This chapter covers the argument, made by ER physician Dr. Mike Min on the 'Don't Die' podcast, that the large majority of disease is eventually preventable and that emergency medicine is structurally positioned downstream of that fact — plus the two self-directed countermeasures proposed alongside it: biometric self-tracking paired with deliberate subjective attention, and using an LLM to generate a differential and a workup list you then push your provider to actually run. It also covers the caveat that undercuts the first countermeasure — the body's ability to compensate silently until it can't — and is honest about the fact that all three ideas come from a single podcast episode.
The chapter's foundation is a positional argument rather than a clinical one, and it comes with a specific attribution: ER physician Dr. Mike Min, speaking on the 'Don't Die' podcast. His framing is worth keeping in his own words, because the complaint is about where in the causal chain his job sits: "Emergency medicine is just like banging your head against the wall all day long... fighting issues that are secondary to a culture that doesn't value health."
The numbers attached to that framing in Preventable Disease & the Reactive-Medicine Framing are large. More than 90% of disease is described as eventually preventable in general, and roughly 80% of ER cases specifically. The distinction that makes those figures coherent is between acute and chronic: trauma, appendicitis, and infections are genuinely hard to prevent, but chronic disease — described here as the majority of what modern medicine treats — is held to be largely preventable through lifestyle. The claim is therefore not that acute-care medicine is useless. It is that the center of mass of what walks through the door was decided years earlier, somewhere the system has no reach.
A system comparison is offered as evidence of the load this creates. The US ER system is described as chaotic and overburdened relative to Australia and New Zealand, with US doctors reportedly seeing about 30 patients a day against roughly 10 a day in New Zealand — a threefold effort differential, attributed partly to the preventable-disease burden. That hedge is in the source material and worth preserving: nothing here separates the preventable-disease contribution from staffing, funding, or how each system routes patients.
It is also worth naming what kind of claim this is. These are a practicing physician's impressions delivered in a podcast conversation, not a cited epidemiological estimate — no definition of "preventable," no time horizon, no study behind the 90% or the 80%. By the yardstick this book sets up in Evaluating Evidence & Supplement Quality, that places the framing in the category of expert opinion: useful for orienting, not for settling. Everything that follows in this chapter is a countermeasure proposed on top of it.
If the preventable share of disease is the target, the first proposed countermeasure is measurement you run on yourself. But the version described in Body Awareness via Biometric Self-Tracking is not simply "wear a ring." It is a pairing: objective readouts — heart rate, sleep score, biological-age tests — held deliberately against subjective attention to your own state, repeated until the two calibrate each other. Bryan Johnson describes the endpoint of that process as a trained sense rather than a dashboard habit: "I know after doing my own health protocol for the past couple years, I'm intensely body aware... I can feel my heart rate at any second... with pretty good accuracy."
The worked example is sleep. Sleep score is described as correlating strongly with next-day work quality, creativity, and cognitive capacity — and, importantly, with effects that accumulate across multiple nights rather than hanging on any single one. Johnson's phrasing puts the outcome in terms of agency rather than fatigue: "When my sleep score is low, I lack agency... When it's high, I can power through anything." That framing matters for how you'd act on the number. A single bad night is not the unit of analysis; a run of them is.
The underlying logic is the same one that motivates the whole chapter. A reactive system measures you when you present with a complaint. Self-tracking measures you continuously, on the theory that the preventable share of disease is preventable precisely because it announces itself gradually — which only helps if someone is watching during the gradual part. That is the connective tissue back to Preventable Disease & the Reactive-Medicine Framing: the tracking isn't presented as a hobby, it's presented as the individual's substitute for surveillance the system doesn't provide.
The most methodologically interesting piece of Body Awareness via Biometric Self-Tracking is the instruction not to take any single device's output at face value. The example given is Andrej Karpathy's own two-month test comparing four sleep trackers — Aura, Whoop, Eight Sleep, and Apple — against his subjective work quality and cognitive performance. The design is the point: the devices are not treated as instruments that report reality, they are treated as competing claims about reality, and the arbiter is a logged outcome the person actually cares about.
That inverts the usual relationship. A sleep score is only useful if it predicts something; running several scores against the same lived outcome tells you which one, if any, is tracking anything real for you. The chapter's practical instruction follows directly: pair wearable readouts with deliberate in-the-moment awareness of your subjective state, and periodically run comparative self-experiments rather than trusting one device. Note that the material records the experiment's design but not its verdict — which tracker won is not stated here.
