By Dr. Michael Schmidt, Director of the Neurological Intensive Care Unit Neuromonitoring and Informatics program at Columbia University Medical Center
The goal of personalized medicine is to tailor all decisions and practices to the specific needs of an individual patient. Anyone that has gone to the store to purchase a customized birthday cake rather than buying one already in a box, or went to a flower shop to pick out the exact flowers you wanted rather than buying a premade floral arrangement, already understands the benefits and associated costs of customization. First and foremost customization requires time to understand the specific needs of the customer. Sometimes communication breaks down and what you thought was made perfectly clear is not what gets conveyed to the person helping you. This results in a mistake leaving you to either start from scratch or walk away with not exactly what you wanted. A customized experience offers you the benefit of getting exactly what you want or need, but at the cost of more time invested, greater risk of mistakes, and of course at greater monetary cost.
In most respects evidence-based medicine as it is currently practiced favors consistent care over customized care. Clinical protocols specify how patients with certain conditions are to be treated in a one-size fits all fashion. We know from rigorously conducted clinical trials that more people will benefit from our treatment protocols than not. It is not a bad model and many studies have shown that treatment protocols benefit patients. Patients receive treatments that have been shown to be beneficial, mistakes are minimized because the components of treatment are pre-specified, and it reduces the time needed to treat patients properly in an already overloaded medical system. In fact a lot of resources are devoted to ensuring that clinical protocols are followed more meticulously rather than there being inertia to move away from protocols towards customization.
Having said that, the underlying infrastructure and knowledge base to support personalized medicine is starting to take shape. Characterization of the human genome has led to some of the most noticeable examples for personalized healthcare. For example for some kinds of cancers physicians are now able to prescribe the best cancer-fighting drug for your personal genetic profile, such that someone else with the same cancer but different genetics would receive a different drug.
In an intensive care unit, such as the one I work in, people have life-threatening injuries and illnesses. We monitor our patients intensively looking for any signs that our patients are having problems, and hopefully evidence that our treatments have been effective and that our patients are recovering. Our physicians and nurses are confronted with more than 200 individual data points about every aspect of the patient each and every morning. Yet cognitive research tells us that people are not capable of understanding the relatedness of more than 2 variables without help, and unlike cancer patients who are generally stable, ICU patients are changing dynamically hour-to-hour and sometimes minute-to-minute.
The result is that ICU care is highly dependent on standardized treatment protocols, and with good reason. Clinicians are able to institute primarily beneficial treatment protocols based on results of specific tests without having to understand the interconnectedness of all the tests to each other that personalized medicine would require. This also means, however, that ICU care is largely reactive with clinical staff needing to wait to observe something abnormal so that they can administer a treatment designed to normalize it.
Our goal is to utilize analytics to more fully understand the hidden complexities and interconnectedness embedded within the data of each individual patient. This information presented to clinicians in a timely and easily digestible form would enable us to better customize our treatment to the needs of the individual by identifying patient-specific physiological endpoints to guide goal-directed therapy. We could identify patients at high risk for catastrophic complications before they occur so that we can intervene and prevent the complication from ever occurring in the first place. Instead of reactive standardized care patients would receive proactive customized care that ideally would lead to better patient outcomes for everyone rather than better patient outcomes, on average, that standardized medicine promises.
I firmly believe that personalized medicine will be the way future medicine is practiced. Clinical decision support tools that analyze, synthesize, and present real-time data to clinicians in form that enables them to make the best decision for each individual patient in a fast and accurate manner are an essential component of that future. Our work with IBM is moving us closer to that goal.
To learn more about how IBM is helping clients overcome big data challenges with analytics technology, check out this video: