Meditations On Theranos

The COVID-19 situation, as well as my own interest in neuroimaging / healthcare imaging technologies, has caused me to think about Theranos a lot lately. The connection between the present context of my life and Theranos is admittedly fuzzy, but I think that I associate the two primarily due to Theranos' status as a health tech company.

I find myself to be largely supportive of Theranos' stated goal of "democratizing medical testing" with their Edison machines. The single largest contributor to the current COVID-19 crisis in the United States has proven to be a lack of diagnostic tests, which I think is ultimately attributable to:

  1. Political incompetence on the part of the Trump administration.
  2. The increasing bureaucratization of hospitals (best exemplified by technologies such as Epic, which are less tailored to the needs of doctors and patients and more tailored to billing departments). This bureaucratization, and the rise in a management class in hospitals has led to effects similar to those in academia (as described in Benjamin Ginsberg's The Fall of the Faculty; a book I have not read, but which I imagine should be taken with a grain of salt). Namely, there is a large body of people that actively impede progress in responding rapidly to patient and public needs. Many of these people likely have the bullshit jobs bemoaned by David Graeber.
  3. The tendency for Americans to not seek out medical care early enough due to our Frankenstein of a healthcare system.

Ultimately, I don't have any confidence in the ability of large institutions of any stripe (private or public) to deal with the most challenging problems of our time. In the case of COVID-19, I believe that this crisis could have been averted if testing kits were widely available and readily accessible in pharmacies. If any given U.S. citizen was able to just walk into a pharmacy and pick up a testing kit, without the overhead of having to visit a doctor or hospital (and incur some undoubtedly excessive co-pay), this crisis would not be where it is today. Essentially, if medical diagnosis were democratized, barriers were lowered, and testing kit production was adequate (which is admittedly a big if) I don't think I'd be holed up in my apartment currently.

I am not bullish on America's ability to fix its healthcare system, reduce excessive bureaucracy in hospitals or elect knowledgable leaders at any level of government. Currently, I think that the best we can do is continue pushing for change in these arenas (and, perhaps more importantly, in education), while also acknowledging that change is unlikely to happen anytime soon. A better investment of time would seem to be to work on improving medical instrumentation, and ultimately make diagnostic tools like MRI scanners and blood tests as safe, reliable and available as over-the-counter drugs, blood pressure tests and thermometers. Essentially, I'm more optimistic about our ability to shift diagnostic services traditionally provided by hospitals down to pharmacies / a self-service model, although I think that this will still require an extremely substantial amount of work. In the end, I think that this would make your average pharmacy OTC/supplements area much much more useful than it currently is (I've always found it amusing that pharmacies have rows upon rows with myriad variants of the same 3 drugs: ibuprofen, acetaminophen, and aspirin).

From my perspective, Theranos' concept was desirable, but its implementation was impossible. The tragedy of Theranos stems from the fact that the science wasn't there- Elizabeth Holmes was warned by several field experts at Stanford before she dropped out to found Theranos. A drop of blood does not have adequate statistical power to detect anomalies compared to the 3-4 vials typically needed. In Spinozist terms, Theranos was prone to imagination over reason, and derived knowledge from random experience rather than from adequate knowledge. In particular, Theranos was inspired by the zeitgeist of the early 21st century Silicon Valley world, without understanding the basis for the Bay Area's techno-optimism.

An endemic part of the Bay Area's tech industry culture seems to be an overreaching application of technological concepts to societal issues to an almost tragic extent. I won't enumerate the full multitude of sins in this meditation, but I will focus on the sin that I believe Theranos was most guilty of: implicitly applying Moore's law to situations where its assumptions did not hold.

Moore's law (really an observation rather than a law) can only narrowly be applied to transistor density on integrated circuits. It predicts an exponential growth in the number of transistors that can be fit on a circuit as years elapse. Basically, all of Silicon Valley's computer hardware and software-centric tech companies are heavily dependent upon the increased performance that can be reaped as a result of Moore's law, which has enabled the average computer to perform tasks that would have been impractical in the past. Yesterday's supercomputer has been miniaturized to today's desktop PC, and there are doubtless microwaves and disk controllers with more computing power than top of the line machines in bygone eras.

This miniaturization, however, does not apply equally well to medical or scientific instrumentation, however. As obvious as this should be, Moore's law does not apply to all of nature, and was only scoped to a very specific scenario. This fact seems to have been lost on Theranos, who thought that they could condense a full-fledged diagnostic lab into a mini-ITX case (they at least seem to be in good company). In summary, I regard Theranos' chief technical and psychological flaw to have been a fixation on making physical objects smaller, derived from misapplying a cognitive schema that belonged to another context (i.e., something like functional fixedness).

If the tech industry is truly hoping to make a difference in democratizing healthcare, I believe that it needs to focus on applying what it knows best- getting peak performance out of modern hardware. Instead of Theranos, I think that there needs to be software out there that can perform fNIRS/EEG/MRI preprocessing or FWI quickly and accurately enough that it can be of use to everyday folk. This software does not exist yet. I think that there needs to be widely run software and hardware architectures that can simulate protein folding, similar to D.E. Shaw Research's Anton, foldingathome, and Rosetta@home. These are the areas where the tech industry can make an impact. These are areas where programmers are in their element. This is the context where a programmer cannot violate the laws of physics or overestimate our ability to make observations of the physical world.