Managing a Portfolio of Interdependent New Product Candidates

Managing a Portfolio of Interdependent New Product Candidates

in the Pharmaceutical Industry

Gary E. Blau, Joseph F. Pekny, Vishal A. Varma, and Paul R. Bunch

Highly regulated industries such as pharmaceuticals and agrochemicals face the

challenge of maintaining a 0continuous stream of new products. This is difficult

because of low probabilities of technical success, high development costs, uncertain

market impact, a scarcity of good new product ideas, and limited human and capital

resources available to develop them. The problem of evaluating and selecting which

new products to develop and then of sequencing or of scheduling them is complicated

further by the presence of dependencies between products both in the market place

and in the development process itself.

This study proposes a portfolio management approach that selects a sequence

of projects, which maximizes the expected economic returns at an acceptable level

of risk for a given level of resources in a new product development pipeline. A

probabilistic network model of distinct activities is used to capture all the activities

and resources required in the ‘‘process’’ of developing a new drug. A prioritization

scheme suggesting sequences for developing new independent drug candidates with

unlimited resources is generated with a conventional bubble chart approach. These

sequences initiate a genetic algorithm (GA)-based search for the optimal sequence

in the presence of product dependencies and limited resources. By statistically

evaluating the sequences generated during the GA search using a discrete event

simulation model, it is possible to construct an economic reward-risk frontier that

illustrates the trade-offs between expected rewards and risks. The model ideally is

suited to answer various ‘‘what if’’ questions relative to changes in the resource level

on pipeline performance.

The methodology is illustrated with an industrially motivated case study, involving

nine...