Health Care, Big Data, and Black Swans

The movement towards using “big data” in health care has the ability to vastly improve the way medicine is practiced.  The term “big data” is used to describe large databases processed by powerful processors analyzing and recognizing trends and patterns that would not otherwise be apparent.    Today, research into the use of big data is centered on the analysis of the human genome and the resultant individual variation of disease and response to therapy, and in the use of health insurance and electronic medical record data to find patterns of care in order to potentially improve quality and lower costs. 

The genomic research is moving from the description of the genome into the more practical application in the treatment of disease.  The goal is to create “personalized” medical care based on each individual’s specific genetic makeup.    The most obvious use is the application of genomic knowledge to define specific sensitivities to certain drugs for certain illnesses so that the medications can be customized on the basis of this individual data. 

The use of the claims and EMR data is more likely to be used in the realm of developing best practice protocols and safety protocols.  A supplement in the latest issue of the Annals of Internal Medicine with an accompanying editorial show how the use of data combined with expert protocols can lead to greater safety and the avoidance of critical errors in the hospital. 

As complex as these two areas are, it should come as no surprise that they don’t yet meet or coordinate.   We are too early in their development.  We do not yet know how to combine a “big data” approach to populations to the “big data” of the individual.  When dealing with populations, individual variations of the type that we try to define with “personalized” genomic based care, is often just “noise’ in the statistical system of the population.  So the trend towards personalized care may be contrary to the trend towards evidence based care based on large databases. 

Both of these somewhat different approaches also leave out an entire domain of the complex business of treating individual patients.  We may not be asking some of the most important questions.  All of these applications are focused on services or encounters, costs and biology.  There is no focus in either a genomic approach or an encounter, claims approach on people as complex, multifaceted individuals who are living social lives in the context of their own emotions, beliefs, economics, and values.  People are complex and that complexity is only partly related to their biology and their claims records.  

Perhaps we need to ask what defines good medical care as another starting point.  The hallmark of a superb diagnostician, is the one who can recognize and look for black swans.  Nassim Nicholas Taleb, who is a distinguished professor of risk engineering at New York University and the author of the book, “The Black Swan: The Impact of the Highly Improbable” is someone who has studied risk and chance in finance for much of his professional life.  He defines a black swan as an event, positive or negative, that is deemed improbable yet causes massive consequences.   In medicine, these events can have death or chronic disability as the consequence.  For me and my family, I want a doctor who will always be thinking about potential black swans while also minimizing the risk of sending me on diagnostic wild goose chases (to mix bird related metaphors).  I wonder if systems that are too focused on the best practice protocols, developing the evidence based medicine, will have that ability that we see in the truly expert, experienced clinician or if the best practice protocols will actually punish that sort of thoughtful, experienced based approach by seeing it as being inefficient and not in keeping with the evidence that supports the bulk of medical care. 

The hallmark of a master clinician is one who customizes the decisions, in partnership with his or her patient, taking all the data, including the social, emotional, financial, and spiritual into account in managing the problem at hand and in managing the potential for future problems.   I also worry that we don’t have databases that reflect the real reasons people get better or not.  The master clinician asks the right questions and listens for the answers with a skill that currently no computer can match.  Does the patient have the family support that is needed to deal with illness?  Can the patient afford the right diets?  Does their religious belief create a social network of support and a psychological framework to allow someone to deal with their physical condition in a way that fosters healing?  Are they worried about other family members and that is leading to their spending less time caring for themselves?  Do they lack basic shelter?  While this may also suggest the need for yet a third domain of big data we also have to be concerned with the moral implications that this third type of data may bring.  The loss of privacy and the ability to misuse data such as this is great.

 

So I hope that we make good use of the big data systems that we are now building but I also hope that we never let these systems become a new type of electronic care giver that ignores all of the important elements that we are not, and perhaps never will, capture.  Let’s also hope that these efforts at big data and evidence based protocols do not lead to our ignoring the possibilities of those black swans in medicine that can devastate people and families. 

Most of all, let’s make sure that in any system of care, we always ask the right questions to help us understand the people behind the illness and to truly hear their answers.  We will always need to find, as Professor Taleb puts it, the “bird droppings” of the black swan so we don’t miss the illness that can be a catastrophe.