Jan 2, 2019 · 29m · a16z

a16z Podcast | Health Data -- A Feedback Loop for Humanity

Jeff Kaditz · 19m spoken Vijay Pande · 4m spoken Sonal Chokshi · 3m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this episode of the a16z podcast, host Sonal Chokshi, Vijay Pande, and Q co-founder Jeff Kaditz discuss transforming healthcare from reactive, single-point medical care into a continuous, quantitative time-series science. By leveraging longitudinal multi-omics tracking, data science, and value-based care models, they outline a framework for proactive disease prevention and personalized longevity.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The host holds 13.9% of the talking time here. How this is scored →

The host as informed peer 3.8 Guest teaching 4.8 Guest disagreement 1.3 The host pushing back 1.5
05100:0010:0020:002:10–5:24 · The host as informed peer 2/10 Diagnostics as Predictive Time-Series Models The host asks basic introductory and clarifying questions regarding the definition of diagnostics, PSA, and time series. The guest reframes diagnostics from static measurements to predictive models, educating the host on sensitivity and specificity limitations.5:24–9:16 · The host as informed peer 3/10 Healthcare Economics: Fee-for-Service vs. Value-Based Care The host asks organizational and prompting questions to move the topic along. The guest and co-host explain the transition from fee-for-service to value-based care models.9:16–13:28 · The host as informed peer 5/10 Managing False Positives with Data Science The host brings up her psychology background regarding false positives and cites the heated public debate on mammogram screening guidelines. The guest expands from a data science perspective on why single variables shouldn't be discarded.13:28–18:04 · The host as informed peer 5/10 Ethics, Patient Rights, and Primary Care Models The host connects the discussion to developmental psychology and twin longitudinal studies. The guest critique of AMA paternalism regarding patient access to data introduces mild ideological tension.18:04–22:51 · The host as informed peer 2/10 Stethoscopes vs. Modern Multi-Omics Precision The host acts mostly as a listener, reacting with surprise at the simplicity of medical physicals. The guest educates the room on physiological state information content compared to the static genome.22:51–29:52 · The host as informed peer 6/10 Multivariate Baselines and Data Donation for Humanity The host contributes a relevant domain parallel using Landsat satellite raw sensor data storage and challenges the guest on potential sample size selection bias for data donation.2:10–5:24 · Guest teaching 6/10 Diagnostics as Predictive Time-Series Models The host asks basic introductory and clarifying questions regarding the definition of diagnostics, PSA, and time series. The guest reframes diagnostics from static measurements to predictive models, educating the host on sensitivity and specificity limitations.5:24–9:16 · Guest teaching 5/10 Healthcare Economics: Fee-for-Service vs. Value-Based Care The host asks organizational and prompting questions to move the topic along. The guest and co-host explain the transition from fee-for-service to value-based care models.9:16–13:28 · Guest teaching 4/10 Managing False Positives with Data Science The host brings up her psychology background regarding false positives and cites the heated public debate on mammogram screening guidelines. The guest expands from a data science perspective on why single variables shouldn't be discarded.13:28–18:04 · Guest teaching 4/10 Ethics, Patient Rights, and Primary Care Models The host connects the discussion to developmental psychology and twin longitudinal studies. The guest critique of AMA paternalism regarding patient access to data introduces mild ideological tension.18:04–22:51 · Guest teaching 6/10 Stethoscopes vs. Modern Multi-Omics Precision The host acts mostly as a listener, reacting with surprise at the simplicity of medical physicals. The guest educates the room on physiological state information content compared to the static genome.22:51–29:52 · Guest teaching 4/10 Multivariate Baselines and Data Donation for Humanity The host contributes a relevant domain parallel using Landsat satellite raw sensor data storage and challenges the guest on potential sample size selection bias for data donation.2:10–5:24 · Guest disagreement 1/10 Diagnostics as Predictive Time-Series Models The host asks basic introductory and clarifying questions regarding the definition of diagnostics, PSA, and time series. The guest reframes diagnostics from static measurements to predictive models, educating the host on sensitivity and specificity limitations.5:24–9:16 · Guest disagreement 1/10 Healthcare Economics: Fee-for-Service vs. Value-Based Care The host asks organizational and prompting questions to move the topic along. The guest and co-host explain the transition from fee-for-service to value-based care models.9:16–13:28 · Guest disagreement 1/10 Managing False Positives with Data Science The host brings up her psychology background regarding false positives and cites the heated public debate on mammogram screening guidelines. The guest expands from a data science perspective on why single variables shouldn't be discarded.13:28–18:04 · Guest disagreement 3/10 Ethics, Patient Rights, and Primary Care Models The host connects the discussion to developmental psychology and twin longitudinal studies. The guest critique of AMA paternalism regarding patient access to data introduces mild ideological tension.18:04–22:51 · Guest disagreement 1/10 Stethoscopes vs. Modern Multi-Omics Precision The host acts mostly as a listener, reacting with surprise at the simplicity of medical physicals. The guest educates the room on physiological state information content compared to the static genome.22:51–29:52 · Guest disagreement 1/10 Multivariate Baselines and Data Donation for Humanity The host contributes a relevant domain parallel using Landsat satellite raw sensor data storage and challenges the guest on potential sample size selection bias for data donation.2:10–5:24 · The host pushing back 1/10 Diagnostics as Predictive Time-Series Models The host asks basic introductory and clarifying questions regarding the definition of diagnostics, PSA, and time series. The guest reframes diagnostics from static measurements to predictive models, educating the host on sensitivity and specificity limitations.5:24–9:16 · The host pushing back 1/10 Healthcare Economics: Fee-for-Service vs. Value-Based Care The host asks organizational and prompting questions to move the topic along. The guest and co-host explain the transition from fee-for-service to value-based care models.9:16–13:28 · The host pushing back 2/10 Managing False Positives with Data Science The host brings up her psychology background regarding false positives and cites the heated public debate on mammogram screening guidelines. The guest expands from a data science perspective on why single variables shouldn't be discarded.13:28–18:04 · The host pushing back 1/10 Ethics, Patient Rights, and Primary Care Models The host connects the discussion to developmental psychology and twin longitudinal studies. The guest critique of AMA paternalism regarding patient access to data introduces mild ideological tension.18:04–22:51 · The host pushing back 1/10 Stethoscopes vs. Modern Multi-Omics Precision The host acts mostly as a listener, reacting with surprise at the simplicity of medical physicals. The guest educates the room on physiological state information content compared to the static genome.22:51–29:52 · The host pushing back 3/10 Multivariate Baselines and Data Donation for Humanity The host contributes a relevant domain parallel using Landsat satellite raw sensor data storage and challenges the guest on potential sample size selection bias for data donation.

