Lawrence M. Fagan

Use of a domain model to drive an interactive knowledge-editing tool

The manner in which a knowledge-acquisition tool displays the contents of a knowledge base affects the way users interact with the system. Previous tools have incorporated semantics that allow knowledge to be edited in terms of either the structural representation of the knowledge or the problem-solving method in which that knowledge is ultimately used. A more effective paradigm may be to use the semantics of the application domain itself to govern access to an expert systems knowledge base. This approach has been explored in a program called OPAL, which allows medical specialists working alone to enter and review cancer treatment plans for use by an expert system called ONCOCIN. Knowledge-acquisition tools based on strong domain models should be useful in application areas whose structure is well understood and for which there is a need for repetitive knowledge entry.

Episodic skeletal-plan refinement based on temporal data

ONCOCIN is a medical expert system that extends the skeletal-planning technique to an applciation area where the history of past events and the duration of actions are important. The systems knowledge base is designed to reflect a hierarchical model of the domain, and its control and inference mechanisms encourage a mixed-initiative style of interaction between the computer and the user.

The design and implementation of a ventilator-management advisor

VentPlan is an implementation of the architecture developed by the qualitative-quantitative (QQ) research group for combining qualitative and quantitative computation in a ventilator-management advisor (VMA). VentPlan calculates recommended settings for four controls of a ventilator by evaluating the predicted effects of alternative ventilator settings. A belief network converts clinical diagnoses to distributions on physiologic parameters. A mathematical-modeling module applies a patient-specific mathematical model of cardiopulmonary physiology to predict the effects of alternative ventilator settings. A decision-theoretic plan evaluator ranks the predicted effects of alternative ventilator settings according to a multiattribute-value model that specifies physician preferences for ventilator treatments. Our architecture allows VentPlan to interpret quantitative observations in light of the clinical context (such as the clinical diagnosis). We report a retrospective study of the ventilator-setting changes encountered in postoperative patients in a surgical intensive-care unit (ICU). We conclude that the QQ architecture allows VentPlan to apply a patient-specific physiologic model to calculate ventilator settings that are optimal with respect to a decision-theoretic value model describing physician preferences for setting the ventilator.

A Methodology for Generating Computer-based Explanations of Decision-theoretic Advice

Decision analysis is an appealing methodology with which to provide decision support to the practicing physician. However, its use in the clinical setting is impeded because computer- based explanations of decision-theoretic advice are difficult to generate without resorting to mathematical arguments. Nevertheless, human decision analysts generate useful and in tuitive explanations based on decision trees. To facilitate the use of decision theory in a computer-based decision support system, the authors developed a computer program that uses symbolic reasoning techniques to generate nonquantitative explanations of the results of decision analyses. A combined approach has been implemented to explain the differences in expected utility among branches of a decision tree. First, the mathematical relationships inherent in the structure of the tree are used to find any asymmetries in tree structure or inequalities among analogous decision variables that are responsible for a difference in expected utility. Next, an explanation technique is selected and applied to the most significant variables, creating a symbolic expression that justifies the decision. Finally, the symbolic expression is converted to English-language text, thereby generating an explanation that justifies the desirability of the choice with the greater expected utility. The explanation does not refer to mathematical formulas, nor does it include probability or utility values. The results suggest that explanations produced by a combination of decision analysis and symbolic processing techniques may be more persuasive and acceptable to clinicians than those produced by either technique alone. Key words: automated explanation; artificial intelligence; decision theory; decision support systems; medical informatics; stochastic simulation. (Med Decis Making 8:290-303, 1988)

Patient-specific explanation in models of chronic disease

Clinical models of chronic disease characteristically must represent significant uncertainty in both the data input and inferences. This lack of determinism makes it especially difficult for system users to understand and have confidence in the models. This paper presents a representation for uncertainty and patient preferences that serves as a framework for graphical summary and computer-generated explanation of patient-specific clinical decision models. The implementation described is a computer decision aid designed to enhance the clinician-patient consultation process for patients with suspected angina pectoris. The generic angina model is represented as a Bayesian decision network, where the patient descriptors, probabilities, and preferences are treated as random variables. The initial distributions for these variables represent information on the population of patients with anginal symptoms, and the approach provides a method for efficiently tailoring the distributions to an individual patient. This framework also provides metrics for judging the importance of each variable in the model. The graphical interface uses this information to augment the display of a network representation of the model. Variables that are important for clinician-patient communication are highlighted in the graphical display of the network and included in the text explanation in printed patient-education materials. These techniques serve to keep the explanation of the patients decision model concise, allowing the communication with the patient to focus on the most important aspects of the treatment decision.

