Biomedical subjects
Publications and source records attributed to A F Krieg.
During the past 20 years, laboratory systems have evolved to a high degree of complexity. In some cases, continuing addition of new features has adversely affected functionality. There is increasing recognition that the number of features listed on a request for proposal does not necessarily provide an accurate guide to functionality in day to day operations.
Complete evaluations of test performance require data on many test results over many clinical states, not restricted to the traditional 2 X 2 table of two possible test results and two possible clinical states. Published reports on test performance often include dot diagrams, depicting many test results over many clinical states. From such dot diagrams, several methods may be used to obtain numerical data for quantitative evaluations of test performance. Dot diagrams, long used to depict multiple test results over multiple clinical states, can serve as source documents for quantitative evaluations of test performance.
We suggest that evaluations of diagnostic tests start with dot plots that depict multiple test results over multiple clinical states. From this starting point we can calculate posttest probabilities for multiple clinical states at multiple test results. Also, we can project one subset of clinical states as "disease positive" and a second subset as "disease negative" to provide standard analyses such as likelihood ratios, relative operating characteristic curves, posttest/pretest probability plots, sensitivity, specificity, and predictive value. Finally, this starting point provides an excellent basis for comparing multiple studies of diagnostic performance. The advantages of dot plots are illustrated with data on serum ferritin levels over multiple clinical states.
We present a general spreadsheet model for evaluating diagnostic performance of clinical tests. Our model depicts test results as an r X c matrix, with r possible test results and c possible clinical states. Analysis of this matrix is based on the Ri/Cj ratio, calculated as a number of subjects having a specified result Ri within a given clinical state Cj, divided by total subjects within this clinical state. From this model, we can identify three special cases: (1) a 2 X c matrix, with two possible test results of T+ or T-, over c possible clinical states; (2) an r X 2 matrix, with r possible test results, over two possible clinical states of D+ or D-; and (3) a 2 X 2 matrix, with two possible test results over two possible clinical states. Application of the Ri/Cj ratio to the r X c matrix provides a useful approach to graphic analysis of multiple test results over multiple clinical states. The Ri/Cj ratio also provides a general approach to Bayesian analysis, in which likelihood ratio, relative operating characteristic analysis, sensitivity, and specificity represent special cases or special applications.
Explore the source record for details and available documents.
The question "When is a diagnostic test result positive?" can be addressed by clinical decision analysis. We developed two simple decision tree models for selecting appropriate cutoff levels: a net utility model and a threshold model. These models have been incorporated in a software program for desktop computers. We believe it is important for investigators to provide raw data on test performance, for three reasons. First, these data can be used in simple decision tree models to identify "appropriate" cutoff levels. Second, they can be used to evaluate empiric cutoff levels or decision rules. Third, they can be used to evaluate optimal cutoff levels for detailed decision trees depicting specific clinical problems.
We developed a microcomputer program that provides a Bayesian model of diagnostic performance and a simple decision tree model of clinical utility. We have used this program to review diagnostic performance and clinical utility for proposed new services at our 360-bed university hospital. We believe that significant benefits can be achieved if medical journals report complete data on test performance. First, this allows physicians to perform their own evaluations of diagnostic performance. Second, this allows physicians to evaluate clinical utility using either standard decision trees or decision trees that reflect specific clinical problems.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
This article is the fourth in a five part series on laboratory data processing and computers. Preceding the article are enabling and performance objectives, defined by the Laboratory Computer Applications and Data Processing Committee, for the systems analysis and planning competency area. Following the article are questions with which readers can test their understanding of the material.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
A computerized reporting and information system for microbiology employing a relatively inexpensive microcomputer is described. A comprehensive approach to accessioning and result entry for microbiology is presented. A daily laboratory worklist is generated for each work area, providing the responsible technologist with information on previously processed specimens. Manipulation of patient and specimen information permits the performance of various functions, including the generation of billing reports, workload statistics, quality-control summaries, epidemiologic surveys, and cumulative reports. The employment of many user-definable data lexicons allows optimal use of disk space while affording rapid information retrieval. Data file maintenance is automatically accomplished by the system, requiring no user intervention.
The authors of this article review briefly their original classification for clinical laboratory data (February 1979), which they proposed as an aid in analyzing and solving data processing problems, and recommend some extensions based on recent experience.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
This article is the second in a five-part educational series on laboratory data processing and computers. The authors describe the major components of peripheral storage and the limitations of input/output devices, and analyze in terms of laboratory workflow the type and placement of input/output devices. Preceding the article are competency areas and performance objectives, defined by the Laboratory Computer Applications and Data Processing Committee. Following the article are questions with which readers can test their understanding of the material. Next month the authors will focus on software.
Explore the source record for details and available documents.