شاخص های تصمیم گیری بهینه برای ارزیابی پروژه "تحقیق و توسعه" در صنعت داروسازی: شاخص پیرسون در مقابل شاخص گیتینز

This paper examines issues related to various decision-based analytic approaches to sequential choice of projects, with special motivation from and application in the pharmaceutical industry. In particular, the Pearson index and Gittins index are considered as key strategic decision-making tools for the selection of R&D projects. It presents a proof of optimality of the Pearson index based on the Neyman–Pearson lemma. Emphasis is also given to how a project manager may differentiate between the two indices based on concepts from statistical decision theory. This work demonstrates and justifies the correct use of the Pearson index.

This paper addresses the problem of evaluating research and development (R&D) projects in the pharmaceutical industry by using appropriate indices for determining priorities between potential R&D drugs within the Bayesian decision framework. An R&D process is divided into a number of distinct phases that must be implemented sequentially (in a fixed order) and all succeed before a drug is marketed to yield any financial benefits. The phases and sequence of a research project’s stages are explained by Gittins, 1996 and Gittins, 1997. Typically, a project is characterized by the exploratory research stage in pharmaceutical laboratory, which is followed by the development and the marketing stages. There are several important features of the R&D process: It is considered a very uncertain process, it needs at least 15–30 years from the time the research starts until a drug is marketed, and the initial investment is large. Towards the end of each stage the results from the current stage will become known and a decision should be made regarding the fate of the project.

The primary objective was to provide theoretical justification for the use of the Pearson index as an optimal decision tool for the parallel classification of R&D projects. The basic model viewed the project selection as a statistical hypothesis problem in a Bayesian framework. A hypothesis statement is either true or false. In an uncertain environment this is not known and one has to make decisions-based on the available information as to whether the alternative hypothesis is valid or not. In project prioritization, the uncertainty refers to the probability of success of a given project. A project will either succeed or fail, a fact that will be ascertained after the event. The key issue is how the decision maker chooses projects at an early stage (uncertain environment) in order to maximize the probability of stating correctly which projects will eventually be successful given a budget constraint. The Pearson index can be viewed as the right decision rule used when a set of projects needs to be divided into two groups: the projects that are going to be implemented and the projects that will never be considered again. This is equivalent to maximizing the posterior probability that a project is classified correctly when a budget constraint is imposed.