July 23, 2018

Podcast

D.A. Wallach: music, medicine, longevity, and disruptive technologies (EP.06)

The way that anything looks before we understand it is pretty imposing . . . when we have breakthroughs, they feel like an enormous relief, because something that seemed really complicated becomes really simple. — D.A. Wallach

by Peter Attia

Read Time 12 minutes

Recording artist, songwriter, essayist, investor, and so much more: D.A. Wallach is a true polymath. In this episode, among the highlights, D.A. provides compelling and colorful insight into how the music industry works today vs the past, liquid biopsies, how to approach healthspan, and how we can reach a “singularity” in medicine.

Subscribe on: APPLE PODCASTS | RSS | GOOGLE | OVERCAST | STITCHER

We discuss:

  • How to learn music as a kid and an adult [7:30];
  • Chester French’s early struggles and ultimate success [16:45];
  • Learning to learn, fostering curiosity in kids, and balancing creativity with structure [31:30];
  • D.A.’s musical inspirations [44:30];
  • History of the music industry, Spotify, and other disruptive technologies [50:00];
  • The past, present, and future of medicine, hospitals, and healthcare [1:05:30];
  • Investing in health [1:16:30];
  • What D.A. is most excited about in the future of medicine [1:22:00];
  • Liquid biopsies, how to make sense of the morass of sensitivity, specificity, positive predictive value, negative predictive value, true negatives, false positives, false negatives, and true positives in cancer screening…and the Swiss cheese metaphor [1:33:00];
  • The immune system, inflammation, and allergies [2:05:45]; and
  • More.

§

Show Notes

How to learn music as a kid and an adult [7:30]

  • Best way to learn music is to first play intuitively and then study the theory
  • “Singing is just musical speaking”, learning to control the pitch

Chester French’s early struggles and ultimate success [16:45]

Learning to learn, fostering curiosity in kids, and balancing creativity with structure [31:30]

  • Fostering curiosity and creativity is kids is about feeding it and not killing it;
    • Allow kids to ask ‘why’ questions
    • The superpower is having both the ability to practice deliberately while maintaining an exploratory and creative environment
    • The secret to parenting is figuring out how much emphasis to put on each … if pushed too hard it will dampen creativity
  • Herbie Hancock once told Pharrell Williams that learning technical skills won’t kill creativity, ‘It’s just gonna give you more colors to paint with’

D.A.’s musical inspirations [44:30]

Broad inspirations:

  1. The Beatles
  2. Bob Marley
  3. Curtis Mayfield
  4. Stevie Wonder
  5. Motown

Songwriters: Holland-Dozier-Holland (writer for Motown)

Drummers:

History of the music industry, Spotify, and other disruptive technologies [50:00]

  • D.A. got tired of touring and transitions from musician to professional investor
  • Spotify allowed D.A. to pivot into the investment side of the music business
  • Evolution of the way music got to the public for consumption
  • Economics of iTunes
  • Economics of Spotify

The past, present, and future of medicine, hospitals, and healthcare [1:05:30]

Some businesses disrupted by the internet:

  • Amazon to retail
  • Netflix to blockbuster
  • Uber to taxis
  • Travel sites to hotels

Medicine is one of the least disrupted by technology:

  • Hospitals eat all the rent
  • Payers don’t generally get paid that much
  • Providers get less and less on a per encounter basis
  • It’s a similar system to 20 years ago

Any other industries as unchanged as healthcare?

Biggest challenges, not just one problem to be solved:

  1. Quality of care: just keeping people alive vs. improving quality
  2. Cost to the system: 90% of dollars are spent by government, employer, and payer (10% to consumer)
  3. Access

We should be aiming for universal access to current best available healthcare in the world that anyone receives: ‘The future is already here, it’s just not evenly distributed yet.’

D.A. believes we can globally shift towards a large population of highly skilled nurses, fewer overtrained (expensive) doctors, and most exciting: a lot of new medicine.

Investing in health [1:16:30]

Diseases have an explanation, and we can develop interventions that have little collateral damage, so as an investor that’s the most fun place to hang out. Can be impactful and profitable.

Death and turnover may have some utility. D.A. notes the Steve Jobs commencement speech:

No one wants to die. Even people who want to go to heaven don’t want to die to get there. And yet death is the destination we all share. No one has ever escaped it. And that is as it should be, because Death is very likely the single best invention of Life. It is Life’s change agent. It clears out the old to make way for the new. Right now the new is you, but someday not too long from now, you will gradually become the old and be cleared away. Sorry to be so dramatic, but it is quite true.

