Rymon, R. #1993#. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. thesis, The UniversityofPennsylvania.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....explanation seldom consider the e#ect of di#erent possible uses of explanations on which explanation to favor. The in#uence of intended uses for explanations on explanation generation has begun to be investigated in abductive diagnosis for tasks such as integrating medical diagnosis and response #Rymon, 1993#, and performing medical diagnosis within a planning framework #Turner, 1994#. The largest body of arti#cial intelligence research considering the role of goals in explanation, however, is in explanation based learning #EBL# #Mitchell, Keller, Kedar Cabelli, 1986; DeJong Mooney, 1986#. 8 ....

Rymon, R. #1993#. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. thesis, The UniversityofPennsylvania.

....performing simple observations and repair actions, assuming a most likely diagnosis. They do not limit their system to repair alone but rather generalize their goal to some notion of purpose; purpose does not include specification of diagnostic goals. Finally, and perhaps most notably, Rymon [Rym93] developed a goal directed diagnostic reasoner and companion planner, called TraumAID 2.0. The primary task of the diagnostic reasoner was to generate goals for the planner and to reason about whether those goals were satisfied. While the above work has some of the same goals as our work, none ....

R. Rymon. Diagnostic reasoning and planning in exploratory-corrective domains. PhD thesis, University of Pennsylvania, 1993.

....performing simple observations and repair actions, assuming a most likely diagnosis. They do not limit their system to repair alone but rather generalize their goal to some notion of purpose; purpose does not include speci cation of diagnostic goals. Finally, and perhaps most notably, Rymon [Rym93] developed a goal directed diagnostic reasoner and companion planner, called TraumAID 2.0. The primary task of the diagnostic reasoner was to generate goals for the planner and to reason about whether those goals were satis ed. While the above work has some of the same goals as our work, none has ....

R. Rymon. Diagnostic reasoning and planning in exploratory-corrective domains. PhD thesis, University of Pennsylvania, 1993.

....identifying the extent of malfunction, and in prescribing suitable repair. In these real world domains, diagnostic problem solving must involve reasoning about action and change. There have been some attempts at incorporating rudimentary actions into programs which reason to repair (e.g. 2] 3] [13], 16] Friedrich et al. 2] 3] have provided a procedure to choose between performing simple observations and repair actions, assuming a most likely diagnosis. Sun and Weld [16] present a diagnostic reasoner which calls a decision theoretic planner subroutine to plan repair actions. Their ....

R. Rymon (1993). Diagnostic reasoning and planning in exploratory-corrective domains. PhD thesis. Department of Computer Science, University of Pennsylvania.

....been little work which deals with the difficult knowledge representation issues associated with reasoning about action and change, and there is no formal specification of a framework or theory of diagnosis, testing and repair in such a context. The most notable contribution towards this goal is (Rymon, 1993) which details an architecture and system for diagnosis and repair which was applied to medical multiple trauma management. This paper proposes a situation calculus framework which casts diagnosis, testing and repair in the context of a theory of action and change. By formulating these notions ....

R. Rymon (1993). Diagnostic reasoning and planning in exploratory-corrective domains. PhD thesis. Department of Computer Science, University of Pennsylvania.

....with a discussion of an evaluation performed on TraumaTIQ s plan recognition algorithm and its implications for further system development. An Overview of TraumAID 2. 0 The TraumAID system is a tool for assisting physicians during the initial definitive management of patients with multiple trauma (Rymon 1993; Webber, Rymon, Clarke 1992) During this phase of patient care, which often requires urgent action, preliminary diagnoses are pursued and initial treatments are carried out. The current system, TraumAID 2.0, embodies a goal directed approach to patient management. The system architecture links ....

....esophagus repair and drain, while basing its choice of how to address Need Access Chest Cavity on the other currently relevant goals. Another feature of TraumAID s goal posting and planning is that its reasoner embeds a conservative, staged strategy for selecting diagnosis and treatment goals (Rymon 1993): goals whose satisfaction requires expensive and or risky procedures are not included in a plan until they are justified by less costly tests or observations. These strategies appear in the knowledge base as implicitly related management goals, such as a Diagnose Hematuria (blood in the urine) ....