The same logic extends to outcomes over longer horizons through biological-age tracking: measuring a pace-of-aging metric to check whether an intervention is actually slowing measured aging, rather than assuming it is. That is the natural handoff to Longevity Interventions, which is where the interventions worth measuring get classified and where the evidence for them is weighed. This chapter contributes the measurement discipline, not the intervention list — though it takes for granted, without arguing it, that a pace-of-aging metric is trustworthy enough to serve as a scoreboard.
The chapter then supplies the caveat that cuts against its own optimism about body awareness. The body is described in Body Awareness via Biometric Self-Tracking as highly adaptive — capable of suppressing a major problem for a long time before failing suddenly, in a cascade.
The illustration is an inebriation study. At low-to-medium intoxication, the brain compensates well enough to pass behavioral tests even though impairment is visible on scans; only at high intoxication does behavior finally break down. The generalization drawn from it is that cognitive decline begins long before it is behaviorally apparent, and could in principle be caught early by imaging rather than by noticing something is wrong.
Sit that next to the previous two sections and the tension is real, and the chapter does not resolve it. The case for body awareness rests on subjective attention becoming a finely calibrated sense. The adaptive-body caveat says subjective experience is exactly the channel that gets papered over — that you feel fine right up until the compensation runs out. The honest reading is that the two coexist as a division of labor rather than a contradiction: subjective calibration is a fast, cheap daily signal, while the things that actually kill you on the preventable side of Preventable Disease & the Reactive-Medicine Framing need an instrument that doesn't care how you feel. But that reconciliation is inference, not something the source material states.
The second countermeasure addresses the encounter itself rather than the years before it. LLM-Based Medical Self-Advocacy describes feeding your symptoms and complaints into an LLM such as ChatGPT, getting back a probabilistic differential diagnosis and a list of exams that ought to be run, and then using that output to push a provider to actually run them.
The method has a specific shape that is easy to lose. It is not a one-line question. It is a long, thorough intake — on the order of ~100 questions, describing symptoms in detail — and the ask at the end is not "what do I have" but "what workup should be happening for this presentation." The output is then carried into a real appointment as a request for specific exams. Framed by the 'Don't Die' team (Bryan Johnson and Dr. Mike Min), this is explicitly a navigation tool for an overburdened, reactive system: if the doctor has thirty patients today, the patient arrives having done the part that takes time.
The cited success case is Johnson using a ChatGPT intake that correctly flagged his mother's eye condition as inflammation or a corneal scratch, later confirmed by an optometrist and resolved with steroid drops. It's a clean anecdote, and it's worth noticing what kind of case it is — a visible, localized, self-limiting complaint with a fast confirmatory check available. The material offers no failure cases against which to calibrate.
The caveat carried in the concept itself is the important sentence. LLMs reportedly outperform physicians on standardized medical tests, but "life is rarely standard" — which is precisely a statement about how far the approach generalizes past textbook-shaped presentations. The whole thing is positioned as a supplement to advocacy with a real provider, never a replacement for one. What the material does not give you is any way to tell, from inside your own case, whether your presentation is the standard kind or the other kind.
This is the last chapter of a book whose subject is creatine, and none of its three concepts mention creatine at all. That is not an oversight worth papering over, so state it plainly: the mechanism, dosing, and safety evidence all live in Creatine Fundamentals, and nothing in this chapter revises any of it.
What this chapter contributes is posture. Every earlier chapter hands you an intervention — a protein target, a training variable, a recovery practice from Brain, Behavior & Recovery, a compound from Longevity Interventions, creatine itself. This chapter asks the question that sits underneath all of them: would you notice whether any of it worked, and who is watching you between appointments? Preventable Disease & the Reactive-Medicine Framing argues that nobody in the acute-care system is, by design. Body Awareness via Biometric Self-Tracking and LLM-Based Medical Self-Advocacy are the two answers offered — one continuous and self-directed, one deployed at the moment of contact with a provider.
The honest accounting on sourcing: all three concepts derive from a single podcast episode, and the chapter records no contradictions against other material in this corpus — not because the claims have been checked and held, but because this is new ground here with nothing yet to check them against. A 90%-preventable figure, a threefold patient-load differential, a four-tracker self-experiment, and one anecdote about a corneal scratch are not a body of evidence. By the standard set in Evaluating Evidence & Supplement Quality, this chapter is the least evidentially supported in the book, and the appropriate way to read it is as a set of testable practices worth adopting cheaply — not as findings.