The Neuro Experience

Episode: How Artificial Intelligence is Transforming Healthcare
Host: Louisa Nicola & Pursuit Network
Guest: Renee Dehan, VP of Science and Artificial Intelligence, InsideTracker
Date: August 30, 2023

Main Theme & Purpose

In this episode, Louisa Nicola sits down with Renee Dehan from InsideTracker to discuss how artificial intelligence (AI) is revolutionizing healthcare – particularly through the integration of blood biomarkers, DNA, and wearable device data. The conversation explores practical applications of AI, its power in personalization, its role in democratizing health insights, and important challenges such as bias, data privacy, and the careful interpretation of vast datasets. The aim is to break down how AI can empower individuals to take control of their health and optimize their healthspan, not just lifespan.

Key Discussion Points & Insights

1. Measurement and Personalization in Health

• InsideTracker's Approach:

  • Measurement: Blood, DNA, wearable device data, and user-generated information (04:05–11:20).
  • Analysis: AI processes this to give individualized, actionable health recommendations (04:05).
  • Focus: The aim is to help people optimize their healthspan — years spent healthy and active, rather than simply increasing lifespan.

• Quote:

  "Out of all of the years you're going to live, you want those years to be healthy... So we are aiming to do that through essentially techniques that will prevent disease, prevent these diseases of aging..." — Renee (04:36)

2. Types of Data and What They Reveal

• Blood Biomarkers: Up to 48 routinely measured, including standard and advanced markers like APOB and insulin (05:12–08:19).

• DNA Analysis: Offers baseline, inherited risk information — helps explain (and target) persistent health issues (08:22–09:30, 18:55–22:46).

• Wearable Device Data: Passively collected, continuous insights; includes platforms like Oura Ring, Fitbit, Apple Health, Garmin (08:23–09:30).

• User-Generated Inputs: Lifestyle, diet, location, habits add context for recommendations (09:30–11:20).

• Quote:

  "Fitness trackers are wonderful because you just wear them and they passively collect information." — Renee (08:23)

3. The Power and Complexity of AI in Healthcare

• Multiple AI Approaches:

  • Data-Driven AI (machine learning): Finds patterns in huge datasets; predicts outcomes like disease risk (24:17–30:12).
  • Knowledge Representation & Reasoning: Encodes expert-curated literature to inform recommendations; meta-analyses for personalized interventions (24:17–26:50, 44:57–49:17).

• Practical Example: Machine learning predicting 10-year cardiovascular risk from 100 biomarkers (27:46–30:12).

• Quote:

  "We are absolutely drowning and swimming in data. Very good at collecting it, not very good at making sense of it. But machine learning helps." — Renee (24:50)

• Quote:

  "It's just very good at pattern finding. Right, like that's what it's essentially doing…" — Renee (30:13)

4. Marrying Bloodwork and DNA – Why Both Matter

• DNA helps explain “unmovable” health factors (example: stubbornly high LDL) and separates what can be changed via lifestyle from genetic ceilings (17:43–22:46).

• Empowers better decisions: understanding when medication is warranted versus when lifestyle can suffice.

• Quote:

  "It's just, it's good to know that because we don't want somebody bashing their head against the wall trying to do something that frankly, it's not going to pan out for them." — Renee (21:27)

5. Wearable Data: Utility & Confusion

• Wearables (rings, watches, mattresses) offer huge value in trending health metrics, but often differ in algorithms and results (31:23–34:35).

• Stick to interpreting trends rather than obsessing over device-to-device discrepancies.

• Quote:

  "I've definitely worn Aura, an Apple watch, and my Garmin and gotten three very different answers for what happened in my sleep the night before." — Renee (32:24)

6. The Future: Greater Integrations & More Data

• Envisions combining blood, DNA, questionnaire, and advanced imaging (MRI, CT, CAC scores), microbiome, epigenetics — for a truly “high-definition” health picture (34:35–37:12).