[Article contains additional citation context not shown here]

Rymon, R. 1993. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. Dissertation, Department of Computer & Information Science, University of Pennsylvania. Appears as Technical Report MS-CIS-93-84.

....explanations, the use of multiple explanation strategies, or utility based considerations of how the cost of explanatory actions affect the explanation process. Goal based considerations have begun to be examined in research on diagnosis (e.g. Freitag and Friedrich, 1991; Redmond, 1992; Rymon, 1993; Turner, 1994) Such models of diagnosis include doing explanation as part of a cycle of diagnosis and repair (Rymon, 1993) and interleaving reasoning with action (Redmond, 1992; Turner, 1994) and making contextsensitive choices of what to diagnose first (Turner, 1994) A goal of GDIE is to ....

....actions affect the explanation process. Goal based considerations have begun to be examined in research on diagnosis (e.g. Freitag and Friedrich, 1991; Redmond, 1992; Rymon, 1993; Turner, 1994) Such models of diagnosis include doing explanation as part of a cycle of diagnosis and repair (Rymon, 1993) and interleaving reasoning with action (Redmond, 1992; Turner, 1994) and making contextsensitive choices of what to diagnose first (Turner, 1994) A goal of GDIE is to integrate all these aspects of diagnosis into a uniform framework, and to apply that framework to guide explanation and action ....

Rymon, R. (1993). Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. PhD thesis, The University of Pennsylvania.

....and execution of plans. The occurrence of a planning or execution failure triggers explanation. Given the goals of the planner the explanation system has to decide when to explain, what to explain and how to explain in a dynamic environment. Related to earlier work on goal directed diagnosis (Rymon 1993), a major issue being addressed in this research is how contextual changes influence the ongoing explanation process. As an example of the issues to address consider the following scenario implemented in our system. The planner is trying to achieve the goal of catching a plane and generates two ....

Rymon, R. 1993. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. Dissertation, University of Pennsylvania.

....seldom consider the effect of different possible uses of explanations on which explanation to favor. The influence of intended uses for explanations on explanation generation has begun to be investigated in abductive diagnosis for tasks such as integrating medical diagnosis and response (Rymon, 1993), and performing medical diagnosis within a planning framework (Turner, 1994) The largest body of artificial intelligence research considering the role of goals in explanation, however, is in explanation based learning (EBL) Mitchell, Keller, Kedar Cabelli, 1986; DeJong Mooney, 1986) 8 ....

Rymon, R. (1993). Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D.

....conclusions and goals to pursue from the available evidence about the patient, and a planner, that constructs a (partially ordered) plan for how best to address the 1 Currently, the system s knowledge base is restricted to penetrating injuries to the chest and abdomen. currently relevant goals [41, 49]. A more detailed discussion of the system appears in Chapter 4. In response to any new information regarding the patient or the management procedures carried out so far, TraumAID 2.0 outputs a listing of its current recommended management plan, which can be taken literally as orders for ....

....Unfortunately, however, the care given by even experienced trauma surgeons is often sub optimal, although these problems do not always have an effect on the patient outcome. Support for this claim comes from the analysis of data from a study evaluating the performance of TraumAID 2. 0 (see [9, 41]) This analysis suggests that the actual performance of physicians on real cases is not always acceptable to experts in the field of trauma surgery. When expert judges were asked to compare the management plans created by TraumAID 2.0 to the actual care given to patients, they rated the actual ....

[Article contains additional citation context not shown here]

Ron Rymon. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains (Appears as Technical Report MS-CIS-93-84). PhD thesis, University of Pennsylvania, 1993.

....critique is based on the user s intended actions it provides a user oriented focus for its output. This approach has been implemented in TraumaTIQ, the output module for TraumAID, a decision support system for the delivery of trauma care during the initial definitive phase of patient management (Rymon 1993; Webber, Rymon, Clarke 1992) TraumaTIQ is responsible for interpreting proposed actions in the context of the current state of the patient, and producing a critique of those actions. In the next section, we present the TraumAID system in more detail. Section discusses the sequence of plan ....

