Agile IT strategy in pharmaceutical R&D

Agile IT Strategy in Pharmaceutical R&D

Clearly communicated aims should be the goal.

By Andrew Chadwick

March 24, 2009 | As pharma IT groups face tightening budgets and loosening ties between research areas, is an IT strategy spanning multiple research areas still a meaningful aim? To what extent can agile, rapid delivery methods and the increased use of commercial packages ease the burden on internal development, without threatening the coherence and span of the information that scientists need?

Strategy—the “art of war”—is primarily about deploying resources into the right area: fast-moving cavalry to take on local challenges in a flexible and responsive way, while slower-moving forces take strategic objectives over long periods. For IT in drug discovery, a strategic objective is typically to improve creativity and the ability to predict adverse events, which necessitates sharing information between projects and sites. But this will only happen if R&D leaders demand that projects go beyond their own immediate operational needs when recording and organizing their results.

Attempts to define a multi-year, global IT program are typically frustrated by constant change—in research methods, in IT technology standards, and in company structure. Agile development helps to compensate for this. It allows rapid delivery of business benefits by allowing information users and creators to collaborate and find the best means to reach a defined and shared business objective. Rapid prototyping and iterative development methods such as DSDM and Scrum can clarify key requirements quickly, identifying risks in time to overcome them.

However, small, fast-moving teams—the “light horse regiments”—can find it hard to reach, or even identify, objectives that lie beyond the immediate sight of responsible users. Light cavalry alone cannot win a war. The famous mistake during the Charge of the Light Brigade was not a failure in strategy but a failure in communication, the light cavalry charging at the target they could themselves see, instead of the one that their general could see over the hill.

So the longer term aims of a strategy need to be both clear and communicated. What should these aims be for a modern discovery organization?

Capturing Knowledge: Intent and Infrastructure

Sustained improvement in R&D performance requires use of one project’s results, positive or negative, to help improve the conduct and success of future projects. This means pulling information upstream against the flow of the R&D process, which improves the pipeline quality in several ways:

  • Clinical chemistry results may carry clues to mechanism or future side-effects, useful to the next discovery project.
  • Patterns of activity across multiple targets indicate safety margins.
  • Linking targets, pathways, and disease interventions allows repurposing of approved drugs.
  • The prevalence of different failure modes should determine what efforts to apply in “de-risking” projects. The most effective plans examine the high-risk areas first.
  • Comparisons of predictions and outcomes calibrate the reliability of assays, in silico, and human predictions, which influences their priority and best role in pipeline filtering. This may also help to eliminate tests that add no value, or worse, destroy it.

Such upstream flow requires active intervention by management to set a strategic intent. For example, a knowledge sharing contract will encourage project teams to release and organize information needed by others. Such cross-project knowledge management can meet resistance, but this can be overcome if the mutual benefits are made clear. The key to open information doors between projects is investment in metadata:

  • Identify commonalities in pathways and mechanisms to compare projects.
  • Pool reliability and calibration information to compare assay effectiveness.
  • Agree on data definitions to allow the pooling of results obtained at various places, times, experimental systems.

Complete worldwide standardization of data and IT including use of strict data standards is feasible—indeed essential—in Phase 3 clinical trials, but not realistic in a discovery culture. In our experience, a federated approach is more practicable. A federated database architecture is a set of heterogeneous databases that, through common metadata, can provide a unified search, allowing greater use of commercial packages to meet local and tactical needs.

There is growing demand from clients to specify and source commercial packages, e.g. instrument data capture, sample management. These days, who builds a chemical search application in house? Use of such packages saves time, reduces risks, and builds the essential user trust in delivery by the IT organization.

However, choice of a commercial package can frustrate efforts toward a single technical architecture or standard (e.g. open source, or service oriented architecture). Unlike manufacturing and financial, there is no dominant ‘enterprise research system’. Overlaps, for example between the scope of electronic lab notebooks and LIMS, can complicate planning and require clarity on the information model and standards for interfacing.

We typically recommend that the available manpower for IT custom build within Discovery groups be focused into two areas: 1) new science that is a source of R&D process innovation and differentiation, such as reproducible processing and objective interpretation of complex biomarker data; and 2) new knowledge that comes from search over multiple results.

Putting value on an IT strategy is easier if there is an end-to-end view. This forges the essential link between science, information, IT, and the business that must fund the IT investment.

Source: Bio-IT World

© Copyright 2018 Tessella
All rights reserved