Richard A. Bankowitz

A computer-assisted medical diagnostic consultation service. Implementation and prospective evaluation of a prototype.

STUDY OBJECTIVE To evaluate the accuracy of a computer-aided consultation service using academic general internists and the Quick Medical Reference (QMR) diagnostic program: and to study the impact of the consultation on the diagnostic behavior of physicians caring for patients. DESIGN Prospective study of the diagnostic accuracy of computer-aided consultation in 31 cases, as well as a prospective study of ward team diagnoses and opinions before and after consultation. SETTING General medicine wards of two tertiary care centers. PARTICIPANTS Thirty-one patients identified as posing a diagnostic challenge and meeting eligibility criteria, as well as the housestaff caring for these patients. MEASUREMENTS AND MAIN RESULTS After 6 months follow-up, diagnoses were established in 20 of 31 cases. The diagnostic sensitivity of the computer-assisted diagnoses, 85% (95% CI, 56% to 97%), was similar to that of the consult service physicians, 80% (95% CI, 55% to 94%), but better than that of the ward teams, 60% (95% CI, 33% to 81%; P = 0.03 using the binomial test). The consultation influenced the postconsultation differential diagnoses of the ward teams in 26 of the 31 cases (95% CI, 92% to 95%). House officers rated the consultation service as being educationally helpful in 25 of the 31 cases (95% CI, 62% to 94%). CONCLUSIONS Computer-aided diagnostic consultation, when provided by physicians familiar with the limitations of the system and capable of overriding inappropriate suggestions, was both accurate and educationally helpful in most cases. The system provided reasonable diagnostic suggestions not previously considered by the ward teams and these suggestions were valued sufficiently to cause alteration of the original differential diagnoses.

Information Needs of Health Care Professionals in an Aids Outpatient Clinic as Determined by Chart Review

OBJECTIVE To examine the information needs of health care professionals in HIV-related clinical encounters, and to determine the suitability of existing information sources to address those needs. SETTING HIV outpatient clinic. PARTICIPANTS Seven health care professionals with diverse training and patient care involvement. METHODS Based on patient charts describing 120 patient encounters, participants generated 266 clinical questions. Printed and on-line information sources were used to answer questions in two phases: using commonly available sources and using all available medical library sources. MEASUREMENTS The questions were divided into 16 categories by subject. The number of questions answered, their categories, the information source(s) providing answers, and the time required to answer questions were recorded for each phase. RESULTS Each participant generated an average of 3.8 clinical questions per chart. Five categories accounted for almost 75% of all questions; the treatment protocols/regimens category was most frequent (24%). A total of 245 questions (92%) were answered, requiring an average of 15 minutes per question. Most (87%) of the questions were answered via electronic sources, even though paper sources were consulted first. CONCLUSIONS The participating professionals showed considerable information needs. A combination of on-line and paper sources was necessary to provide the answers. The study suggests that present-day information sources are not entirely satisfactory for answering clinical questions generated by examining charts of HIV-infected patients.

A temporal analysis of QMR.

Objective : To understand better the trade-offs of not incorporating explicit time in Quick Medical Reference (QMR), a diagnostic system in the domain of general internal medicine, along the dimensions of expressive power and diagnostic accuracy. Design : The study was conducted in two phases. Phase I was a descriptive analysis of the temporal abstractions incorporated in QMRs terms. Phase II was a pseudo-prospective controlled experiment, measuring the effect of history and physical examination temporal content on the diagnostic accuracy of QMR. Measurements : For each QMR finding that would fit our operational definition of temporal finding, several parameters describing the temporal nature of the finding were assessed, the most important ones being: temporal primitives, time units, temporal uncertainty, processes, and patterns. The history, physical examination, and initial laboratory results of 105 consecutive patients admitted to the Pittsburgh University Presbyterian Hospital were analyzed for temporal content and factors that could potentially influence diagnostic accuracy (these included: rareness of primary diagnosis, case length, uncertainty, spatial/causal information, and multiple diseases). Results : 776 findings were identified as temporal. The authors developed an ontology describing the terms utilized by QMR developers to express temporal knowledge. The authors classified the temporal abstractions found in QMR in 116 temporal types, 11 temporal templates, and a temporal hierarchy. The odds of QMRs making a correct diagnosis in high temporal complexity cases is 0.7 the odds when the temporal complexity is lower, but this result is not statistically significant (95% confidence interval = 0.27–1.83). Conclusions : QMR contains extensive implicit time modeling. These results support the conclusion that the abstracted encoding of time in the medical knowledge of QMR does not induce a diagnostic performance penalty.

Heuristic determination of quantitative data for knowledge acquisition in medicine

Knowledge acquisition for medical knowledge bases can be aided by programs that suggest possible values for portions of the data. The paper presents an experiment which was used in designing a heuristic to help the process of knowledge acquisition. The heuristic helps to determine numerical data from stylized literature excerpts in the context of knowledge acquisition for the QMR medical knowledge base. Quantitative suggestions from the heuristics are shown to agree substantially with the data incorporated in the final version of the knowledge base. The experiment shows the potential of knowledge base specific heuristics in simplifying the task of knowledge base creation.

The Role of Diagnostic Uncertainty in Health Care Outcomes and its Potential for Modification with Decision Support

INTERNIST-1 and its successor Quick Medical Reference (QMR), are computer programs which utilize an extensive knowledge base of diseases in internal medicine to assist clinicians in the task of diagnosis. The provision of medical care, in general, and the utilization of the diagnostic laboratory, in particular, are known to exhibit wide variation among physicians. A neglected reason for some of this variation may be the degree of diagnostic uncertainty accompanying each patient case. We propose that computer programs like QMR may decrease some of this variability by providing more information to the physician, and we discuss a methodology for testing this proposal.