A diagnostic policy specifies what test to perform next, based on the results of previous tests, and when to stop and make a diagnosis. Cost-sensitive diagnostic policies perform tradeoffs between (a) the cost of tests and (b) the cost of misdiagnoses. An optimal diagnostic policy minimizes the expected total cost. We formalize this diagnosis process as a Markov Decision Process (MDP). We investigate two types of algorithms for solving this MDP: systematic search based on AO* algorithm and greedy search (particularly the Value of Information method). We investigate the issue of learning the MDP probabilities from examples, but only as they are relevant to the search for good policies. We do not learn nor assume a Bayesian network for the diagnosis process. Regularizers are developed to control overfitting and speed up the search. This research is the first that integrates overfitting prevention into systematic search. The paper has two contributions: it discusses the factors that make systematic search feasible for diagnosis, and it shows experimentally, on benchmark data sets, that systematic search methods produce better diagnostic policies than greedy methods.

This paper studies the problem of learning diagnostic policies from training examples. A diagnostic policy is a complete description of the decision-making actions of a diagnostician (i.e., tests followed by a diagnostic decision) for all possible combinations of test results. An optimal diagnostic policy is one that minimizes the expected total cost, which is the sum of measurement costs and misdiagnosis costs. In most diagnostic settings, there is a tradeoff between these two kinds of costs. This paper formalizes diagnostic decision making as a Markov Decision Process (MDP). The paper introduces a new family of systematic search algorithms based on the AO* algorithm to solve this MDP. To make AO* efficient, the paper describes an admissible heuristic that enables AO* to prune large parts of the search space. The paper also introduces several greedy algorithms including some improvements over previously-published methods. The paper then addresses the question of learning diagnostic policies from examples. When the probabilities of diseases and test results are computed from training data, there is a great danger of overfitting. To reduce overfitting, regularizers are integrated into the search algorithms. Finally, the paper compares the proposed methods on five benchmark diagnostic data sets. The studies show that in most cases the systematic search methods produce better diagnostic policies than the greedy methods. In addition, the studies show that for training sets of realistic size, the systematic search algorithms are practical on todays desktop computers.