Lynn L. Peterson

Temporal pattern discovery in course-of-disease data

Describes a system that utilizes event-set sequencing for knowledge discovery within a database of human immunodeficiency virus (HIV) patients. The authors use the development of their temporal pattern discovery system (TEMPADIS) to help understand the overall process of knowledge discovery in a medical database environment. Results are presented for a database of HIV patients. The authors discuss the overall process of knowledge discovery including data cleaning and preprocessing, data reduction and projection, and matching goals to a particular data mining method.

Medical information systems: characterization and challenges

This paper examines the characteristics and challenges presented by medical databases and medical information systems. It begins with a survey of medical databases/information systems. This is followed by a list of challenges for database management systems generated by the needs of these systems. It concludes with a look at some systems which address these challenges. In the context of this background information, the database community is asked to consider whether the results of database research are reaching those who are making day-to-day decisions regarding design and implementation of medical information systems.

Artificial neural networks: an emerging new technique

The artificial neural network is at the heart of an emerging technique, which many academicians and practitioners are using very productively due to its high performance in addressing complex problems. Although the ANN has not yet reached its full potential, the technique has demonstrated the capability of enhancing performance in a broad range of problems. This article presents an overview of the artificial neural network, a taxonomy of its learning paradigm, and its application areas.

Development of a case-based expert system: Application to a service coordination problem

Abstract Researchers have recently begun utilizing the case-based reasoning (CBR) technique in the construction of expert systems. Instead of developing a knowledge base that contains explicit rules, CBR involves developing a case with prior cases or example. Retrieving prior cases relevant to the current problem and deciding on a solution on the basis of the outcome of previous cases constitute the use of such an expert system. A major advantage of CBR is that in domains that already have much of the required knowledge in the form of cases a case-based expert system can be easily developed. The purpose of this article is to illustrate the CBR technique as applied to the problem of emulating the decision process of service coordinators. This article describes the architecture of a case-based expert system and the detailed methodology of applying the CBR technique. The benefits and pitfalls of applying the technique in the development of an expert system are also discussed.

Automatic generation of a knowledge-base for a dermatology expert system

A connectionist expert system (ES) in the domain of dermatology which will ultimately be used for instruction of medical students in the diagnosis of papulosquamous skin diseases is presented. The ES performs as an apparatus for diagnosis and elucidates the effects of diagnostic parameters in the derivation of viable conclusions. The ES uses the back-propagation algorithm to automatically generate a knowledge base from patient cases, each consisting of a list of symptoms as input and the experts diagnosis as the desired output. The knowledge base has the form of a multilayered network, and knowledge is distributed over the network so that a implicit specification of decision criteria. The ES uses this knowledge base for diagnosing a patients disease given a list of symptoms.<<ETX>>

An active hypertext path mechanism in a cooperative problem solving environment (abstract only)

As a concept, hypertext was first described over forty years ago; however it has only been in recent years that technology has permitted the im.plementation of actual systems. Hypertext is represented by a network of nodes and links which are traversed by a reader selecting designated keywords or icons located in each node. The result of such a traversal is dependent on the content of the destination node which could conceivably be text, audio, video, or even executable code.

A cooperative software development paradigm for content analysis-based psychology research procedures (abstract)

Results from a pilot study of the Worst Event Test (WET), a computer-aided content analysis-based procedure developed jointly by members of The University of Texas Southwestern Medical Center at Dallas Department of Psychiatry and members of The University of Texas at Arlington Department of Computer Science Engineering, suggest that a semi-automated content analysis is feasible for studying the relationship of cognitive distortion to depression, and is preferable to manual procedures. The cooperative, evolutionary software development approach which emerged from the WET may be seen as a paradigm useful in assisting in the development of other initially ill-defined psychology research procedures, and has the potential to benefit psychology researchers who might find manual development tasks prohibitive in terms of time and cost. The development of the WET, a multi-step procedure containing an initially ill-defined information system as well as human reasoning processes to be determined and standardized, was divided into three phases: (1) achievement of component reliability, (2) system prototype testing, and (3) integration. Each phase corresponded to a known psychology research need, and determined both knowledge engineering and information system design activities required. Software design and development were carried on concurrently with cyclic psychology procedure development, facilitating vital interaction between psychology activities, knowledge engineering activities, and information system design. Rapidly prototyped software provided timely development assistance for the psychology researchers at each step, as well as software usable during system prototype testing. Following procedure validation, the kernel software will be integrated with other data-handling processes specified during high-level information system design to create a custom workstation environment for psychology researchers engaged in similar research.