DxEM started over thirty years ago when NASA began planning trips to the moon and Mars. This was the time when 2001: A Space Odyssey by Stanley Kubrick was still fresh in everybody’s mind (1968). A grant proposal was submitted by the College of Medicine at the University of Florida (UF) to NASA Kennedy Space Center for the construction of medical HAL, the paranoid computer in the Kubrick film. Nobody thought it could be done but they were willing to let UF try. For twenty years a team of physicians, programmers and textbook specialists worked to produce HAL, and it actually succeeded. This was long before the Internet and Google. At that time, the UF team was working with mainframe computers and rudimentary word processors. By the late 1990’s the project was ready to be tested in space with astronauts. Then something happened that changed its future. The NASA manned missions were cancelled, and robots and remote vehicles were introduced. The 20-year multimillion-dollar project was disbanded and all the material, staff, and programs were terminated. A great deal was learned in this twenty-year period about how to make computers medically intelligent and empower them to think and reason with sound medical logic. This core knowledge has now been incorporated into a new company called KQI, Knowledge-Quest Inc. (KQI).

One major spinoff of this early NASA experiment is KQI’s design of its medical databases. The UF staff learned early that computers can’t “reason” or comprehend medical textbooks and complex English text construction. Computers need data organized and logically constructed as modules and factoids that can be assembled into a complex search engine. The best way to do this is to construct a relational database of medical facts. All symptoms need to be identified individually for every condition and sorted separately. All lab tests need to be separated and linked to symptoms and diseases. All words need to be searchable and able to be sorted by multiple criteria. All references need to be attached to individual facts and separated from each other. Facts need to be given qualifiers like frequency of occurrence and clues on how they should be interpreted.

To make this new database possible you need a whole new structure and a new authoring tool that can enforce editing and authoring uniformity. This new authoring system needs to have quality control features and programs to check spelling and content integrity. It also has to be automated so that the data collection process is highly efficient. This tool was constructed and now powers DxEM.

If you can create the database, the next task is to build an application that makes it useful. That program exists, and it is called DxEM.

DxEM operates in two forms: the editing version and the production version. What you see as a user of the application is the production version. What authors see is the editing version. We have an automated program that takes our editing database and populates our production system. This conversion process takes less than 30 seconds. This rapid updating means that we can republish our entire production database as often as needed without any additional expense. No paper, no shipping, no delay. With each update you get the most current snapshot of all aspects of medical diagnostics. Instead of months or years of delay, current knowledge and wisdom can transfer instantaneously and be there when you need it.

Beyond the original goal of creating an intelligent assistant to astronauts, part of our charge was to build an intelligent electronic medical record system (EMR) that could travel into space and oversee astronaut day-to-day healthcare. We actually did this using available technology. At that time there were no cell phones, no laptops, and no Internet. Our early solution required mainframe computer support and was not able to transition into space. We did build this intelligent EMR and it was highly successful. With the termination of the NASA project our EMR needed to find a new home. It was eventually sold by UF to a nationally known medical computer company.

To update this story, DxEM has now been installed in a new intelligent medical record system for primary care and small hospital applications. Our testing and validation process of this new system has shown that we can save up to 60% of physician data entry time in routine charting. In addition, the clinical support information in DxEM can be delivered within the medical record so that physicians get current data at every stage of their critical order and planning process. We can now demonstrate a closed system from knowledge creation to knowledge delivery that can be reduced from months and years down to days and hours. By breaking the barrier on knowledge transmission and directly reaching the patient care frontier, we can truly see a revolution in healthcare delivery that goes way beyond continuing medical education. In industry they use a term called JIT (just-in-time). Stores scan product codes at checkout so that the JIT delivery system can have stock on the shelves when you arrive. In medicine, we use POC (point-of-care) to describe technology that is at the patient’s bedside for immediate application. DxEM’s EMR provides JIT knowledge delivery at the POC. This is a quantum leap that uses computer speed and accuracy for the patient’s immediate benefit.

Dr. Ralph Grams was the author and project manager of the original NASA research grant from the College of Medicine at the University of Florida (UF). He was Professor of Pathology and Director of the Medical Systems Division for most of his career. He retired in 2015 after 42 years at the University of Florida College of Medicine. He now devotes full time to KQI so that this technology can become commercially available. He is the sole author and editor of this vast array of medical data and has used modern technology to amplify his authoring speed so that a large professional staff is not required. The continuity and cohesion of all the software design and database changes have come by testing and validating each step of the process. Multiple versions of this database have been tested until an ultimate design was reached. In addition, multiple program changes have occurred over the years until we have our current cloud-based platform. The focus has been on distilling the best clinical data into the shortest packets of knowledge that can be clinically useful. This is not a book but a database of highly structured medical facts. Every fact has multiple references to support its authenticity. No personal biases or editorial comments are allowed. Everything must be traceable to its parent sources. This system must stand the legal test of authenticity and the ability to defend its output based on sound clinical logic which physicians can support and duplicate.