Friday, March 14, 2008

The Barbell Strategy

The barbell concept: essentially, a barbell strategy is an asymmetric diversification scheme - it's a hedged portfolio. The bells receive most of the weight, while the bar contributes a small fraction to the whole (note: as we shall see, we leverage the bar, as demonstrated above).

Physician blogger, The Happy Hospitalist, speaks to the barbell strategy in one of his latest and greatest posts, "The Smokey Zebra on the River Congo" (click here to read). Dr. Jerome Groopman, one of my heroes, built a splendid case for this hedging approach as well in his most recent book, How Doctors Think - he captures the fallibilism (here) perspective in his mantra, "What else could it be?" In The Black Swan, Nassim Taleb, my hero of heroes, advocates for the barbell strategy in constructing financial portfolios (invest 80-90% of your $ in safe securities like T-Bills and the other 10-20% in an array of aggressive options: stocks, out-of-the-money options, venture capital, start-ups, and others) in order to limit known downside risks (as much as possible), insuring your portfolio against blowups, while maintaining ample exposure (as diversified as possible) to unknown, scalable upside payoffs.

In essence, the barbell strategy is an attempt to construct a portfolio, whether in finance, clinical medicine, personal skills, or the hunt for knowledge, that diversifies our investments, activities, and perceptions around "unknowledge," out of respect for the limits to our predictions about the future, our specialized skills, and our awareness of the true textures that govern and drive our increasingly complex and recursive world.

In his splendid Fractal Press blog (click here), Navanit Akeri, a Black Swan-platform tinkerer and thought leader, presents "25 rules of thumb for acting under uncertainty" (please click here to read). I encourage all to read them over a couple of times. Here are a few of Navanit's wise words of wisdom: "Be suspicious of stability or the appearance of stability - there are higher dimensions of risk lurking in the shadows." "Invest in preparation - insure against the worst-case-scenario." "Diversify. Massively." "Disconfirmation is more powerful than confirmation - the problem of induction." Much of what Navanit captures in these wonderful rules of thumb speaks to Nassim Taleb's primary mission to "factor for the invisible." Keenly, Taleb builds a portfolio that is robust to and insured against negative Black Swans, yet remains responsive to the positive ones - serendipitous events and unexpected breakthroughs. Clearly, factoring for the invisible is not an easy task ... but it is a noble goal.

Joseph Lister factored for the invisible when he discovered that bacteria cause surgery site infections, not "bad air" as the dogmatists of his day purported (click here). In breaking from conventional wisdom, Lister acted on a leap of intuitive faith and pursued his sepsis hypothesis to its full conclusion. Through this process, "the unseen became visible" and Lister revolutionized medicine forever - a fracture in healthcare history. Lister too advocated for a barbell strategy of focusing on and emphasizing asepsis (preventing bacteria from entering wounds via hand-washing regimes for surgeons, meticulously clean wards and instruments, and carbolic acid preparations to clean wounds), while using and prescribing anti-sepsis (killing infective agents in wounds) when preventive measures failed. When Lister published his empirical research in the Lancet in 1867 about bacterial agents causing infections, he detailed the immense reductions in morbidity and mortality that resulted from his asepsis and anti-sepsis surgery practices. Yet, as a pioneer in his field, he received many comments and critiques like the following response: "Where are these little beasts? Show them to us, and we shall believe in them. Has anyone seen them yet?" Within his own hospital, he encountered difficulties implementing his life-saving aseptic practices, most likely due to the problem of silent evidence: we scorn the abstract and the unseen ...

