October 8, 2002

Central Intelligence

In life sciences as in other industries, sharing integrated business and competitive intelligence is a valuable competitive advantage

by Adrienne Tannenbaum and Elby Nash

The life sciences industry — which comprises biotechnology, pharmaceutical, and medical devices firms — is in the business of discovering, manufacturing, and distributing new drug compounds, molecules, therapies, or technologies. Production of physical products by these firms depends on, and generate, new knowledge and information related to the discovery and application of additional molecules, compounds, therapies, or devices. This knowledge and information is always the basis for creating other, new products.

This knowledge, usually internal, can become part of a well-defined business intelligence (BI) environment when it is gathered, categorized, and made accessible to a company's key internal stakeholders. The same knowledge, when augmented with information about the external marketplace's interests and activities in related knowledge, can become part of an organization's competitive intelligence (CI) capability. Simply supplementing categorized internal BI with competitor product development and delivery, adverse event information, articles in peer-reviewed medical journals, and recruiting patterns expands the internal view to one that can be used competitively.

The objective in either case is to get a competitive edge by making more informed, faster decisions about new or potential products. Life sciences organizations are already thriving on volumes of information — structured, unstructured, internal, and external. What's lacking is the foresight to put it all together in a fashion that guarantees its reusability for continued and accurate justification of corporate decisions.

In the life sciences industry and elsewhere, it's generally accepted that most CI needs can be satisfied by information already contained within the organization. Unfortunately, because this information is often not identified or easily locatable, it's pursued from external sources at an additional cost.

The Origins

Pharmaceutical companies are under particular stress to maximize their return on each new drug introduced to the market. The cost of researching compounds that fail because they're either too toxic, produce undesired side effects, or simply don't demonstrate efficacy against a disease has been directly affecting profitability. Industry financial analysts estimate that each new approved drug must return $300 million to $600 million just to cover the costs of the research failures. To recover these costs, pharmaceutical companies now spend almost twice as much of the allocated R&D budget (and sometimes more) on marketing and related administrative costs.

With the economic stakes so high, the need becomes more imperative to make timely, well-informed, strategic business decisions about:

• Continuing, pursuing, or terminating projects (compounds/devices/molecules)
• Evaluating companies that might be appropriate partners, or merger and acquisition candidates
• The competition — their current research or product in therapeutic areas and its impact on current research or product pipelines
• Identifying different or novel ("off-label") uses for existing products
• Locating researchers who may become important opinion leaders for the company or its competitors.

Only when information about these impacts is aggregated, analyzed, and made available to key decision-makers can a life sciences company (or any company) truly make informed and timely decisions about its strategic responses.

Information in the Life Sciences Industry

The life sciences industry depends on both structured and unstructured data. Because most of this information originates from diverse unstructured sources (such as patents, patent literature, post-marketing Phase IV adverse event reports, trade and industry journals, and so on), it is often unconnected and inaccessible. Likewise, traditional data is collected from internal studies, manufacturing processes, clinical trial results, and sales numbers. Most of this structured data is stored and made available for subsequent analysis. However, the types of analysis required rarely affect the data before it's collected — hence the need for a broader view of data collection.

Within the life sciences industry, collected information has traditionally focused on that which is structured. Commonly, these databases result from single-project events, usually focused on one compound's life cycle. Much of this data is "personal" in the sense that it's collected by individual researchers as a means of supporting and validating the identification of a potentially new product. As the product's life cycle moves forward, data is collected during clinical trials, often in a separate database geared toward retaining required FDA statistics. Included in this data capture are manufacturing costs, timeframes, patient trial characteristics, and end results.

Finally, when a decision to go ahead has been reached, the product is marketed according to its assigned trade name and the collection of sales data begins. Each of these data collection landmarks is typically isolated, in that it is designed for and focused on a particular product and event in its development and distribution life cycle.

Based on these isolated data collections, the need to evaluate an entire product's transition from research through sales and distribution usually requires the need to look at isolated sporadic files: documents, spreadsheets, custom and commercial databases, and various other isolated files. The perpetuation of these inconveniences led to the development of the data warehouse in many life sciences organizations.

Most data warehouses result from renewed and extended information consolidation based on one product's collected structured data view. In some situations, corporate data warehouses focus more on the revenue and sales results of major firm products. In neither case, however, does an organization have the ability to cross-analyze the life cycles and phase-based characteristics of multiple products. Data warehouses, therefore, are limited in their contribution to BI or CI environments, particularly when they result from data that has already been collected to support an already initiated process.

In separately initiated integration efforts, unstructured information has often been organized via a document management system. In these scenarios, research documents, FDA application support, patents, and product evaluations are organized so that they can be located via keyword searches, minimally the associated compound and trade names. Again, however, the document management system is restricted in the type of contents represented and organizes unstructured information represented by internal documents. Packages typically encourage or restrict (or both) the implemented document search criteria, and the ability to connect existing structured data with documents located via this type of application is usually attempted only as an afterthought.

In many organizations, the corporate research library is the BI and CI environment. Here, both competitive and internal information is stored, focusing primarily on the information that substantiates or defends the organization's products. Because much of this information is external to the organization, instant Web-based research usually supplements corporate library support. Regardless, corporate library-based research efforts are individual: Research librarians often gather information upon request and individual employees conduct their own information gathering. In neither case is the gathered information organized, related, or made available to other employees as the result of their own supplemental keyword-based queries. From a cost-benefit perspective, repeated information gathering prohibits timely and complete decisions.

The Missing Components

Isolated information, structured and unstructured, is clearly not earning its keep. Aside from the fact that it's often gathered after the fact, even those efforts requiring and evaluating industry information before a product development effort is undertaken are isolated and not reused. Therefore, a BI-CI environment requires a specific set of architectural qualities:

• Metadata. Information is great when you find it; it is even more valuable when you understand its origin and meaning. Metadata — the information that details the components of your potential BI-CI environment — must be as accessible as the instance data that is being sought. The completeness of BI-CI metadata focuses on the ability to answer the 5 Questions™ we've developed at Database Design Solutions with respect to life sciences information: What data or information do I have? What does it mean? Where is it? How did it get there? How do I get it?
• Logical and physical connectivity. Keeping potential BI-CI information sporadic and unconnected requires repeated searches and evaluations over time. All existing and competitive information is in fact logically related, often by product, research compound, manufacturer, life science organization, or internal organization. Logically, these connections are often determined via the creation of a simple enterprise data model comprising the major subject areas within an organization and their logical relationships. (See Figure 1.) In many organizations, the need to pursue a "data focus" as opposed to a "process focus" is a prequel to this exercise. Physically, unstructured and structured information must also appear to be connected so that a researcher can easily get from one BI-CI component to another, based on information currently under evaluation.
• Subject area-based access. Enterprise data models can form the search beginnings within a BI-CI environment. Each of these high-level subject areas should represent tag-based identifiers within the organization's intranet-based BI-CI framework. Consider a subject area of "Product" and realize the need to identify all product-related data by its generic or trade product name, as required. Likewise, a subject area of "Department" would be treated similarly.

Using What You Have (And Don't Have)

As external information is gathered, it needs to become part of a defined internal BI framework. At the heart of this framework is a metadata-based model that relates each piece of BI-CI information to its appropriate search key. This metamodel identifies and relates search keys so that the analyst can easily move from one BI-CI component to a logically connected supplement when necessary.

• Privacy
• Your California Privacy Rights
• Copyright © 2008 United Business Media LLC