The More Data You Have, The Further You Are From The Truth

Produced by @chubbyemu
0:00 Pretext and context MD Anderson Cancer Center
1:42 The Early Experiences That Shaped Dr Msaouel
3:39 The System of Cancer Research In the United States
5:25 FDA Drug Approvals and Special Designations For Oncology
11:09 A trade-off in our current system
12:35 Whoops 😞 I misspoke
14:35 Randomized trials are not supposed to be representative of a population
18:10 Inference for the treatment of Katie Coleman (metastatic oncocytoma)
25:12 It is much more valuable to refute hypotheses than to confirm them
28:05 All Models Are Wrong, But Some Are Useful
30:05 Recasting the Bias-Variance Trade-off as the Patient Relevance-Robustness Trade-off
33:38 High Relevance, Low Robustness: Subjective Bayes
34:46 High Robustness, Low Relevance: Pure Randomization
41:47 Signal To Noise Ratio
45:49 The Big Data Paradox
53:41 How the Big Data Paradox fits in with Katie Coleman's case
54:33 How the Big Data Paradox fits in with other branches of medicine
57:54 How do we rectify subjectivity in The Truth?
01:00:43 How do we address the problems introduced by the Big Data Paradox?
01:04:21 Duality in nature and a Twitter argument re: the Evidence-Based Deep Medicine Iceberg
01:07:03 Why the ideas coming from MD Anderson are coming from MD Anderson
01:09:39 Where you can find Dr Msaouel

Production Assistant: Nick Brown
Secret channel: @BigEmus

References:

Stephen Senn. Statistical Issues in Drug Development.

Causal inference and the data-fusion problem.

[prapanthebachelorette]

“All models are wrong, but some are useful.” That is a powerful statement that makes me rethink a lot of things 😮

[cadevywilliams]

I would recommend every student wishing to get a MD/PhD degree in any aspects of biology to watch this video. Super informative. At least for me, it lets me reflect on the way I design experiments and analyze data.

[lautreamontg]

Man, this discussion was so erudite and wide ranging that I was having trouble keeping up with it. And that's a good thing.

[celiatking]

I had a very strong feeling that you were talking about MD Anderson in Houston, TX! My daughter was also treated there for DIPG and they were amazing! Unfortunately she passed away there 11 yrs ago when she was 6 yrs old and this type of brain-stem cancer is still considered rare. Back in 2011, there were under 200, 000 cases worldwide with the life expectancy of no more than 6 to 18 months from diagnosis. Nowadays there have been strides where expectancy has been 3 - 5 yrs from diagnosis.
I recommend this hospital to everyone I meet!!!

[priwncess]

You have just given a young student the sign she needed to determine what she wants to do with her life! I will work on achieving an M.D. ❤

[benwlee]

I remember the days I was involved in marketing research, one of the foremost the boss would stress is to pick a good focus group. He always says if you pick focus group correctly, you do not need a large data size, hence the importance of questionnaire design, and that is deeply involved with domain knowledge.
I guess this coincides with his argument that Bayesian and random sampling are two sides of the same coin. If you dont know what you are doing, it always be garbage in, garbage out.

[Typw]

Thank you for such a cathartic video. The philosophical and medical rigor in this conversation needs to be celebrated.

[izzieb]

As someone who is currently going through (seemingly endless) investigations to determine the cause of my illness, this podcast is extremely relevant and raises important points.

It seems to me that many conditions are often missed or misdiagnosed due to being compared to an average that does not actually exist in reality. Such things should be taken in the context of the individual patient, as stated in this interview. Each patient is individual.

Furthermore, as raised here and relating to the recent video on the Chubbyemu video, it seems that some medical professionals are reluctant or unable to extrapolate on their knowledge and follow their intuition - particularly if there has been very little literature on the subject, as with the patient in the Chubbyemu video. I would imagine this leads to a lot of unnecessary hardship and treatment for patients, had they been caught earlier.

Sorry for the long comment, thank you for the great videos!

[sirunknown]

I love how excited you are talking to him. It makes me really happy to see the just bright smile on your face.

[j.d.]

Very philosophical, but interdisciplinary science is the only way to go.

[ItsAsparageese]

Listened on my podcatcher, came here to like the video, add it to my general "favorites to share" list, and commend this absolutely stellar interview. I'm a huge medicine nerd, patient education nerd, and fledgling data science nerd, and this is just some of the best material on these logic/research topics I've ever heard. I'm sending the link to several of my professors and public health friends. Thank you SO MUCH for the awesome work you do!

[ana-zbix]

This is a treasure to humankind. Thanks for all the scientists and physicians trying to give people the best and most efficient treatment for cancer, which is such a heartbreaking disease to our species.

[TheRealGamer]

Wow this entire video was very well made, thank you Dr. Bernard.

[kennysproat]

This podcast makes me really think how important personalized medicine along with pharmacogenetics is especially when the tests to put these drugs on the market are not the most representative

[lastnamefirstname]

nice long-form discussion, thanks heme review!

[Decasia]

Please keep doing these amazing interviews/discussions. And thank you!

[yengsabio]

Thank you for sharing this video! I will re-view it as many as I can. A lot of things are to be absorbed out of this interview.

Lots of love, cheers, & mabuhay, from my end--the Philippines!

[extendedblundering]

This is amazing content; seriously grateful to see such high value and informative interviews available. You are a hero for taking this time to produce intelect inspiring videos; thanks!

[anniebeanie]

I had a whole genome analysis done last year and uploaded results in multiple sites. Found out i have very high predisposition to breast cancer. But also found out i have almost every possible gene (2 alleles not just 1) for bad BPA detoxication. Obvious risk factor at play. Studying now very thoroughly what are the practical steps to avoid all bisphenols A, S, F or whatever. As is not practical to throw away anything from plastic, i am mostly focused on finding out where the risk ACTUALLY is and how to avoid it. For example no need to remove existing dental fillings but i might want to have a talk with the dentist what does she know about fillings she is using and if she could be placing fillings carefully as the risk is mainly at placing the fillings. And contact with thermal receipts is very high risk etc ....

[therabbithat]

What if the rarest cancers seem rare because they're actually the most common cancers...and the body is just really good at catching and eliminating them as a result