Following Joseph Lister's lead, physicians could engage barbell strategies for clinical judgment and decision making: 80 to 90% of cases, as Happy Hospitalist confirms, fall within known patterns of disease presentations and acuity levels - clinical decision tools, algorithms, and protocols prove useful in these cases involving mild uncertainty - while 10-20% of cases present atypically, fall outside normal templates of care, and require extended investigation and suspension of judgment - the "What else could it be?" medicine domains. Perceptively, Navanit Akeri suggested that a healthcare institution, such as a hospital, could achieve this barbell strategy by employing, possibly 1 or 2 out of every 10, hospitalists as "What else could it be?" specialists who round the wards looking for complex cases, poking holes in other physicians assumptions, and actively hedging against wild uncertainty in patients' disease processes. Building on this insight, I constructed some barbell strategies for medicine, applying the "80-20" rule as a conceptual, intuitive framework for tractability; please, do not platonify these numbers as concrete, universal realities - they are simply descriptive, qualitative statistics.

For instance, 80% of medical care could be provided by community-based primary care physicians, as they do in Cuba (click here). This emphasis would match people's healthcare needs most of the time, and then 20% of medicine could be provided by specialists.

Similarly, 80% of our healthcare efforts could be spent on prevention, proactive prophylaxis (click here), while 20% could focus on curative, reactive measures.

In building a personal skills portfolio, a doctor could practice medicine 80-90% of the time, seeing patients one by one, assuring a relatively steady income and job stability and satisfaction. Then, this doctor could write essays and do research with the other 10-20% of his extra work/hobby/free-time (Atul Gawande has leveraged this type of personal portfolio extremely well). The next book that I plan to read, Intern, by Dr. Sandeep Jauhar (click here), illustrates the positive outcome of employing this type of barbell strategy. Sandeep, an insightful humanist, composed his marvelous book while training and practicing medicine (non-scalable), producing a magnificent piece of scalable work in his extra work/hobby/free-time that will contribute to medicine today and in the future. Dr. Jauhar also writes for the New York Times, and his latest essay, "Many Doctors, Many Tests, No Rhyme or Reason," is worth reading (click here to read). His blog is exemplary as well (click here to read). As an aspiring essayist (via this blog and other interfaces), I have attempted to diversify my personal skills portfolio in a manner similar to how Sandeep Jauhar, Atul Gawande, and Jerome Groopman have done. Through my writing, I hope to contribute to our Advanced Knowledge Civilization in scalable ways. Luckily and thankfully, I did receive positive feedback about my blog from Dr. Sandeep Jauhar, so hopefully I can continue to follow his modus operandi and employ this barbell strategy in medicine successfully (I encourage others to do the same).

In the open world of personal health and wellness, Art DeVany's wonderful Evolutionary Fitness concept and practice (click here) respects the fractal, self-affine complexity of our physiologies (read about fractals, self-affinity, and power-laws by clicking here) by applying a barbell strategy to exercise, diet, and lifestyle patterns: "In using Power Law Training, we are affectively mimicking our ancestral dynamics. That may be why it produces optimal health and body composition." Power Law Training emphasizes intensity, variability, and novelty (Art DeVany's site is worth perusing if these dynamics interest you): most of your activities involve low-grade movements - the Tiger stalking his prey - and resting - the Tiger sleeps deeply - and then, all of the sudden, you engage in Power Law Training (an intense 12 minute interval/strength session on an elliptical, followed by continued efforts for 35 to 45 minutes lifting weights, for instance) and fracture your activity level, pushing your physiology into far from equilibrium processes. 

