MULTISTAGE VIRTUALLY EXPERIENCED PATIENT DECISION MAKING

Purpose: Create an interface to represent a multistage decision process, including choice points, decision points, and outcomes, and to facilitate the communication of probabilities through an extension of the research paradigm known as Decision from Experience

Method: We have developed a novel patient decision aid in the form of an interactive animated decision tree.. This activity is enabled through controls which activate individual trials similar to a Monte Carlo simulation. Modes include single stepping though each branch of the tree or through entire paths from the root to probabilistically determined outcomes. Counts of all outcomes reached are displayed numerically as decimal probabilities as well as automatically recalibrated and redrawn horizontal bar charts, both intended to lower the cognitive load on patients by serving as memory devices. Other modes of interaction include autonomous stepping through collections of single paths, allowing observation by the user, and that of displaying all probabilities and bar charts after running 1,000,000 trials, the duration of which is approximately one thousand milliseconds, depending on browser and platform. The tool is implemented in Java, and can be hosted on a web site and runs in all browsers with Java virtual machines at least supporting jre 1.4. Definition of the tree is enabled through a custom XML grammar created for representing animated decision trees. The seed for the pseudo-random number generator may be supplied in order to produce identical runs when required.

Result: Test users have provided useful feedback. The visual interface of the tree shown uses standard decision tree symbols; therefore, trees look familiar to physicians educated in reading decision trees. Although designed primarily for patients, clinical personnel who are board-certified in internal medicine and gerontology have suggested application of the tool as a decision aid by medical staff as well. Pilot studies have begun but the current paucity of results does not yet justify statistical analysis.

Conclusion: no conclusions can yet be drawn.