Abstract
Several diagnostic decision support systems have been proposed. Yet, such systems are not commonly used partly due to insufficient temporal reasoning. We have developed a clinical problem-oriented diagnostic decision support model that employs temporal reasoning by using extended temporal database functionalities developed at our lab. In addition, our model supports the workflow of the diagnostic process and employs detailed clinical data. We developed a prototype system that implements the concepts of the model.
Introduction
The diagnostic process of investigating a clinical problem includes processing of subjective information acquired by inquiring the patient (i.e. symptoms), objective findings obtained by performing physical examination (i.e. signs) and various laboratorial and imaging data. The number of diagnoses that might fit (i.e. differential diagnosis) should decrease as the process progresses. The temporal relationships between the clinical data are often critical for obtaining the most proper set of diagnoses. Several diagnostic decision support models and systems have been proposed and developed. However, inadequate temporal reasoning was recognized as a significant factor for their relatively low penetration1. Thus, the goal of this project was to develop and implement a decision support model for investigating clinical problems based on time oriented set of clinical data. In addition, we aimed to devise a model that follows and supports the workflow of the diagnostic process, and employs as much patient’s history and physical examination data as possible.
Methods
The temporal reasoning of the model is based on extended update functionalities in temporal databases developed at our lab. We also developed a non-Bayesian scoring scheme that incorporate semi-quantitative scales to express the probabilistic association of clinical data with particular diagnoses. The model was designed to enable usage of detailed patient’s history and physical examination data gathered in the diagnostic process, as well as lab and imaging data. Accordingly, the user interfaces were designed to follow the diagnostic workflow.
We developed a prototype system that implements the concepts of the model using Java™ (Sun microsystems), JDBC and MySQL database.
Results and Discussion
The system enables an expert administrator to set a weight for the significance of clinical data in a certain clinical problem domain, and to set a weight for the strength of temporal relationships between them.
The prototype system supports the user in gathering time oriented clinical data in consistent with the diagnostic workflow. The user can get a list of suggested differential diagnosis at any point in the process.
We believe that this problem-oriented model that incorporates temporal reasoning and is adjusted to the diagnostic workflow might be helpful in supporting physicians.
We will shortly conduct an evaluation study to examine the power of the system to support clinicians in investigating the clinical problems of Syncope and Jaundice.
References
- 1.Berner ES, Webster GD, Shugerman AA, Jackson JR, Algina J, et al. Performance of Four Computer-Based Diagnostic Systems. N Engl J Med. 1994;330:1792–6. doi: 10.1056/NEJM199406233302506. [DOI] [PubMed] [Google Scholar]