Rymon, R. 1993. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains (Appears as Technical Report MS-CIS-93-84).

....demands of the task, and the reluctance of certain kinds of users to accept advice from computers. We have attempted to minimize obstacles by providing support in the form of concise, relevant, user focused critiques. 1. 2 The Need for Decision Support for Trauma Management In his thesis, Rymon [75] presents a good argument for the usefulness of knowledge based decision support in the trauma domain. His work and the system he implemented, TraumAID 2.0, served as the starting point for my own work in this area. TraumAID was designed to help physicians during the first few hours after an ....

....Assurance (Q.A. by monitoring deviations from standards of care. The TraumAID system, which has been under development at the University of Pennsylvania for over ten years, is a tool for assisting physicians during the initial definitive management phase of patients with severe injuries [75, 92]. 1 During this phase of patient care, which is often characterized by the need for urgent action, preliminary diagnoses of the patients are pursued and initial treatments are carried out. The current system, TraumAID 2.0, embodies a goal directed approach to patient management. The system ....

[Article contains additional citation context not shown here]

Ron Rymon. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. PhD thesis, Department of Computer & Information Science, University of Pennsylvania, 1993. Appears as Technical Report MS-CIS-93-84.

....diagnostic and therapeutic activity into a single management plan, in the remainder of the paper, we will use the term management strategy rather than diagnostic strategy. Such strategies are embodied in TraumAID, the expert system to which TraumaTIQ serves as an interface [ Webber et al. 1992; Rymon, 1993 ] We will discuss some of TraumAID s strategies in the next section, in which we describe how TraumAID reasons and develops plans in trauma cases. 4 Planning in TraumAID This work is being done in the context of the TraumAID 1 project [ Clarke et al. 1993; Rymon et al. 1993; Webber et ....

....[ Webber et al. 1992; Rymon, 1993 ] We will discuss some of TraumAID s strategies in the next section, in which we describe how TraumAID reasons and develops plans in trauma cases. 4 Planning in TraumAID This work is being done in the context of the TraumAID 1 project [ Clarke et al. 1993; Rymon et al. 1993; Webber et al. 1992 ] whose overall goal is to improve the delivery of quality trauma care during the initial definitive phase of patient management. Initial definitive management encompasses a complex range of diagnostic and therapeutic procedures. Resolving often conflicting demands for ....

Ron Rymon. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. PhD thesis, Department of Computer & Information Science, University of Pennsylvania, 1993. Appears as Technical Report MS-CIS-93-84.

....monitored in order to possibly refine the target concept as explanation proceeds, based on information gathered during explanation. The use of explicit goals to direct the processing of our system is in the spirit of [Krulwich et al. 1990] and the more recent work on goal directed diagnosis (GDD) [Rymon, 1993]. Our approach differs from theirs in that we use a multi strategic framework and interaction with the environment is an integral part of our model. Also, in other approaches a single hypothesis is examined until it is either accepted or ruled out. Our approach involves the parallel utilitybased ....

Rymon, R. (1993). Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. PhD thesis, University of Pennsylvania.

....Shortliffe, 1984 ] for a sampling of some of these approaches) However, such models do not include the introspective reasoning about needs for information that is fundamental to goal driven learning. The interaction of task goals and explanation has been stressed by [ Leake, 1991; Leake, 1992; Rymon, 1993 ] and the role of interactive dialog in learning has been investigated by Redmond (1992) Research on expert system explanation has also considered the role of dialog in generating useful explanations for human questioners. For example, Moore and Swartout (1989) argue that expert system ....

Rymon, R. 1993. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. Dissertation, The University of Pennsylvania.