— Steve Jobs

D.A. is more concerned with people suffering less as they age (i.e., healthspan).

What D.A. is most excited about in the future of medicine [1:22:00]

  • The Holy Grail for D.A. is a singularity-like moment in biomedicine. A moment where everything changes and everything accelerates our capacity to something. The moment where we can digitally represent complex biology. We can then study it at zero marginal cost.
  • D.A. says, in biology now, we want to understand how the machine of a living system works, which is incredibly complicated. The ways in which we try to represent them is incredibly complex, typically exercises in cybernetics.
  • D.A. is looking to Systems Biology: translate mechanistic connections into formal code, a language that captures how something works in a way that a computer can get, an unambiguous language, this will allow us to simulate biological systems in their full complexity.
  • Once we can do that, we can effectively run intervention-based experiments on digital systems at zero cost. That will moment we can understand biology at an extraordinary rate.
  • Most of D.A.’s investments are in things he thinks are “stepping stones” towards that singularity.

Liquid Biopsies [1:33:00]

The big three things that are going to kill most people:

  1. Atherosclerotic diseases
  2. Neurodegenerative diseases
  3. Cancer

The best blood tests might offer you 70-80%, 60-70%, and <30% predictive value for atherosclerotic diseases, neurodegenerative diseases, and cancer, respectively.

The gaping hole: we can’t do a liquid biopsy for cancer. But there are people working on it.

ONCOblot

ONCOblot is looking into protein Enox2, a protein assay, the belief was that the protein found exclusively on malignant tumors but may not have been the case.

Grail

A spinout of Illumina (who make genome sequencing machines).

  • The premise is that by sequencing peripheral blood, you can find all kinds of cellular refuse from somatic tissues, and if you are able to amplify this refuse, it will tell you about the cells it’s coming from
  • With Grail, if you sequence with enough depth, you can detect just the DNA coming from the tumor, and you get to know where it is, and what it’s doing

Glympse

Another approach is by the Boston company called Glympse, founded by Sangeeta Bhatia.

  • When cancer is in a tissue, it remodels the microenvironment
  • There are 550 endoproteases in the genome (construction workers of the gene), and Sangeeta is identifying which proteases are implicated in the early stages of diseases
  • Sangeeta wants to engineer nanoparticles to send in the body (synthetic biomarker) designed to break apart if the encounter the bad enzymes — a “swat team” — that will then signal in the urine when it finds something

Freenome

Freenome says don’t look for the tumor DNA in the blood, look at entire genome for “smoke” (i.e., the body’s systemic response) from the fire (i.e., the cancer). Try to detect the body’s response to the fire rather than look for the fire itself.

The Swiss Cheese Metaphor

An application for liquid biopsy is part of the Swiss cheese metaphor for cancer screening:

  • Mammograms are good for some things and bad for others (ignoring cost for now)
  • Good for calcified lesions but not for non-calcified
  • Conversely, MRI is the opposite of this
  • Layer mammogram, DWI MRI and liquid biopsy, and lastly the Bayesian piece, where we know her family history and risk factors, and perhaps you can improve the predictive value

Diagnostics 101—True positives, false negatives, true negatives, false positives, sensitivity, specificity, and predictive values [1:49:00]

§

Diagnostics in context

  • Sensitivity, specificity, PPV, and NPV shouldn’t be judged in isolation
  • Blindly sending a “you tested positive for cancer” letter, for example, to 500,000 random people is guaranteed to have 100% sensitivity!
    • The fact that this test is 100% sensitive is useless in light of it having 0% specificity
    • It has zero SnNout (i.e., the ability to rule out cancer with a negative test)
    • Conversely, if a sensitive (Sn) test is negative (N), rule the diagnosis out (OUT) (i.e., SnNOUT, rule it out)
  • Likewise, blindly sending a “you tested negative for cancer” letter, to 500,000 random people is guaranteed to have 100% specificity!
    • The fact that this test is 100% specific is useless in light of it having 0% sensitivity
    • It has zero SpPIN (i.e., the ability to rule in cancer with a positive test)
    • Conversely, if a specific (Sp) test is positive (P), rule the diagnosis in (IN) (i.e., SpPIN, rule it in)
  • Diagnostic measures help determine if screening tools are effective at detecting when cancer is present or not
  • These measures can be misleading when evaluated in isolation; context matters

True, False, Positive, Negative, Sensitive, Specific…

True or “False is another way of saying the test is correct or incorrect.

