger self?

Interviewer: David Wu @davidjhwu - Twitter

Producer: Aaron Schumacher @a_schu95 - Twitter

Art: Saurin Kantesaria @@saorange314 - Instagram

Dr. Faisal Mahmood is an Assistant Professor of Pathology at Harvard Medical School and Computational Pathology at Brigham and Women’s Hospital. Dr. Mahmood recently published an exciting new paper this year where he and his team built a deep learning model to accurately identify tumors of unknown origin on pathological slides (Lu et al., Nature 2021)

Pathology is one of the central pillars of medicine and here we really dive deep into how machine learning is pushing the boundaries of the field and our abilities to diagnose and recognize tumors. Enjoy!

Twitter: @TheMaMLPodcast

Interviewer: David JH Wu (@davidjhwu)

Producer: Aaron Schumacher (@a_schu95)

Cover Art: Saurin Kantesaria

1:20 Background in computational pathology

5:30 Interest in Pathology

10:20 Modern algorithms detecting biomarkers to better educate physicians

12:05 Using AI to identify tumors of unknown origin

17:10 Building the TOAD AI Model

21:50 Assessing the validity of the Toad Model

23:30 Determining inputs for the TOAD Model

26:30 Diversity with the TOAD Algorithm

27:40 Next steps for the project

33:15 Using AI to augment physicians' abilities

35:06 Advice for physicians interest in AI

37:00 Dream Research Project

38:40 Will AI make medical discoveries in the future?

40:38 Advice you would give to yourself in your 20's

41:55 Obtaining a Ph.D. in Japan

44:00 Closing thoughts

Paper: Lu et. al, “AI-based pathology predicts origins for cancers of unknown primary.” Nature, 2021

Dr. Nneka Comfere is a Dermatologist and Dermopathologist at the Mayo Clinic in Rochester, MN. We discuss Dr. Comfere's discovery of visual beauty within dermatology and how this can be applicable in a machine learning setting. We also talk about the possible uses for dermatoscopes and artificial intelligence to fill gaps in care based on location. Dr. Comfere's take on AI from a clinician's perspective is accessible to not only medical professionals, but also those seeking to learn more about how machines are becoming part of the healthcare system. Enjoy!

Interviewer: Maddie Ahern

0:25 - Journey to Dermatology and Dermopathology, Integration of AI

9:35 - Articles in Journal the American Academy of Dermatology

19:25 - Initial Venture into AI, Building a Dermatological Database

28:32 - Future of AI in Medicine

32:15 - What is Next for Dr. Comfere?

36:51 - Advice for Students/Learners

Ian Pan, MD, is a Kaggle Grandmaster, radiologist resident at the Brigham and Women’s Hospital, and a rising star in the medicine and AI field.

Kaggle competitions are international data science competitions that are both very competitive and prestigious. We talk about Ian's path to medicine and AI as well as the various strategies he’s used to become one of the top coders globally in this burgeoning new field. Ian also gives some great advice on how to get started and we close with some of his exhortations against poor practices in ML today.

This interview was a lot of fun and if you are curious about Kaggle competitions or how to be the best at them, this interview is for you.

Time-Stamps

Anouk Stein, MD, is a radiologist and AI Data Specialist at MD.ai, a healthcare start-up based in NY. We discuss Dr. Stein's journey to MD.ai as well as her current work in the medical AI space. Dr. Stein provides some great advice for anyone looking to get started in practical machine learning. We also talk about some of the exciting kaggle competitions held by MD.ai as well as the importance of external data validation. We close with some great advice for our listeners from Dr. Stein on how to embrace the exciting new changes taking place in medicine today. Dr. Stein is a terrific teacher and I learned a lot from her in this interview. I hope you all enjoy!

Timestamps 

1:10 Individual path to healthcare and AI 

3:58 What does MD.ai do? 

5:00 Stanford Design-Your-Life course 

7:40 AI and Radiologists 

12:00 Combining algorithms and the necessity of a meta-algorithm 

14:48 External Validation of Data  

16:50 Practical Machine Learning 

22:10 What is external validation  

25:10 Controversy over generalization  

29.10 Accomplishments of MD.ai 

32:30 What is it like to work in industry after medical education  

37:45 How MD.AI got started  

40:25 Fields of Medicine looking towards AI 

43:40 The future of AI in medicine over the next 10 - 20 years  

46:25 What advice would you give to yourself in your twenties  

47:00 Any advice for young physicians

Resources Mentioned: