A decision-driven system to collect the patient history

Area of Science:

• Medical Informatics
• Pulmonary Medicine
• Artificial Intelligence in Healthcare

Background:

• Hospitalized patients with pulmonary disease require accurate and efficient diagnostic histories.
• Traditional paper-based questionnaires can be lengthy and may not optimize question selection.
• Computerized systems offer potential for improving the efficiency of medical history taking.

Purpose of the Study:

• To develop and evaluate a computer-administered history system for hospitalized patients with pulmonary disease.
• To compare the diagnostic accuracy and efficiency of a decision-driven system versus a paper questionnaire.
• To assess the utility of a cognitive model and Bayesian scoring for question selection in medical diagnosis.

Main Methods:

• Development of a frame-based decision system with a cognitive model for question selection.
• Implementation of a Bayesian scoring algorithm to generate differential diagnostic lists.
• Comparison of the decision-driven system with a standard paper questionnaire in terms of questions asked and diseases captured.
• Evaluation of the number of questions asked (mean +/- SD) and the percentage of pulmonary diseases captured.

Main Results:

• The decision-driven system asked significantly fewer questions (50.7 +/- 31.0) compared to the paper questionnaire (182 questions).
• Both systems captured a similar percentage of pulmonary diseases (74% for the decision-driven system vs. 75% for the paper questionnaire).
• The decision-driven system identified 74% of 61 pulmonary diseases, while the paper questionnaire identified 75% of 85 diseases.

Conclusions:

• Computerized diagnostic knowledge bases can direct question selection to reduce the number of questions needed for a diagnostically useful patient history.
• A decision-driven approach enhances the efficiency of pulmonary disease diagnosis in hospitalized patients.
• This technology has the potential to streamline the medical history-taking process, improving patient care and clinician workload.