Imperfectly Precise

In the Fractal Nature of GeometryBenoit Mandelbrot stated, “clouds are not spheres, mountains are not cones, coastlines are not circles, and bark is not smooth, nor does lightning travel in a straight line.” In other words, nature is not simple and regular but complex and filled with irregularities and roughness. The human brain reductively categorizes this complexity in order to extract meaningful information and make predictions about the environment. Technology and the scientific revolution has increased our capabilities of systematically smoothening, sorting, and simplifying clouds, mountains, and lightning into neatly defined boundaries of spheres, cones, straight lines. The tools of the scientific method have exponentially increased our modeling capabilities. In the same vein, medicine has also advanced with a largely reductive approach as the human body and diseases are classified into distinct organ systems, cell types, and disease subtypes. However, there is a two-fold cost in fitting complex, gradable phenomenon into discrete, discontinuous categories. Firstly, since these categorizations are approximations, there is the loss of boundary information and if our categories do not capture the phenomenon at the appropriate level of the prediction, our models and predictions become imprecise and error-prone. Secondly, and maybe more insidiously, once categories crystallize, they become reality rather than a representation of reality. Thereafter, they frame the scope of our discussion and limit our ability to find solutions. The goal of precision medicine is in large part to improve the former from the relatively coarse-grained designations to more refined states.

In approximately 230 BCE, the Greek physician Apollonius of Memphis is credited to first use the term “diabetes” which translates “to pass through.” Thereafter, in the 5th century CE, two Indian physicians – Sushruta and Chakrata – were the first to differentiate between the two types of diabetes mellitus. They noted that thin individuals with diabetes developed the disease at a younger age [Type 1 diabetes] in contrast to obese patients who were diagnosed in their later years and lived longer following the diagnosis [Type 2 diabetes]. The silk road polymath, Avicenna (Ibn-Sina), compiled a medical text titled the Canon of Medicine in which he included a detailed description of diabetes. Clinical features such as sweet urine, increased appetite, diabetic gangrene, and sexual dysfunction were described in detail in this text. This classification remains to the present day and has been utilized as the framework to elucidate the mechanisms of these diseases. This, in turn, has led to impressive gains in diagnostics and therapy.

In a recent study in Lancet, researchers in Scandinavia proposed reclassifying diabetes into five categories. In this classification, tScreen Shot 2018-03-28 at 8.50.43 AM.pnghree of the categories (SIRD, MOD, and MARD) were related to Type 2 Diabetes and two (SAID and SIDD) was similar to Type 1 diabetes. MARD and MOD contribute to almost 60%of disease burden. These new categories were created bDM_Subtypes.pngy considering six different metrics ranging from demographics (age of onset and body mass index)  and traditional serum markers (hemoglobin A1c) to non-routine markers such as (measures of insulin sensitivity and presence of autoimmune antibodies). Furthermore, genetic markers such as TCF7L2 gene mutation was evident in clusters 2, 4, and 5 but was not seen in cluster 3. Additionally, cluster 3 subtype of patients are most vulnerable to developing devastating sequelae such as vascular disease, and renal failure whereas, cluster 1 and 2 were at higher risk of diabetic ketoacidosis. According to the authors, “an important goal is…to identify more refined subtypes to accurately predict clinical outcomes and identify targeted therapies that ameliorate them.”

Marvin Minsky stated, “a definition is enclosing a wilderness of ideas within a wall of words.” Definitions serve to frame, and oftentimes limit our thoughts, solutions, and perspectives. For the most part, they are necessary to reduce the immense complexity present in the natural world and extract information to make predictions. Moreover, these models have become exponentially more precise and our predictions more accurate through the methodologies of the scientific method in tandem with technologies. However, we mistake our categories of nature as inherent properties of nature, rather than representations of reality. Nature is not discrete nor absolute and thus our discrete and absolute categories are approximations. Nonetheless, as we continue to improve our tools and apply the scientific method to map, measure, and categorize every nook and cranny of the “mountain, coastline, and bark,” our predictions will continue to improve. But in that asymptotic but non-linear procession towards perfect precision, it is equally important to recognize that our models are representations of nature and not nature itself.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s