Positive or “Negative is another way of saying detected or not detected.

Sensitivity is the true positive rate: it represents the percentage of people with cancer who are correctly identified as such.

Specificity is the true negative rate: it represents the percentage of healthy people who are correctly identified as such.

Therefore, in the context of cancer screening:

True positive: a positive test in a person with cancer (i.e., truly detected)

False negative: a negative test in a person with cancer (i.e., missed detection)

True negative: a negative test in a person without cancer (i.e., truly not detected)

False positive: a positive test in a person without cancer (i.e., false alarm)

Sensitivity

Sensitivity is the true positive rate, or probability of cancer detection in someone we know has cancer.

To determine the sensitivity of a test, we must know whether the person truly has cancer, which means you need a sample of historical data to calculate it. Example: we take 1,000 women who truly have breast cancer, and we have each of them undergo a mammogram. We see the following results:

Positive (this is True) test: 840 women = 840 True Positives

Negative (this is False) test: 160 women = 160 False Negatives

The sensitivity of the test is 84%. It detected 84 out of 1,000 women with cancer.

Specificity

Specificity is the true negative rate, or probability that people without cancer are correctly identified as such.

To determine the specificity of a test we must know the person truly does not have cancer, which also means you need a sample of historical data to calculate it. Example: we take 1,000 women who truly do not have breast cancer, and we have each of them undergo a mammogram. We see the following results:

Positive (this is False) test: 90 women = 90 False Positives

Negative (this is True) test: 910 women = 910 True Negatives

The specificity of the test is 91%. It correctly identified 910 out of 1,000 women without cancer.

The sensitivity and specificity tradeoff

For any test, there is usually a tradeoff between avoiding false positives and false negatives. For example, in airport metal detectors looking for a gun, if the machine is extremely sensitive, individuals carrying virtually any metal will set off the detector (i.e., low false negatives). However, the more sensitive the test, the greater the risk for false alarms (i.e., high false positives), and therefore the lower the specificity. The higher the sensitivity, the lower the specificity.

The same is true in cancer screening. If an MRI can detect tumors < 1 mm in size, it’s highly sensitive. However, the test is going to be less specific. More false alarms. The opposite is true as well. If an MRI only detects tumors > 5 cm, the test will be highly specific (i.e., low false positives), but have low sensitivity (i.e., high false negatives). The higher the specificity, the lower the sensitivity.


High sensitivity test is more likely to detect cancer when cancer is truly present and less likely to miss cancer when cancer is truly present.


Low sensitivity test is less likely to detect cancer when cancer is truly present and more likely to miss cancer when cancer is truly present.


High specificity test is more likely to not detect cancer when cancer is truly not present and less likely to detect cancer when cancer is truly not present (i.e., fewer false alarms).


Low specificity test has a lower proportion of negatives that are correctly identified as such and is more likely to detect cancer when cancer is truly not present (i.e., more false alarms).


Predictive values

Positive predictive value: Positive predictive value (PPV) is the probability that subjects with a positive screening test truly have the disease.

Negative predictive value: Negative predictive value (NPV) is the probability that subjects with a negative screening test truly do not have the disease.

Here’s an example: 500,000 people are screened for thyroid cancer with a population prevalence of 0.16%. The screening tool used is ultrasound with 98% sensitivity and 54% specificity.

We can see from this example that the NPV is virtually 100%. What it appears to be showing is that if you test negative, you almost assuredly don’t have thyroid cancer. While true, this is misleading. Only 0.16% of the population tested has thyroid cancer. Because 99.84% of the people tested don’t have thyroid cancer, the NPV can’t be less than 99.84%.

We can also see that even with a high sensitivity, which tells us the true positive rate, the PPV is a remarkably low 0.3%. This means that if a person tests positive for thyroid cancer, the likelihood that he or she has thyroid cancer is about 1-in-300, or 0.3%.

Here’s a different example: 500,000 women are screened for breast cancer with a population prevalence of 11%. The screening tool used is Vigilance MRI with 92% sensitivity and 99.6% specificity.

The NPV is 99%. Because population prevalence in this example is higher (i.e., 11%), the NPV is more informative. However, even in this population, the number of false negatives is fixed to a low percentage (e.g., if Vigilance MRI had no ability to detect cancer, the NPV would be 89%).

