Samuel Holtzman

Structuring conditional relationships in influence diagrams

An influence diagram is a graphical representation of a decision problem that is at once a formal description of a decision problem that can be treated by computers and a representation that is easily understood by decision makers who may be unskilled in the art of complex probabilistic modeling. The power of an influence diagram, both as an analysis tool and a communication tool, lies in its ability to concisely summarize the structure of a decision problem. However, when confronted with highly asymmetric problems in which particular acts or events lead to very different possibilities, many analysts prefer decision trees to influence diagrams. In this paper, we extend the definition of an influence diagram by introducing a new representation for its conditional probability distributions. This extended influence diagram representation, combining elements of the decision tree and influence diagram representations, allows one to clearly and efficiently represent asymmetric decision problems and provides an attractive alternative to both the decision tree and conventional influence diagram representations.

R&D Decision Advisor: An interactive approach to normative decision system model construction

Abstract This paper describes the architecture of R&D Decision Advisor ™ , a commercial intelligent decision system for evaluating corporate research and development projects and portfolios. In analyzing projects, R & D Decision Advisor interactively guides a user in constructing an influence diagram model for an individual research project. The systems interactive approach can be clearly explained from a blackboard perspective. The system guides the user to choose among general project features but offers flexibility to capture unique project details. Related research generalizes the underlying architecture and addresses resource-constrained decision situations.

Exact Reasoning about Uncertainty: On the Design of Expert Systems for Decision Support

This paper focuses on designing expert systems to support decision-making in complex, uncertain environments. In this context, our research indicates that strictly probabilistic representations, which enable the use of decision-theoretic reasoning, are highly preferable to recently proposed alternatives (e.g., fuzzy set theory and Dempster-Shafer theory). Furthermore, we discuss the language of influence diagrams and a corresponding methodology – decision analysis – that allows decision theory to be used effectively and efficiently as a decision-making aid. Finally, we use RACHEL, a system that helps infertile couples select medical treatments, to illustrate the methodology of decision analysis as a basis for expert decision systems.

Field testing as clinical trial methodology in periodontics.

In the early 1950s the randomized control trial (RCT) was introduced and became widely accepted as the definitive proof of efficacy of a specific medical treatment. In fact, the acceptance and application of this methodology were instrumental in converting medicine from an unpredictable art to a science. At present no other methodologies exist that allow the evaluation of therapeutic efficacy with confidence comparable to that achieved with randomized controlled trials. In recent years researchers have applied new experimental designs and data analysis techniques to clinical trials conducted in a field trial environment to facilitate the understanding of proper use of new therapeutic agents and procedures. Since many of the new methodologies are still evolving or have only recently been introduced, this review considers some of the major trends and developments, as well as experiences of the authors, in field trial methodology. This manuscript addresses the following questions: 1) Are there current clinical trial needs that are not met by RCT? 2) If so, what considerations are necessary for new approaches to have scientific usefulness? and 3) What are the strengths and weaknesses of the field trials setting relative to an institutional environment? J Periodontol 1992; 63:1064- 1071.

R&D analyst: an interactive approach to normative decision system model construction

This paper describes the architecture of R&D Analyst, a commercial intelligent decision system for evaluating corporate research and development projects and portfolios. In analyzing projects, R&D Analyst interactively guides a user in constructing an influence diagram model for an individual research project. The systems interactive approach can be clearly explained from a blackboard system perspective. The opportunistic reasoning emphasis of blackboard systems satisfies the flexibility requirements of model construction, thereby suggesting that a similar architecture would be valuable for developing normative decision systems in other domains. Current research is aimed at extending the system architecture to explicitly consider of sequential decisions involving limited temporal, financial, and physical resources.

Activity graphs: A language for flexible consultation systems

An intelligent decision system (IDS) uses artificial intelligence principles to deliver automated, interactive decision analysis (DA) consultations. Network methods adapted from operations research underlie two key IDS components: influence diagrams and activity graphs. Influence diagrams, which are familiar to DA researchers and practitioners, represent decision problems inevent space. Activity graphs, which are introduced in this paper, represent processes inaction space. While activity graphs can represent any process, we use them as a knowledge-engineering and programming language to represent the process knowledge of skilled decision analysts in the context of a specific class of decisions. This paper defines activity graphs as an extension of directed AND-OR graphs. Anactivity tree is a directed AND-OR tree consisting of nodes, which may contain activities (small computer programs) and connectors that establish logical relationships among nodes and define logical resolution agendas. Anactivity graph is a directed, multiply connected network of activity trees. Activity graphs may involve recursion. Development of the activity graph language is motivated by our desire to enable professional decision analysts — or other experts — with limited advanced programming experience to design and build consultation systems that combine the guidance offered by protocol systems with the flexibility and generality of transaction systems. This paper defines the activity graph language in detail. A simple example illustrates key concepts. The paper also discusses our experience using a computer system that implements activity graphs for developing commercial IDSs.