speaking balance: gold is the host, purple is the guest (3 minute bins)

0:00 · the host 22.6% · guest 77.4%0:00 · the host 22.6% · guest 77.4%3:00 · the host 2.1% · guest 97.9%3:00 · the host 2.1% · guest 97.9%6:00 · the host 6.8% · guest 93.2%6:00 · the host 6.8% · guest 93.2%9:00 · the host 11.3% · guest 88.7%9:00 · the host 11.3% · guest 88.7%12:00 · the host 10.5% · guest 89.5%12:00 · the host 10.5% · guest 89.5%15:00 · the host 2.4% · guest 97.6%15:00 · the host 2.4% · guest 97.6%18:00 · the host 19.6% · guest 80.4%18:00 · the host 19.6% · guest 80.4%21:00 · the host 5.3% · guest 94.7%21:00 · the host 5.3% · guest 94.7%24:00 · the host 24.4% · guest 75.6%24:00 · the host 24.4% · guest 75.6%27:00 · the host 35% · guest 65%27:00 · the host 35% · guest 65%
Sharpest disagreement ▶ 16:53 Calling out medical establishment paternalism

Jeff forcefully rejects the traditional medical model and AMA stance, arguing it is hypocritical to deny citizens access to their own biological data while allowing unhealthy behaviors like smoking.

Hardest push from the host ▶ 28:42 Challenging sample size validity in data donation

Sonal presses Jeff on how a small, self-selected sample of data donors can yield statistically valid population insights.

Biggest teaching moment ▶ 2:41 Reframing diagnostics as predictive models

Jeff corrects Sonal's basic definition of diagnostics, explaining that medical tests are actually predictive algorithms whose utility changes when evaluated across longitudinal time series.

The host holds their own ▶ 27:21 Drawing Landsat satellite data parallel

Sonal demonstrates expert synthesis by introducing the Landsat project's practice of storing raw sensor readings to be retroactively analyzed as technological capability improves.

the scores for every segment, with the reasoning behind each
ChapterTopicThe host as informed peerGuest teachingGuest disagreementThe host pushing backWhy
Diagnostics as Predictive Time-Series Models 2611 The host asks basic introductory and clarifying questions regarding the definition of diagnostics, PSA, and time series. The guest reframes diagnostics from static measurements to predictive models, educating the host on sensitivity and specificity limitations.
Healthcare Economics: Fee-for-Service vs. Value-Based Care 3511 The host asks organizational and prompting questions to move the topic along. The guest and co-host explain the transition from fee-for-service to value-based care models.
Managing False Positives with Data Science 5412 The host brings up her psychology background regarding false positives and cites the heated public debate on mammogram screening guidelines. The guest expands from a data science perspective on why single variables shouldn't be discarded.
Ethics, Patient Rights, and Primary Care Models 5431 The host connects the discussion to developmental psychology and twin longitudinal studies. The guest critique of AMA paternalism regarding patient access to data introduces mild ideological tension.
Stethoscopes vs. Modern Multi-Omics Precision 2611 The host acts mostly as a listener, reacting with surprise at the simplicity of medical physicals. The guest educates the room on physiological state information content compared to the static genome.
Multivariate Baselines and Data Donation for Humanity 6413 The host contributes a relevant domain parallel using Landsat satellite raw sensor data storage and challenges the guest on potential sample size selection bias for data donation.