Linking Evolutionary Fitness' positive health benefits to cutting-edge physiology research in the Margret & H.A. Rey Institute for Nonlinear Dynamics in Medicine is fascinating (please click here - warning: if you peruse their outstanding website, you just might think about "homeostasis" in new and exciting ways). These curious investigators operate an open-source "laboratory without walls" - inspired by the spirit of Curious George, "that playful primate who serves as a most fitting emblem of scientific curiosity and exploration" - and these researchers work to find "hidden information in complex physiological dynamics." Read this article (click here) to learn more about healthy fractal physiological complexity versus the "collapse of complexity" in disease states like Congestive Heart Failure. Disease states and aging are associated with losses of nonlinear, multifractal complexity and with increases in periodicity (read here) - which, in real-world terms, bolsters DeVany's lifestyle message to avoid entrenched, linear routines; your mind and body thrive on nonlinear complexity. For example, Congestive Heart Failure patients' heart rates exhibit simplistic periodic rhythms (read more here). Our brain-heart-pulmonary connections are extremely complex, exhibiting self-affinity and scale-invariance in healthy people. Heart Songs (click here) represents one way that researchers at this lab have revealed hidden information: they map "heart rate time series of healthy and diseased hearts into musical notes" in an effort that, "hopefully, provides a way to begin appreciating the differences in the dynamics of health and disease." Creating tools for people to "tap into" the hidden information in their complex physiological dynamics and synch their heart, pulmonary, and neurological rhythms represents an exciting frontier for health progress.

Marching to a similar drum, HeartMath (click here), an innovative company that empowers "heart-based living" and builds biofeedback tools to help us unite the self-affine properties of our heart and brain functions - starting with our "heart brain" (ganglia within our hearts that send numerous afferent signals to our brains, communicating intimately with our brains) - helps us enhance our minds and spirits and unleash our abilities to think and act to the best of our abilities. This institution represents another intriguing research group that is trying to help people improve their health and wellness by helping them better understand their bodies. 

Overall, Evolutionary Fitness provides a wonderful platform for better matching our lifestyles, exercise habits, and nutritional choices to our evolved physiologies, maintaining and enhancing the healthy fractal complexity of our body rhythms along the way. Stepping back, I am amazed at the extent to which we modern humans have moved thoughtlessly and strongly to impose linearity, routines, and concreteness on our lives and world: as Taleb notes, "information asks to be reduced," and living in the Information Age, we seem to do a lot of reducing to the detriment of our health, wealth, and well beings - we build and organize far too many social and cultural institutions from the top-down that inhibit our creativity, intuition, and physical and mental development. Bottom-up construction, informed by the healthy nonlinear dynamics of our heart and brain functions, would create more fluid, dynamic, and flexible systems, similar to the Institute for Nonlinear Dynamics in Medicine, for people to live, work, and interact in.

In continuing my studies of predator hunting behaviors and foraging strategies, I have been reading about Levy walks (click here) - the strategies that predators have evolved to forage for food - which are random walks under Levy distributions (click here), creating Levy flights (click here), as described by the great Benoit Mandelbrot (click here) in his seminal book, The Fractal Geometry of Nature. Levy walks are barbell strategies. In a recent paper in Nature (click here), "Scaling laws of marine predator search behavior," researchers presented findings from studying marine predators' foraging decisions: "Many free-ranging predators have to make foraging decisions with little, if any, knowledge of present resource distribution and availability. The optimal search strategy they should use to maximize encounter rates with prey in heterogeneous natural environments remains a largely unresolved issue in ecology. Levy walks are specialized random walks giving rise to fractal movement trajectories that may represent an optimal solution for searching complex landscapes." As Navanit Akeri points out (click here to read his blog), though, "the optimality of Levy walks for predators may hinge on the prey population itself being Levy-distributed" - which means that information must be scarce and patchy in nature. Navanit Akeri posted his latest perceptive hypothesis - "amongst all stable distributions, the Levy-distribution is entropy maximizing (click here) under path-length constraints. That is, given finite resources and a fixed distance that a predator can forage, the Levy walk may be entropy maximizing (click here) under the set of all possible random walks within the constraint." Clearly, Levy walks represent one way to analyze the decision making problem of "what to do when you don't know and are constrained by finite resources." Learning from millions ... make that billions ... of years of evolutionary history seems reasonable and worthwhile, and, it turns out that a diverse spectrum of organisms engage Levy processes to forage for food successfully. 