TQuery: a context-sensitive temporal query language

Users of electronic medical databases request pertinent information by recasting their clinical questions into a formal database query language. Because the query language is the users only access to the data, the query language must be powerful enough to enable users to express their data requirements. However, a competing need is for the query language to be restrictive enough so that queries can have unambiguous semantics and the query processor can generate correct answers. We describe a query language, called TQuery , that was designed specifically to formulate database queries that are dependent on temporal and contextual relationships. TQuery specifications express contextual constraints without the need to explicitly reference calendar dates. TQuery is the database query language used to retrieve patient data from an object-oriented electronic patient medical-record system called the temporal network (TNET). TNET and TQuery were developed to support the real-time temporal reasoning and representation needs of a LISP workstation-based medical expert system.

Methods for Semi-automated Indexing for High Precision Information Retrieval

OBJECTIVE To evaluate a new system, ISAID (Internet-based Semi-automated Indexing of Documents), and to generate textbook indexes that are more detailed and more useful to readers. DESIGN Pilot evaluation: simple, nonrandomized trial comparing ISAID with manual indexing methods. Methods evaluation: randomized, cross-over trial comparing three versions of ISAID and usability survey. PARTICIPANTS Pilot evaluation: two physicians. Methods evaluation: twelve physicians, each of whom used three different versions of the system for a total of 36 indexing sessions. MEASUREMENTS Total index term tuples generated per document per minute (TPM), with and without adjustment for concordance with other subjects; inter-indexer consistency; ratings of the usability of the ISAID indexing system. RESULTS Compared with manual methods, ISAID decreased indexing times greatly. Using three versions of ISAID, inter-indexer consistency ranged from 15% to 65% with a mean of 41%, 31%, and 40% for each of three documents. Subjects using the full version of ISAID were faster (average TPM: 5.6) and had higher rates of concordant index generation. There were substantial learning effects, despite our use of a training/run-in phase. Subjects using the full version of ISAID were much faster by the third indexing session (average TPM: 9.1). There was a statistically significant increase in three-subject concordant indexing rate using the full version of ISAID during the second indexing session (p < 0.05). SUMMARY Users of the ISAID indexing system create complex, precise, and accurate indexing for full-text documents much faster than users of manual methods. Furthermore, the natural language processing methods that ISAID uses to suggest indexes contributes substantially to increased indexing speed and accuracy.

Collaborative conversational interfaces

This paper proposes a method of designing human-computer speech interfaces based on principles of human conversation. It argues that conversation is the primary mode of language use and that it is fundamentally collaborative. Speech interfaces should therefore be designed to recreate the collaborative nature of natural conversations. The paper presents five strategies for designingcollaborative conversational interfaces, and it describes the principles of human-language use that underly these strategies. The paper also argues that collaborative conversational interfaces have a crucial advantage over other kinds of interfaces in that they are readily adaptive to different levels of experience and styles of use. The paper gives examples of collaborative conversational interfaces that we have developed, and discusses the ways in which these interfaces have been made adaptive.

OPAL: Toward the Computer-Aided Design of Oncology Advice Systems

Creating the knowledge base of an expert system, as when developing any model, requires the abstraction of some reality. The important aspects of a problem area must be identified and extracted. The often difficult process of identifying, extracting, and representing those important domain aspects for use by an expert system is called knowledge acquisition. Successful knowledge acquisition is often considered the major obstacle in the construction of knowledge-based advice systems.1

Building a speech interface to a medical diagnostic system

A description is given of the design of an interface for QMR-DT, an evolving decision-theoretic version of Quick Medical Reference, which performs diagnostic reasoning about diseases in internal medicine. QMR-DT encompasses the part of QMRs functionality that provides differential diagnoses for a set of patient characteristics, but it uses a different algorithm to compute diagnoses. The work described includes two programs that integrate off-the-shelf speech technology with programs that manipulate medical terminology. The term identifier program uses an isolated-word, speaker-dependent speech product to provide an interface for entering medical findings into QMR-DT. Frame browser is an auxiliary program used by the developers of term identifier to examine frame structures. Frame browser was also used to experiment with a continuous-speech system.<<ETX>>