The human organism is a highly complex system with an intricate network of various interrelated causes and effects. Diagnosis, prognosis, and therapeutic planning require a high level of medical expertise and experience. Due to the complex effects a disease exerts on the human organism, it is generally necessary to sift through of a lot of data in order to provide a good diagnosis and prognosis in any individual. Accordingly, the physician uses an extensive set of rules herein.

Mathematical modeling is often a complex, longwinded interactive process between experts from within the context of application and the mathematicians producing the model. Once a model is found it is evaluated and analyzed using mathematical methods. Nowadays the computer plays a large part in this. Without it, many of the models would be far too complex and therefore useless. In the evaluation, visualization is often the initial step for translation back into the specific context of application. This entire process is referred to as simulation—its computational part as scientific computing.

The diagnosis and therapy of diseases can be conceived as regulatory processes. In individual diagnosis, information on the actual disease state is achieved initially by assessing the medical history of and by questioning the patient, and also through a set of targeted procedures such as radiography and computed tomography or ultrasound. This information serves as the foundation upon which the physician, using his knowledge and experience will come up with a diagnosis and initiate therapeutic measures.

Expert systems can now be of manifold value in the diagnostic and therapeutic setting. A well-functioning expert system can guide doctors to the correct diagnosis by delivery of concrete suggestions for diagnosis, as well as supporting accumulation of the physician’s own experience. In other instances, a different group of doctors can play a part in the development of an expert system—knowledge can thereby be integrated and made objective.

Expert systems and data mining are pillars that provide the basis for computer-supported medical diagnosis and therapy, which has been gaining considerable importance in recent years. This novel medical field requires the intensive and interdisciplinary collaboration of physicians, mathematicians, computer specialists, physicists, and engineers. In addition to utilizing previously existing medical information more effectively, and learning from it through combination with medical expertise, expert systems could serve as integral components of a system that is yet to be constructed, consisting of interlaced medical data banks that would be provided online for physicians for diagnosis and therapy.

One example is the “Mycin” system developed in 1972, which helps the physician choose the type and dose of antibiotics. This system is made up of three main components:

•   Consultation component After insertion of patient data and data pertaining to, say, the bacterial types detected in the blood sample, the system proposes an antibiotic for therapy, including suggestions for dosage. It is crucial to understand that data fed into the system may be vague: in this case, a morphological description of the organisms found in the sample can be given. The system tries to identify these through combination with other information.

•   Explanatory component Following suggestion for therapy, it can be revealed which data and information was used.

•   Acquisition component This component enables new expert knowledge to be integrated into the existing system, or current information to be edited. Entering new information does not require any knowledge of programing, and can be done entirely by the expert.

The actual simulation of experts—the query of a specific situation within the area of expertise of the system—occurs through the consultation module. Following entry of data describing the situation, the interpreter analyses these, and compiles the rules needed, then applies these and finally displays the result in a user friendly manner.

Typically, the entered data are not precise. The interpreter must be capable of dealing with poorly delineated situations. Contrary to knowledge, strategies for entry-dependent collocation of the rules are permanently integrated within the system, or more precisely, in the interpreter, and cannot be altered by the user.

For various ends, it is necessary to fully understand the route from given input data to output data within the expert system: controllability of the system should be mentioned, as should applications in expert training or the test of newly to be integrated knowledge. A structured depiction of the routes from entry to output offers a module for reconstruction of decisions.

The use of expert systems in medicine does not differ much from that in other specialist areas. The following formulated areas of application can easily be transferred to other fields.

•   Relief for the doctor of performing routine tasks The expert system can take over the classification of patient data for diagnostic purposes, including concrete diagnostic suggestions for the physician. A further field of application is the choice of a specific therapy, oriented according to the individual circumstances of a particular patient.

•   Improved availability of medical expert knowledge An expert system can be used, among other things, as a specialist book or databank. The internet allows the functionality of expert systems to become available to the wider general public who in turn can use it to improve their health-related behavior and the surveillance of their own health.

•   Objectifying knowledge using formalized depictions The construction of an expert system requires clear and formal depiction of existing subjective knowledge on a particular circumstance. This enables knowledge to be comprehensible and traceable enabling full utilization of functionality of the expert system. The collaboration of multiple experts on a system makes this approach very effective.

•   Facilitation of validation of knowledge or rather verification of hypotheses.

Since the demands that medicine places in terms of complexity and operational reliability are considerably higher than in the technical–scientific area, a warning should be given: none of the above mentioned points should be misinterpreted to the extent that competency of a medical expert can be replaced by a computer program.