VTE comprises deep vein thrombosis (DVT) and pulmonary embolism (PE). DVT involves the formation of a clot in a deep vein, like those found in the leg and pelvis. If the clot dislodges, it can travel to the lungs and become a PE. The condition is deadly and kills more people than AIDS, breast cancer, prostate cancer, and traffic accidents combined in the US and the UK (Johns Hopkins Medicine 2014).
Many factors put an individual at risk of DVT including inherited blood-clotting disorders, prolonged bed rest or lack of movement, hormone replacement therapy, smoking, cancer, family history, and age (Mayo Clinic 2014). Current tests for the condition run the gamut from expensive and accurate (e.g. contrast venography) to inexpensive but inaccurate (e.g. observation and assessment) (Schumann, MD and Ewingman, MD MSPH 2007). Among those who survive a DVT episode, half will have long-term complications and about one-third will have a reoccurrence within 10 years (Centers for Disease Control and Prevention 2012).
A statistical tool called the Wells Score was developed to predict the probability of DVT. This score presents a combined total across ten variables that impact an individual's risk of developing the condition. The Wells Score is often calculated in conjunction with blood tests to determine a possible DVT risk level. The challenge with this approach is that it is one-dimensional.
As with all statistical algorithms, the Wells score cannot account for context and multiple variables hidden within an individual's health history. For example, the Wells score increases with the presence of cancer. The problem is that it treats all cancers the same when in fact different types of cancers - such as ovarian cancer - carry with them a greater DVT risk. Additionally, other aspects of a patient's history including gender, age, race, socioeconomic status, and recent travel history are not included in the weighting. While the Wells Score is an important tool, it simply falls short in capturing the complexity that underlies an individual's risk of DVT and subsequently PE.
Enter predictive analytics. New solutions that use advanced machine learning and much of the same technology that underlies sophisticated search engines are making their way onto the healthcare market. These solutions go far beyond what is captured by established statistical measures like the Wells Score to account for the millions of variables hidden within a patient's medical history. They do this by analyzing attributes, defining patient clusters, and predicting risk. The best of these technologies include the clinical rules and proven, evidence-based measures captured by things like the Wells score. But they go far beyond their predecessors to see patient risk within the context of the entire population. The results provide a much more accurate and meaningful picture into actual disease risk to enable more effective clinical action.
Centers for Disease Control and Prevention. Deep Vein Thrombosis (DVT) / Pulmonary Embolism (PE) — Blood Clot Forming in a Vein. 06 08, 2012. http://www.cdc.gov/ncbddd/dvt/data.html (accessed 11 13, 2014).
Mayo Clinic. Diseases and Conditions. 07 03, 2014. http://www.mayoclinic.org/diseases-conditions/deep-vein-thrombosis/basics/risk-factors/con-20031922 (accessed 11 13, 2014).
Schumann, MD, Sarah-Anne, and Bernard Ewingman, MD MSPH. "Is it DVT? Wells score and D-dimer may avert costly workup." National Center for Biotechnology Information. 12 2, 2007. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3183832/ (accessed 11 13, 2014).