Statements from this episode (14)

Assertion Supported
Kaditz: Annual physicals show no correlation with better patient health outcomes
“If you think about what the annual physical consists of, and actually if you ask most physicians and most research, there is actually no real correlation between outcomes and consistently doing annual physicals.”
Jeff Kaditz Jan 2, 2019 ▶ 0:41
Assertion Partly supported
Kaditz: Dental care costs are flat or falling while oral health quality improves
“By most metrics, even adjusted for inflation, the cost of dental care is staying flat or going down. But at the same time, the quality of our dental health is going up”
Jeff Kaditz Jan 2, 2019 ▶ 1:24
Insight
Kaditz: Longitudinal physiological data contains more information than genetics alone
“Even if you're a twin, even if two people have the same genome, they actually won't necessarily express the same phenotypes, which to us means that there's more information encoded in the evolution of your physiological state and the trajectories in your physi…”
Jeff Kaditz Jan 2, 2019 ▶ 4:33
Insight
Kaditz: Genetic code alone cannot be the silver bullet for personalized medicine
“Human physiology is a long tail distribution. So your genetic code cannot be the kind of, you know, silver bullet that is magically going to bring us to the age of personalized medicine.”
Jeff Kaditz Jan 2, 2019 ▶ 4:53
Insight
Kaditz: Value-based healthcare requires continuous monitoring to define success metrics
“If you're going to have a value-based care model, if you're not monitoring the population continuously and regularly, you really don't have a way to know if, you know, somebody is getting healthier or not. So unless we have a way to quantify what it means to g…”
Jeff Kaditz Jan 2, 2019 ▶ 7:17
Assertion Supported
Kaditz: Most lethal diseases reach advanced stages before symptoms appear
“Most of the most lethal diseases, by the time you're symptomatic, you're in an advanced stage.”
Jeff Kaditz Jan 2, 2019 ▶ 8:27
Insight
Pande: Low-utility diagnostic tests become valuable when tracked as longitudinal data
“If you think of this test from a one off point of view, it may look less useful. But if you think of it from a longitudinal point of view or longitudinal in the context of many other tests, suddenly now this is actually potentially useful information.”
Vijay Pande Jan 2, 2019 ▶ 12:01
Insight
Kaditz: Data scientists would never discard 70-percent accurate diagnostics like mammograms
“If mammograms have a predictive power information value in terms of defining or predicting whether or not somebody has breast cancer with the 70% accuracy, no data scientist in the world would throw out a variable that gave you That amount of information. They…”
Jeff Kaditz Jan 2, 2019 ▶ 12:45
Assertion Not checkable as stated
Kaditz: Discarding raw medical data prevents the building of predictive models
“Like right now in medicine, a lot of times we store the result of a diagnostic, but we don't keep the measurement, which actually prohibits us from going back and building models that, you know, in reusing the measurement.”
Jeff Kaditz Jan 2, 2019 ▶ 14:17
Assertion Supported
Kaditz: Dentists often provide the first detection of asymptomatic heart disease
“There are papers and research studies that show that there's a lot of diseases, you know, that somebody can be, you know, have no symptoms for like cardiac disease that dentists get the first look at cause it's correlated to gum disease.”
Jeff Kaditz Jan 2, 2019 ▶ 15:50
Opinion
Kaditz: Patients should own and control all of their personal health data
“People should own and control information about their own bodies and have a right to whatever information they want about it, especially if it's, can be non-invasively gathered and, you know, they're willing to pay for it.”
Jeff Kaditz Jan 2, 2019 ▶ 17:47
Assertion Not checkable as stated
Kaditz: Physiological states contain a million trillion times more information than genomes
“The best estimates we have for a complete representation of your physiological state at a point in time Is that it's a million trillion times more information.”
Jeff Kaditz Jan 2, 2019 ▶ 20:20
Insight
Kaditz: Tracking individual baselines is more sensitive than population-wide clinical thresholds
“Doing a clinical study and coming up with an absolute threshold is inherently, we know from a lot of other sciences, less sensitive than tracking deltas in the same system.”
Jeff Kaditz Jan 2, 2019 ▶ 23:16
Insight
Kaditz: Donating personal health data impacts humanity far more than organ donation
“If I donate an organ, I can save one person's life. If I donate my data, I can help save the lives of every person who ever is born ever after me.”
Jeff Kaditz Jan 2, 2019 ▶ 28:28
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 1,000 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.