• Quote:

  “We will do despicable things for data. Like, the more, the better... the more of that we can capture, the more of a high definition picture we're going to have of what's actually happening.” — Renee (35:07)

7. Drawbacks, Bias, and Fairness in AI

• Leaky Inputs: Most historic health data — and thus training datasets — come from white men, risking biased predictions in others (37:24–40:07).

• Results can be less accurate (or even harmful) for underrepresented groups (women, minorities, trans, older, younger people).

• Quote:

  “But the first thing that comes to mind is bias, right? So if you are trying to build a machine learning model... you do require training data, and that data is subject to bias.” — Renee (37:24)

• Women in Research:

  • Underrepresentation due to biological complexity (cycling), leading to greater risk of adverse drug events (39:40–40:07).

• Data Standardization: Time of day and other confounders are not always captured, adding more variables to analyze and control (40:07–40:40).

8. Data Privacy & DNA Fears

• Concerns about genetics information being tied back to individuals and misused by employers or insurers; people are rightly protective (41:10–42:34).

• Importance of vetting company data policies is stressed.

• Quote:

  “Any company that is collecting sensitive data like DNA sequencing information, you should do your homework and make sure that they're taking this very seriously.” — Renee (41:10)

• Some people also fear learning about genetic risks, but Renee and Louisa advocate for knowledge as a path to healthspan, not just longevity (42:34–44:30).

9. Education and Accessibility

• AI, including tools like ChatGPT, can democratize understanding but must be met with a healthy skepticism and education on how to spot good sources (44:57–49:17).

• InsideTracker’s system acts as an always-on expert meta-analyst — parsing literature, as a clinician would, to make recommendations.

• Quote:

  “That whole process, just for that one question, can take weeks. And I don't have that time. And I don't have that time to do that for every blood biomarker or every lifestyle intervention. But AI enables you to take that and capture it and store it and then reuse it over and over and over again...” — Renee (48:09)

Timestamps for Key Segments

• Blood vs. Wearable Data & Data Age: 00:00–00:33, 24:50–25:21
• What is InsideTracker? 03:25–06:50
• Adding DNA & Wearable Integration: 08:19–09:30
• AI in Analysis (Data-driven & Reasoning): 24:17–26:50
• Personal Health Team by AI Analogy: 14:28–17:43
• Genetics & Personal Examples: 18:55–22:46
• Bias and Underrepresentation: 37:24–40:05
• Privacy Concerns Around DNA: 41:10–42:34
• Using AI to Make Results Accessible: 44:57–49:17

Notable Quotes & Memorable Moments

• On Health Optimization:

  "That's like having a board of directors for your health. Really?"
  Louisa, 15:34

• On AI Pattern Recognition:

  "It's just very good at pattern finding. Right, like that's what it's essentially doing."
  Renee, 30:13

• On Data Privacy:

  "Any company that is collecting sensitive data like DNA sequencing information, you should do your homework and make sure that they're taking this very seriously."
  Renee, 41:10

• On Genetics and Health Outcomes:

  "It's good to know that because we don't want somebody bashing their head against the wall trying to do something that, frankly, it's not going to pan out for them."
  Renee, 21:27

• On Wearable Data Frustration:

  "I've definitely worn Aura, an Apple watch, and my Garmin and gotten three very different answers for what happened in my sleep the night before."
  Renee, 32:24

Conclusion & Takeaways

This candid, expert-driven conversation highlights not just the potential of AI to revolutionize personal health but also the pitfalls and ethical considerations that come with such transformation. The major through-line is that AI's real value is in delivering practical, evidence-backed, and individualized insights, bridging big data and real-life interventions to empower people toward proactive longevity. However, the field must address bias, protect privacy, and focus on inclusive, understandable progress to realize its promise for all.

Guest/Resource Links:

• InsideTracker
• Find Louisa on Instagram: @louisanicola_