Concept formation vs. logistic regression: predicting death in trauma patients.

This study compares two classification models used to predict survival of injured patients entering the emergency department. Concept formation is a machine learning technique that summarizes known examples cases in the form of a tree. After the tree is constructed, it can then be used to predict the classification of new cases. Logistic regression, on the other hand, is a statistical model that allows for a quantitative relationship for a dichotomous event with several independent variables. The outcome (dependent) variable must have only two choices, e.g. does or does not occur, alive or dead, etc. The result of this model is an equation which is then used to predict the probability of class membership of a new case. The two models were evaluated on a trauma registry database composed of information on all trauma patients admitted in 1992 to a Level I trauma center. A total of 2155 records. representing all trauma patients admitted for more than 24 h or who died in the Emergency Department, were grouped into two databases as follows: (1) discharge status of 'died' (containing 151 records), and (2) any discharge status other than 'died' (containing 2004 records). Both databases contained the same variables.

Extracting knowledge from large medical databases: an automated approach.

Tools which can uncover patterns in patients' records and then make predictions based on that knowledge are and will continue to be high priority in many medical informatics groups. These tools are impacting the performance of outcome studies by discovering patterns which can then be verified with standard statistical tools. This paper demonstrates INC2.5, a general classification system, as a tool for assisting physicians in the decision making process. INC2.5 gathers information from patient records and builds a decision tree which is used to assist physicians in predicting the outcome of new patients. The decision tree will also reveal any patterns which the system found in the data. Successful results of such a system can be used to enhance outcome studies as well as to spread clinical information to areas with fewer resources.

Concept formation vs. logistic regression: predicting death in trauma patients.

This paper discusses two classification models, one based on concept formation and the other using standard logistic regression. The models are first explained in some detail and then evaluated on the same population of trauma patients. The goal of both systems is to predict the outcome of those patients. The results are summarized and explained in terms of differing algorithms of the two models.

Information, intelligence, and interface: the pillars of a successful medical information system.

This paper addresses three key issues facing developers of clinical and/or research medical information systems. 1. INFORMATION. The basic function of every database is to store information about the phenomenon under investigation. There are many ways to organize information in a computer; however only a few will prove optimal for any real life situation. Computer Science theory has developed several approaches to database structure, with relational theory leading in popularity among end users [8]. Strict conformance to the rules of relational database design rewards the user with consistent data and flexible access to that data. A properly defined database structure minimizes redundancy i.e.,multiple storage of the same information. Redundancy introduces problems when updating a database, since the repeated value has to be updated in all locations--missing even a single value corrupts the whole database, and incorrect reports are produced [8]. To avoid such problems, relational theory offers a formal mechanism for determining the number and content of data files. These files not only preserve the conceptual schema of the application domain, but allow a virtually unlimited number of reports to be efficiently generated. 2. INTELLIGENCE. Flexible access enables the user to harvest additional value from collected data. This value is usually gained via reports defined at the time of database design. Although these reports are indispensable, with proper tools more information can be extracted from the database. For example, machine learning, a sub-discipline of artificial intelligence, has been successfully used to extract knowledge from databases of varying size by uncovering a correlation among fields and records[1-6, 9]. This knowledge, represented in the form of decision trees, production rules, and probabilistic networks, clearly adds a flavor of intelligence to the data collection and manipulation system. 3. INTERFACE. Despite the obvious importance of collecting data and extracting knowledge, current systems often impede these processes. Problems stem from the lack of user friendliness and functionality. To overcome these problems, several features of a successful human-computer interface have been identified [7], including the following "golden" rules of dialog design [7]: consistency, use of shortcuts for frequent users, informative feedback, organized sequence of actions, simple error handling, easy reversal of actions, user-oriented focus of control, and reduced short-term memory load. To this list of rules, we added visual representation of both data and query results, since our experience has demonstrated that users react much more positively to visual rather than textual information. In our design of the Orthopaedic Trauma Registry--under development at the Carolinas Medical Center--we have made every effort to follow the above rules. The results were rewarding--the end users actually not only want to use the product, but also to participate in its development.

Turning medical data into decision-support knowledge.

Advances in information collection and analysis are reaching the point of providing physicians with the help of computer-based assistants. These systems will provide rapid second opinions to physicians in a clinical setting as well as assist them in the analysis of large sets of patient descriptions for research purposes. This paper presents INC2.5 as such a decision-support system. INC2.5 extracts information from databases of previously seen patients to build a decision tree which is used to predict the outcome of new patients on a chosen variable. The concept of matching new patients with the most similar previously seen patient, on which INC2.5 is based, can be easily understood by its users. Further adding to INC2.5's ease of use is its flexibility in allowing users to customize decision trees to their liking. In order to convey the uncertainty of the environment, INC2.5 presents all decisions with a confidence factor.