DEMONSTRATION OF A BALANCE BEAM AID FOR INSTRUCTION IN DIAGNOSIS

Background: Understanding the impact of clinical findings in discriminating between possible causes of a patient’s presentation is an essential component of clinical judgment, but students start with no intuitions on these issues. A balance beam offers a visual and physical analogue of the impact of a patient’s signs and symptoms' impact on competing disease hypotheses’ probabilities. We describe a Balance Beam Aid for Instruction in Diagnosis (BBAID) for training medical students to diagnose a common presentation.

Methods: A balance beam, with one of two competing diseases on each end, and with weights representing particular pieces of evidence placed at appropriate locations on the beam, is a natural physical analogue that can accurately represent the combination of several pieces of evidence with varying ability to discriminate between the disease hypotheses. Calculating Bayes’ Theorem using log(posterior odds) as a function of log(likelihood ratios) and log(prior odds) maps onto the physical forces of weights and helium balloons affecting a balance beam. 

Results:  Our working version of the BBAID for Acute Chest Pain presents a list of potential findings and allows the user to specify whether each is present, absent, or unknown. It displays the likelihood ratios corresponding to positive (LR+) or negative (LR-) observations for each symptom, for any pair of diseases. For each specified finding, a token is placed on the beam at a location whose distance from the fulcrum is proportional to the finding's log(LR), with a downward force (a weight) if the finding is present and an upward force (a balloon) if it is absent. A representation of the summary impact of all evidence is displayed. In a separate matrix display, the current probabilities of all diseases and their pairwise comparisons are shown. Clicking a cell in this matrix causes the corresponding pair's data to be displayed on the balance beam and LRs for that pair to be displayed next to the symptom names.

Conclusion: It is feasible to represent clinical diagnosis on a balance beam, providing a mathematically accurate display of the impact of clinical evidence upon the relative probabilities of two diseases. This holds promise as a tool for research, education, or clinical practice. It is limited by the quality of the input information (sensitivities for relevant symptoms and diagnoses). Its reliance on naïve Bayes' theorem (assuming that symptom impacts are independent) is the same as in other applications.