James E. Matheson

Influence Diagrams

Innuence diagrams are graphical models for structuring decision scenarios, particularly scenarios consisting of a predeened sequence of actions and observations. Innuence diagrams were originally introduced by 3] as a compact representation of symmetric decision trees 8] but they may also be thought of as extensions of Bayesian networks. This article is based on the article on Bayesian graphical models (referred to as BGM), and the reader is advised to read BGM before proceeding. UTILITIES The basis of innuence diagrams are probabilities and utilities. Utilities are quantiied measures for preference. That is, a real number is attached to each possible scenario in question. The beliefs in the scenarios are expressed as probabilities. (1) EXAMPLE The Bayesian network in BGM Figure 1 can for example be used to calculate the expected distribution for River ow given various observations. Assume that we can decide on a set of diierent types of Land use, and we have to balance the decision with the impact on River ow. This can be modelled by extending the Bayesian network framework with nodes for decisions and utilities: The Land use node is changed to a decision node, and we give River ow as well as Land use a diamond shaped child indicating that we attach utilities to these variables (see Figure 1). With this representation, the computer can easily calculate the impact from the various decisions and thereby the expected utilities that arrive from taking each decision. The user is advised to take the decision with highest expected utility.

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.

Influence Diagram Retrospective

Since the invention of Influence diagrams in the mid-1970s, they have become a ubiquitous tool for representing uncertain situations. This single diagram replaced awkward manipulations of decision trees and natures trees with a single representation that displays both the sequential and informational structure of decisions. The diagram permits high-level graphic communication, clear assessments and computation in a single graphical system. This retrospective discusses the evolution and application of influence diagrams.

Using Decision Quality Principles to Balance Your R&D Portfolio

OVERVIEW:Achieving a high return from your R&D investment cannot be accomplished simply by focusing on individual projects. To maximize your return, you need to make quality decisions at the portfolio level—ensuring the proper mix of high-risk, high-potential R&D with lower-risk projects that produce near-term returns through incremental improvements to existing products and processes. Using proven decision-quality techniques to analyze your portfolio and balance risk and return enables you to fund and manage projects more efficiently and cost-ejfectively, and ultimately play a key role in maintaining and enhancing your companys success.

Making Excellent R&D Decisions

Doing more with less and getting the right products to market at the right time require exceptional RD for example, deciding whether to invest in a given program area or project, or choosing one technology over another. This part of the process can be broken into four components: * Decision basis--Are the inputs to your R&D decisions of high quality? * Technology strategy--Is R&D strategy in tune with business goals at the highest level? * Portfolio management--Are your R&D portfolio and pipeline well balanced? * Project strategy--Is each individual project being managed to its full potential? The second broad category is organizing for decision quality. The focus here is on what a company can do institutionally to maintain a high level of decision qualify. This, too, is divided into four components: * Organization and process--Does your company have a good brain? Do decision makers select the right tools and think about them well? * R&D culture and values--Does your company have a strong heart? Does it hire the best people, train them well, and empower them to perform to their full potential? * Relationship with internal customer--Does R&D work closely with product marketing and other internal customers? * Relationship with end customer--Can R&D look ahead to anticipate and meet the needs and desires of the ultimate users of its results? The final component of excellent R&D decision-making is improving decision quality. …

Describing and Valuing Interventions That Observe or Control Decision Situations

The value of information and value of control calculations have long been two separate parts of a decision analysts efforts to extract as much insight as possible from a decision model. This paper unifies these concepts as interventions that modify the structure of the original problem, which have two key properties, purity and quality. Purity is an idealization that leads to Howard canonical form, clarifies the definition of control intervention, and allows us to extend and correct the calculation of the value of control. Quality is a characteristic that leads to generic models of imperfect intervention, which, because of their equivalence to any pure intervention, prevent misguided recommendations when the value of a perfect intervention is high but the value of a somewhat imperfect intervention is low. Quality is a number between 0 and 1 that normalizes and allows comparison of imperfect interventions between applications having very different value scales.

Smart Organizations Perform Better

OVERVIEW: Identifying linkages between the use of best practices and overall measures of corporate performance is difficult. Studies of several hundred companies, however, show that underlying cultural and organizational patterns lead to effective implementation of many best practices, which the authors call the “principles” of a smart organization. These patterns, which are measured with an organizational IQ test, correlated positively with overall corporate performance, leading to the conclusion that smart organizations perform better.

Outside-In Strategic Modeling

The roots of good strategic decision making are in cultural and organizational norms and patterns. One principle, which we call the outside-instrategic perspective, is essential for excellent strategy. This perspective led a company to recognize that its current strategy was counterproductive and that it must make almost a 180-degree turn. While many companies start with themselves and project market shares, earnings, and so forth out into the environment, the outside-in strategic perspective reverses this to start with the environment and work inwards to the company. In this case, we used a framework that starts with consumer spending in various entertainment areas and divided revenues along a simple value chain of retailers, wholesalers, and producers. An influence-diagram-based model led to counter-intuitive insights about the most valuable segments.

Comment on Influence Diagram Retrospective

Influence diagrams were first used in 1973 as a way to model political conflicts in the Persian Gulf and measure the value of information collected by the Defense Intelligence Agency. The number of scenarios for events in the region was too large to be represented as a conventional decision tree model. Influence diagrams were initially conceived as a way to create smaller, coalesced decision trees that required fewer probability assessments. However we found that they also facilitated communication between analysts, experts, and policy makers. Influence diagrams later became the basis for new ways of solving decision models.

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.