The PPV in this example is 96.6%. The probability that a woman has cancer if she tests positive is about 97-in-100, or 97%. In contrast, the PPV for mammography screening, for example, is ~ 10.6%. However, the population prevalence of breast cancer is ~ 1% for asymptomatic women when regular mammography screening intervals are recommended, which skews the PPV lower (e.g., if the population prevalence for the Vigilance MRI example was 1%, the PPV would be ~ 70%). In this light, the Vigilance MRI example may represent a follow-up screen for a population of women who had a positive mammography.

Putting it all together

Sensitivity, specificity, PPV, and NPV shouldn’t be judged in isolation. At first blush, it may appear that a test with 100% sensitivity is the perfect test to determine if someone truly has cancer. However, imagine that this test consists of sending an identical letter to 500,000 random postal addresses and informing the head of household that he or she is positive for cancer. This test would guarantee 100% sensitivity: any person in the sample who truly has cancer is a true positive. There are zero false negatives in the sample since no letters informed the recipient that he or she is negative. Therefore, a test with 100% sensitivity can be completely useless in practice without considering specificity. The test would have 0% specificity.

The converse is true as well. Imagine the letter sent to 500,000 was instead a negative letter. It would identify 100% of the true negatives and would yield zero false positives. The test would have 100% specificity. But it would also carry with it 0% sensitivity.

Similarly, in the positive-letter example, 100% sensitivity may be misleading in terms of its ability to predict someone has cancer. If the population prevalence was 0.5%, the PPV would also be 0.5%, meaning the test is no better than, well, sending random letters to postal addresses. The same holds true for the negative-letter example.

Diagnostic measures help us determine if our screening tools are effective at detecting when cancer is present or not. Often, these measures can be misleading, especially when evaluated in isolation. It’s important to look at diagnostics in context.

§

The immune system, inflammation, and allergies [2:05:45]

  • We’re still learning a lot about cardiovascular disease around inflammation
  • Sean Parker is working on allergies, and this led him to the immune system therapies: he saw a crossover of allergies, the immune system, and cancer
  • Allergies, as mundane as they are, still can’t be addressed in a precise way

§

 

Selected Links / Related Material

§

 

People Mentioned

§

D.A. Wallach

D.A. Wallach is a recording artist, songwriter, investor, and essayist who Kanye West and Pharrell Williams discovered while he was an undergraduate at Harvard College. He has been featured in GQ, Rolling Stone, Vogue, and numerous other publications, and has toured with N*E*R*D, Lady Gaga, and Weezer. D.A. has also performed on TV Shows including Jimmy Kimmel Live and Late Night with Jimmy Fallon.

As one half of Chester French, D.A. has released three full-length albums, and has written and performed on records with Janelle Monae, Rick Ross, Diddy, and many others. His solo debut for Capitol Records, Time Machine, is available at: www.TimeMachineAlbum.com

Beyond music, D.A. invests in and advises several start-up technology companies, including SpaceX, Doctor On Demand, Ripple, Emulate and Spotify, where he was the official Artist in Residence. Forbes selected D.A. as one of its 30 Under 30 and Fast Company named him one of the 100 Most Creative People in Business. In 2015, he launched Inevitable Ventures, an investment partnership with multibillionaire Ron Burkle that supports radical entrepreneurs in areas including health care, the life sciences, and financial technologies.

In 2016, D.A. made his feature film debut in La La Land, which won a record number of Golden Globe Awards and received 14 Academy Award nominations, making it one of the 3 most nominated films in history. He frequently publishes essays on media, technology, and philosophy at dawallach.com and Medium. [dawallach.com]

DA on Facebook: D.A. Wallach

DA on Instagram: @dawallach

DA on Twitter: @dawallach

DA’s website: dawallach.com

Disclaimer: This blog is for general informational purposes only and does not constitute the practice of medicine, nursing or other professional health care services, including the giving of medical advice, and no doctor/patient relationship is formed. The use of information on this blog or materials linked from this blog is at the user's own risk. The content of this blog is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Users should not disregard, or delay in obtaining, medical advice for any medical condition they may have, and should seek the assistance of their health care professionals for any such conditions.

Comments

Read Our Comment Policy
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]
[8283]

Send this to friend

Facebook icon Twitter icon Instagram icon Pinterest icon Google+ icon YouTube icon LinkedIn icon Contact icon