Art DeVany picked up on this Nature article as well, posting an excellent commentary that links the properties of Levy walks to the concepts he advocates in Evolutionary Fitness (please click here to read his post) - plus, another study showed that human movement patterns follow Levy walks as well (click here). In short, Levy walks describe power-law activity patterns; my next step is to apply these dynamics to clinical judgment and decision making in order to optimize medical practice. The scarcity and patchiness constraint seems reasonable for medicine: information distributions about patients and their illnesses/disease processes are often scarce and vary across time, presenting new information in nonlinear ways, with floods of information cascading into the picture from time to time (patchiness observed as blood pressures dropping and ventilation rates suddenly increasing in coding patients, for instance). Lab values and disease rates (see epidemiology here) are patchy as well ... so I suspect that pursuing these Levy walks, entropy processes, and hunting behaviors / foraging strategies further could prove fruitful for clinical (wilderness) medicine (hunting). To develop a tractable version of this hunt for knowledge problem in medicine, clinicians could study test choices and test results as a Mutual Information (click here) problem under finite resources, in the face of real-world uncertainty ... more to follow at a later date.

I made a connection between foraging strategies and clinical decision making: when a patient's illness conforms perfectly to a common prototype and the doctor has all the patient's information collected and readily available to him, then the doctor is effective by following established protocols and algorithms - linear decision tools - for diagnosing and treating this patient. However, this sterile scenario erodes quickly in the real-world: patient's symptoms do not match templates and doctors work with incomplete knowledge and information about their patients' lives, minds, and biochemical individualities. In this resource limited ecology (doctors certainly must order tests parsimoniously), doctors must employ non-linear searching strategies to diagnose and treat patients effectively. Thus, informed by Nassim Taleb's book title, Dynamic Hedging: Managing Vanilla and Exotic Options, doctors must manage "vanilla" - common, prototypical illnesses - and "exotic" - variable or atypical presentations of diseases - options (health states) in order to keep their minds open to less common and novel possibilities within an ever-changing hunting environment. Conceptually, I suspect that a doctor could employ a barbell strategy in hedging against uncertainty while practicing the stochastic art we call medicine. In analyzing her historical case mix, a doctor may find that about 80% of her cases presented and unfolded as expected (roughly), with mild uncertainty. In these garden-variety cases, the available "evidence-based" medical decision tools probably prove useful and supportive. Linearity typically suffices when established, repeatable patterns present clearly. However, suppose that about 20% of the time (which is about the number of cases that doctors misdiagnose, interestingly), patients present with illnesses that are atypical or their bodies react differently to treatments due to biochemical individuality and unknown effects. In these cases, protocols hold little value and doctors must rely on astute clinical judgment and other tricks-of-the-trade learned from others and developed through experience. This is why Jerome Groopman always ask, "What else could it be?" while hunting for solutions in order to keep, as Richard Feynman suggests, "the door to the unknown ajar" and search widely for new clues and connections. This latter thought process, which I feel is the key to being a good doctor and hunter - follows Levy walk strategies and cannot be boiled down into simplistic, "cookie-cutter" templates. Clean, step-by-step algorithms for hunting prey (patients' illnesses) in the wild (healthcare settings) do not exist.

I shared this management of vanilla and exotic health states options barbell strategy with Navanit Akeri, and he linked this concept to Karl Popper's (click here) hopes for science in his book, The Logic of Scientific Discovery (here). In this masterpiece, Popper (an advocate for negative empiricism) discusses the fact that scientists are resistant to move to skeptical states and poke holes in their own theories. However, other scientists exist to do the poking for them. Applying this dynamic to healthcare institutions, Navanit perceptively suggested that an institution like a hospital could establish an overall 80-20 distribution, with 80% of doctors practicing protocol-driven medicine, while the other 20% specializing in "What else could it be?" medicine.

