Summary of Expert Systems in the Nursing Domain

This summary on the nursing domain, although clearly stated by the author, Mr. Keith Darlington, characterizesfurther the concepts of expert systems rather than proving the application of the concepts in helping the domain, in my opinion. Nevertheless, some rare possibilities surfaced in an effort by the author to prove the relationship and effectiveness of expert systems and the nursing sphere. Perhaps, with all rights due to him, Mr. Darlington intentionally and successfully outlined the basic concepts of expert systems.

First, the author mentioned the Inference Engine which is an established set of rules to be tested based on certain conditions. If a certain condition is true after a series of questions, then a specific result is set to be triggered; else, do otherwise. The user communicates with the system via the user interfacethat can be manipulated with the use of a mouse, the keyboard, light pen, touch-sensitive screen, and voice input. While the expert system will help the nurse in answering questions through the user interface, it will also provide mechanisms for the nurse to ask questions as well. Because patients who are diagnosed with the same sickness may experience different level of comforts or pain, it is good practice to incorporate an uncertainty degree that will provide more flexibility to the nurse in providing answers to the system. Thus, by generating certainty factors of 0 to 100, where 50 is the median, is a good mechanism to measure certainty levels where as the 0 mark indicates no certainty, and the 100 mark indicates a high degree of certainty.

While such backward chaining mechanism can be useful not only in the nursing domain, it is also a known fact that it thrived in other areas that use the same heuristic approach. To determine, however, whether a system will succeed or not, the telephone test is a great tool. That is if a human expert can solve a problem over the telephone, equally, the system will too. Otherwise, the system will most likely fail. As a result, it is judicious for a computer scientist, or any team of computer scientists to employknowledge engineering prior the manufacture of any expert system. The aforementioned technique is accomplished in two steps. One has to acquire the relevant data and enter them in the knowledge base for which the system is being fabricated.

Obtaining the relevant data, yet, requires knowledge acquisition which is attainable through researchand interviewingthe expertsin the field. For instance, there are three knowledge sources in the nursing field. The clinical and literature data which can mainly be obtained through research, and the expert data which can be obtained by the most famously used knowledge acquisition which is interviewing. However, it may quiet difficult to gather the needed information from the nursing experts themselves because they me be either unmotivated, or not having the time it requires for them to be interviewed. Of course to successfully conduct the interviews one must be thoroughly prepared and collect as much as accurate information as possible.

Once the information needed is assembled, the computer scientist(s) need to figure out what kind of tools will be needed to build the system. Programming languages such as C and Pascal, and AI languages such as LISP and Prolog are among the diverse languages used to build expert systems. Nevertheless, currently, expert system’s shells are available for both computer scientists and non-computer scientists like skilled nurses as apparatus to create expert systems. The shells are liquidated expert systems from their knowledge base. They are non-flexible, but useful.

A variety of the abovementioned shells come equipped with knowledge acquisition tools. This is done by an induction engine that reads a set of rules and tries to generate rules that are secured to the domain in question.

Here are some rules that a nursing expert system may consider.

  1. If an old man is a smoker, then he is a high risk patient.
  2. If a middle man is a non-smoker, then he is a low risk patient.
  3. If a young woman is a non-smoker, then she is a low risk patient.

Over the years, many expert systems such as Expert Nurse, CANDI (Computer AidedNursing Diagnosis and Intervention), and BABYthat monitors ICU(Intensive Care Unit) babies, have been developed to support nurses in their fields. While each one of such system may have a particular use, the concepts are comparable.

Reference:

Darlington K. Basic Expert Systems. ITIN 1996; 8.4:9-11.