Laboratory information processing systems

The dissertation research is concerned with Information Processing Systems in laboratories in fields such as medicine and microbiology. The processing in such laboratories is unstructured and existing information modeling techniques cannot not be used. An initial attempt was made by the development of a Laboratory Information Processing System (LIPS) at the University of Michigan in 1988. Its goal was to provide computerized support for studies on proteins that generate gels. A gel is a two dimensional object derived from a tissue cell using a process called 2-D Electrophoresis, that consists of a pattern of spots. Each spot on a gel represents a protein. A computer readable image is obtained from a gel and analyzed. By comparing spots across gels, researchers are able to identify the causes of the malfunction of a protein and prescribe the necessary cures. Mechanisms were needed so that researchers could verify their hypothesis at various stages of the study without having to wait for the study be completed. Results and conclusions had to be stored in a form so that researchers can compare results across studies. LIPS was implemented using a relational database networked on Sun workstations and personal computers. Experience with the present system has exposed limitations resulting from current information technologies such as the use of a relational model which limits the system's capability to model and store complex data. A generic model known as LIM (Laboratory Information Model) that unifies Data, Rules, and Interfaces is proposed. The prototype of LIM model uses emerging object-based technologies in databases (GemStone ODBMS), user interfaces (ObjectBuilder/C++ GUI), programming languages (C++) and software engineering. The prototype demonstrates the value of such technologies to substantially reduce the large amount of code normally required to build such systems. The dissertation provides a next-generation architecture of Laboratory Information Processing Systems and a generic Information Model for medical laboratories. The design methodology and the models are generic, i.e., they can be applied to situations in other fields that have large amounts of complex data that needs analysis for unstructured hypothesis.