Abstract. This paper describes the student modeling component of ANDES, an Intelligent Tutoring System for Newtonian physics. ANDES' student model uses a Bayesian network to do long-term knowledge assessment, plan recognition and prediction of students' actions during problem solving. The network is updated in real time, using an approximate anytime algorithm based on stochastic sampling, as a student solves problems with ANDES. The information in the student model is used by ANDES' Help system to tailor its support when the student reaches impasses in the problem solving process. In this paper, we describe the knowledge structures represented in the student model and discuss the implementation of the Bayesian network assessor. We also present a preliminary evaluation of the time performance of stochastic sampling algorithms to update the network.

No abstract

Abstract. Objective: A computer-based system to apply trauma resuscitation protocols to patients with penetrating thoracoabdominal trauma was previously validated for 97 consecutive patients at a Level 1 trauma center by a panel of the trauma attendings and further refined by a panel of national trauma experts. The purpose of this article is to describe how this system is now used to objectively critique the actual care given to those patients for process errors in reasoning, independent of outcome. Methods: A chronological narrative of the care of each patient was presented to the computer program. The actual care was compared with the validated computer protocols at each decision point and differences were classified by a predetermined scoring system from 0 to 100, based on the potential impact on outcome, as critical/noncritical/no errors of commission, omission, or procedure selection. Results: Errors in reasoning occurred in 100% of the 97 cases studied, averaging 11.9/case. Errors of omission were more prevalent than errors of commission (2.4 errors/case vs 1.2) and were of greater severity (19.4/error vs 5.1). The largest number of errors involved the failure to record, and perhaps observe, bedside information relevant to the reasoning process, an average of 7.4 missing items/patient. Only 2 of the 10 adverse outcomes were judged to be potentially related to errors of reasoning. Conclusions: Process errors in reasoning were ubiquitous, occurring in every case, although they were infrequently judged to be potentially related to an adverse outcome. Errors of omission were assessed to be more severe. The most common error was failure to consider, or document, available relevant information in the selection of appropriate care.

No abstract