....Furthermore, in many diagnosis and repair domains multiple diagnostic and therapeutic needs require, and compete for, the agent s activity. In This work was supported in part by a graduate fellowship, ARO Grant DAAL03 89 C0031PRI, and by a National Library of Medicine grant 1RO LM051217 01. Rymon, 93] we propose an Exploratory Corrective Management (ECM) architecture (Figure 1) employing a basic cycle of diagnostic reasoning, planning and action. Reasoner Planner goals Physician evidence plan characterization TraumAID 2.0 Figure 1: The ECM Architecture In this architecture, diagnostic and ....

....asked to blindly compare the actual care in 97 trauma cases to the management that would have been proposed by TraumAID 2.0. The judges indicated a significant preference for TraumAID 2.0 plans over actual plans by a ratio of 64 to 17 with 16 ties (p 0. 001 by binomial test) For more details see [Rymon, 93] 2 Related Work 2.1 Formalizations of diagnosis Recent years have seen significant advances in formal approaches to diagnosis. A large number of approaches and frameworks have been suggested: probabilistic classifiers and discriminators, logical consistency based and abductive paradigms, ....

[Article contains additional citation context not shown here]

Rymon, R., Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph. D. Thesis. In preparation. Department of Computer and Information Science, University of Pennsylvania, 1993.

....concludes that the patient has a ureteral injury if he tests positive for one by IVP. 2 Because this architecture ties together diagnosis ( exploration ) and treatment ( correction ) and only allows the former to proceed to the extent that it can affect the latter, we have called it elsewhere [29, 31] an Exploratory Corrective Management or ECM architecture. 3 Other rules tend to be more complicated than these: see [34] 6375: UreteralInjury(SIDE=S) IVPUreteralInjury(SIDE=S,TEST RESULT= POSITIVE) 2. Goal setting rules map evidence and conclusions to goals (either diagnostic or ....

....diagnosed: 6370: ROUreteralInjury(SIDE= LEFT) Wound(WOUND TYPE= STAB,WOUND LOCATION= LLQ) Hematuria. 6376: RxUreteralInjury(SIDE=S) UreteralInjury(SIDE=S) Explicitly reasoning about goals allows us to encode a variety of useful strategies that improve the efficiency of TraumAID s plans [29]. For example, by explicitly reasoning about goals, we can allow the relevance of one goal to inhibit consideration of another one e.g. further investigation of a possible duodenal injury can be inhibited once the need for a laparotomy has been established for any other purpose, since knowing ....

[Article contains additional citation context not shown here]

Ron Rymon. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. PhD Dissertation. Dept. Computer & Information Science, University of Pennsylvania, November 1993. (Appears as Technical Report MS-CIS-93-84).

....7 A laparotomy is abdominal surgery. Goal Directed Diagnosis (GDD) a diagnostic reasoning framework which emphasizes diagnostic and therapeutic goals as the important aspect of solving diagnostic problems. The companion planner, and other aspects of TraumAID 2. 0 are discussed elsewhere [49, 50]. The paper is organized as follows: Section 2 contrasts our view of the goals of diagnosis with that of related formulations. To situate GDD s reasoning style, Section 3 overviews TraumAID 2.0 and its ECM architecture. Then, in the paper s main body, Sections 4 5 present the GDD formalism and ....

....concurrent diagnostic and therapeutic needs. Each consultation session in TraumAID 2.0 consists of several such cycles. In this paper, we focus on a formalization of TraumAID 2.0 s GDD based reasoning component. For description of other parts of the system, the interested reader is referred to [49, 50, 54]. TraumAID 2.0 was originally implemented in Common Lisp on a Symbolics platform, and was recently ported by Jonathan Kaye to a Sun based X windows environment as well as to a Macintosh platform. The new system currently consists of some 685 rules, out of which 184 are goal setting rules (cf. ....

[Article contains additional citation context not shown here]

Rymon, R., Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph. D. Thesis. Department of Computer and Information Science, University of Pennsylvania, 1993.

No context found.

Rymon, R. 1993. Diagnostic Reasoning and Planning in Exploratory-Corrective Domains. Ph.D. Dissertation, University of Pennsylvania.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC