Differential Diagnosis in Evidence-Driven Clinical Decision Support 

In the medical diagnosis process, a medical provider forms a differential diagnosis using the information garnered from a conversation with the patient, physical examination, and laboratory test results.

The information shared between the diagnostic indicators represents a source of statistical noise that can significantly reduce the accuracy of the resulting differential diagnosis. In practice, modern differential diagnostic systems fail to provide a timely informed differential diagnosis because they lack the ability to efficiently process “noisy” data.

Diagnostic “Noise”

Due to inherent time constraints and the emerging availability of electronic medical record (EMR) data, health care providers are readily employing clinical decision support systems (DSS) to assist in the diagnostic process. Southwest Research Institute® (SwRI®) is committed to developing technology that reduces diagnostic error in clinical decision support (CDS), telemedicine, medical training, and biosurveillance.

Grid Computing Architecture

Clinical Decision Support Prototype

SwRI engineers have developed a novel medical software algorithm capable of automating the interpretation, simplification and noise reduction of diagnostic data for the clinician, thus requiring minimal human intervention to produce an accurate differential diagnosis.

 A software prototype serves as a testing and demonstration vehicle for the new software technology. The prototype is a rich, visual DSS developed using the Java® programming language, R statistical package, MySQL® database management system, and Swing® graphical toolkit.

Diagnostic Knowledge Base

The diagnostic knowledge base for the software prototype was constructed using the Centers for Disease Control (CDC) National Ambulatory Medical Care Survey (NAMCS) dataset and relational database technology. The three major knowledge “focus” areas are:

• Diagnostic Data – Diagnoses, causes and cases (i.e., diagnostic records)
• Empirical Data – Model variables, principal components and regression equations
• Heuristic Data – General disease statistics and population statistics

Diagnosis	Multivariate Analysis
Symptoms	Analysis Results

Explanatory User Interface

SwRI engineers have developed a rich explanatory model to better communicate how clinical signs and symptoms influence the diagnostic selection model. The technology will help both medical students and experienced medical providers reach a more sound diagnosis within the time constraints of a patient consultation.

High-Performance Computing

The software prototype is designed to execute as a high-performance computing grid built on the JavaSpaces™ distributed computing platform. The grid solution implements an efficient read/write/take architecture to support the rapid uptake of new diagnostic cases in the diagnostic knowledge base.

Decision Support Capabilities

The software prototype implements a multivariate analysis algorithm to assist medical providers in rapidly generating a differential diagnosis, referring a patient to a medical specialist, and qualifying a patient for hospital admission. This novel software algorithm is capable of:

• Analyzing over one million diagnostic cases
• Reducing diagnostic noise and simplifying the prediction model
• Building a multivariate prediction model for over one hundred diagnoses
• Computing the diagnosis odds for each analyzed model
• Effectively communicating the rationale for a diagnosis recommendation

Automation and Data Systems Division Brochures
SwRI Brochures SwRI Publications SwRI Home