Shamsul Chowdhury

Extracting knowledge from a large primary health care database using a knowledge-based statistical approach

Clinical databases from automated medical records represent a growing resource for deriving new medical knowledge. In this study a large primary health care database was explored with respect to the association between hypertension and diabetes. Data collection was made with a query language, and data analysis performed with an interactive knowledge-based statistical tool, MAXITAB, employing a multivariate tabular analysis technique. In the study population of 6660 patients the prevalence of diabetes was almost three times higher for hypertensive patients than for those with no hypertension. Conversely, the prevalence of hypertension was 2.6 times higher for diabetic patients than for those with no diabetes. The results support the assumption of a relationship between hypertension and diabetes, although the question of causality between the two diagnoses remains unsolved. Knowledge-based statistical tools of this kind may be feasible for exploring large clinical databases and may result in new medical hypotheses, worthy of further investigation.

A statistically rule-based decision support system for the management of patients with suspected liver disease

The paper describes how a decision support system in liver diseases, mostly oriented to prediction of the necessity for liver biopsy, has been developed. The system designed is a hybrid one and consists of two parts: logical and statistical. The logical part contains rules, formulated on the basis of current medical knowledge, which enables recognition of clear cases; diseased or non-diseased. The unclear cases are classified on the basis of rules statistically extracted from databases. These rules have been reached after a comprehensive exploratory analysis of the sample of 165 patients with slightly to moderately raised levels of routine liver tests but without signs or symptoms of liver diseases. The extracted decision diagrams which simulate traditional medical diagnosis conduct have been found to be superior to discriminant analysis and probabilistic inductive learning. They use only a limited number of laboratory tests to detect the necessity for biopsy.

Design and delivery of information resources and knowledge bases for the diagnosis and management of liver disorders

The paper presents how a design and implementation of a system for the investigation of patients suffering from liver disorders were done. The system is developed on the Apple Macintosh using the Hypercard system. Hypercard is a general purpose application development tool based on the hypertext technique. The system consists of a collection of chunks of text (medical knowledge, patient management procedure, statistical rules etc.) linked appropriately. The emphasis is on allowing the users to get access to the information they needed. The contents of the system are text and active links. By simple navigation commands-such as merely clicking a mouse button, the users are allowed to move around in the information base and get hold of the information sought.<<ETX>>

Artificial Intelligence Methods in Data Analysis and Interpretation

Knowledge-based interfaces to existing statistical software are one way of applying artificial intelligence methods in data analysis and interpretation. In this paper we discuss our experiences from the implementations of knowledge-based systems for time series analyses, in particular for eliminating seasonal variations in statistical studies of industry and trade, and for multivariate tabular analysis. We further discuss the problems of realizing knowledge-based systems in a production environment and present an integrated conceptual model for knowledge-based data analysis and interpretation.

Extraction of Knowledge from a Database

Statistical packages are generally employed to derive or infer facts about a Universe of Discourse stored in a database. These packages are rather powerful concerning their computational capabilities, but they often lack knowledge about the statistical aspect of a computation. As such, they can not be fruitfully utilized by statistically naive users. We propose the use of AI techniques in order to provide intelligent interfaces, which would support this category of users in their statistical evaluations.