Beyond Logic & Knowledge-based Systems in Clinical Decision Support

Date: July 17, 2014

Bill Maris, Managing Director of Google Ventures recently stated in an interview with Re/code that “Medicine needs to come out of the Dark Ages now”[1]. Of course he wasn’t inferring that modern medicine was still based on superstitions as most Dark Ages practices were. But in the Dark Ages, after the fall of the Ancient civilizations of the Romans, Greeks and Egyptians, there was very little advancement in medical knowledge; advancements in medicine had stalled[2]. Since the end of the Dark Ages, there has been a massive movement forward in medicine and technology. However, some antiquated practices have remained, even in the midst of such advancement. These practices revolve around recording, managing and effectively using clinical data. Many hospitals still record patient data on paper and treatment decisions are still often made based on the “expertise” or the “knowledge in my head,” as Maris put it, without outside reference or decision support tools. Maris hopes that those days will soon disappear and so do I.


I believe we are starting to “come out of the Dark Ages”. With the advent of portable smart devices, supercomputers, cloud computing and a host of other technologies, medicine is beginning to make strides in recording, managing and making more effective use of data. Over the last decade or so, many new EMR/EHR, middleware, interoperability and other clinical data collection companies have formed to solve these problems. But all of these technologies bring more and more data that clinicians have to digest and make decisions from. This can be incredibly overwhelming. That is why the crucial part of the solution is the Clinical decision support (CDS) tools that aid clinicians in making better diagnosis and treatment decisions.


CDS tools are becoming more advanced as they gain access to more clinical data, though CDS is still in its nascency both in its maturity and adoption. Worldwide spending on CDS tools was only around $340 million in 2013[3], with most current CDS tools sold being knowledge-based systems that aggregate clinical knowledge, patient data and nationally recommended guidelines. The other type of CDS tool is the non-knowledge based CDS systems that rely on rules and logic to help direct clinicians to the best possible diagnosis or treatment. These types of tools are important for clinical practice, but the logic and rule-based math that they use is simple, they have limited ability to simultaneously analyze multiple streams of data, and much of the support is reactive (such as an alert when something has already happened). These tools are far behind decision support tools in other industries that can analyze multiple streams of data, perform highly complex computations, in real-time and provide predictive analytics.


With such real-time, predictive tools, clinicians can predict the onset of disease before visible symptoms occur. This will have an immense impact on healthcare, improving patient outcomes and reducing healthcare cost. The good news is that at Medical Informatics Corp, we’ve already built a platform called Sickbay for developing, testing and deploying real-time, predictive CDS tools. Sickbay is a grid-computing system that was built to collect, store, and process the high-resolution physiological data from all bedside monitors and devices from every patient bed, across entire hospital systems. This is the 95% of physiological data that isn’t being collected by any EMR or EHR on the market today. Sickbay has already been used to detect the onset of sepsis, sleep apnea and heart attacks an hour before they happen in certain patient populations. We’ve already begun developing the first set of CDS applications and though we have over thirty applications in our development pipeline, there are countless diseases and conditions for which applications can be developed.


Given that healthcare spending is astronomical and growing, hospital budgets are being squeezed and there are favorable governmental programs for implementing CDS tools, CDS tool adoption must grow to help reduce the cost of care, though the lengthy sales cycle into healthcare and regulatory requirements may retard the growth. According to a Markets and Markets report, the CDS market is expected to grow at a CAGR of 10.4% through 2018. This would put the worldwide market at about $558 million in four years. This number may be on the low side, as the total Healthcare Analytics market is estimated to be $4.4 billion and growing at 25.2% through 2020[4]. With so much potential for market growth and possibilities for developing disease predicting CDS tools, I’m excited to see the great impact to the Healthcare industry – that should shed some light on the Dark Ages.


[2] “Medicine in the Middle Ages”. 2005. Web.



Share This Post

  • Subscribe to Blog Alerts

  • Blog Categories

  • New call-to-action