In response to this splendid insight, I have developed the concept of overlapping barbell strategies. Overlapping barbell strategies in medicine could reduce medical errors and could ensure/better continuity of care. Here is my thinking. In the US, everyone should have a personal physician, as they do in Cuba's system that emphasizes primary care (retainer practices appear to be free-market, US versions of the Cuban model). As practiced in Cuba, this personal family physician would only treat about 500 to 800 patients, depending on geography and other factors. In Cuba, these community-based, local family physicians make house calls, provide personalized preventive services, and then accompany their patients when they visit specialists or enter emergency rooms and hospitals. Cuba's family physicians provide about 80% of health and wellness services as well as public health measures. These family physicians know the most about their patients, so they would be in the best positions - most informed - to follow algorithms and protocols correctly. When their patients fail to respond to their care or when they feel their patients need specialized attention, these family physicians access specialists with their patients. To me, at this interface, there exists a crossover: generalist physicians, conceptually, employ 20% of their efforts mediating the "What else could it be?" care with the specialists who perform 80% "What else could it be?" care and 20% protocol-driven care. This overlapping barbell strategy achieves, in the outpatient setting, a 80-20 distribution for managing vanilla (80) and exotic (20) health conditions, while simultaneously employing generalist physicians to focus on vanilla (80) but still consider exotic (20) and specialists to explore exotic (80) while still considering vanilla (20). These interactions between generalist and specialist physicians would hopefully set up ecologies for medical judgment and decision making that would parallel Popper's ideals for science (possibly curbing the number of unnecessary tests that physicians order); and, each doctor would still engage "What else could it be?" thinking, just to varying degrees. Ideally, this same continuity of care barbell overlap would occur in hospitals as well, but only if these personal physicians had patient panel sizes and practices, as they do in Cuba, which allowed them to accompany and treat their patients in hospitals. Since, in the U.S., we have moved to "hospitalist" programs for primary inpatient care, hospitalists (80-20) and specialists (20-80) could engage overlapping barbell strategies as well. Alternatively, a hospital could also create two types of hospitalists: one that focuses on protocols and standards, and one that pokes holes in and probes deeper into diagnoses, treatments, and care decisions.

Since doctors vary in their abilities and desires to practice "What else could it be?" medicine (and the associated skepticism), I suspect that healthcare institutions could benefit tremendously - via reductions in misdiagnoses, drug errors, and incorrect or unnecessary treatments/procedures/tests - from achieving 80-20 distributions overall by employing two types of physicians (generalists and specialists, as one possibility, or two types of hospitalists, as another) to practice according to two different, overlapping barbell strategies.

In the end, the barbell strategy reminds us of the power-law dynamics that drive our world, our lives, our bodies, and our minds. In short, the barbell strategy conceptualizes a platform to apply broadly as we stretch the limits of being human while simultaneously respecting the humble limits of being human - the ultimate paradox. Stochastic tinkering upon a platform that will not fall out from under you unexpectedly (i.e. you will not blow up) provides entrepreneurs, innovators, and trial-and-error solution searchers with maximum exposure to the envelope of serendipity - it is better to pursue many small speculative investments than a few large directed ones: "Ah-Hah!" / "Eureka" moments will trigger transitions from extended searching to bursts of acting, and the unseen will be made seen, the invisible made visible, and the breakthroughs that change our lives for the better will engulf us, fracturing the textures of our world forever (click here to read more great insights from Nassim on this topic). This, my friends, is an optimistic view: the envelope of serendipity produces positive Black Swans that are far better, far more positive, and far more earth-shattering than we could have ever imagined (just "Google it").

My new hobby is collecting barbell strategies; let me know if you have one. I like to diversify ... opportunistic generalists tend to survive.

1 comment:

  1. FYI I've posted the first half of a refutation of Taleb's bar-bell strategy. I'm afraid you can't have your diversification cake and eat it too, and a tiny amount of geometric intuition should convince any serious student.

    http://quantapology.blogspot.com/2011/10/bar-humbug-debunking-bar-bell-portfolio.html

    ReplyDelete