Building an analytics engine capable of mastering and integrating large data sets to solve for knotty business questions is an enormous challenge for life sciences companies. But those that have managed to build the engine often find the next step -- turning the analytics generated into actionable insights that drive results – an equally tall order.
Just as, alas, the data don’t magically clean themselves, valuable analytics can wind up parked on a desk gathering dust, absent some forethought into how they’re plugged into workflow across different units. Here are a few ways companies can ensure value generation at the implementation stage:
- Opening internal lines of communication around data and analytics, with deliberate expectations management and outputs that speak to stakeholders across units and at different levels.
- Establishing processes for plugging insights into practice and ensuring that teams are aware and incented to make use of them. For example, using predictive modeling to power a “Next best action” or “Suggestion” function allows for data-driven nudging, and studying the comparative impact of data-driven decisions can promote broader adoption.
- Cross-pollinating the knowledge sets of your data nerds and the business team. Data teams can get disconnected from the mission if they’re not equipped with deep knowledge of the commercial context underpinning their work.
When analytics work is grounded in deep clinical and commercial expertise, powerful tools can be brought to the task of improving patient health. An example :
Recently, we worked with a diversified pharma preparing to launch a product for a seriously underdiagnosed condition. An analysis using DRG’s multi-year claims data repository, covering over 300 million lives, showed only 150,000 diagnosed patients in the U.S. – but the company’s clinical team projected a prevalence more in the range of 2-3 million individuals.
Identifying these undiagnosed patients posed a tricky puzzle. Working in close collaboration with the company’s brand and medical teams, DRG data scientists and epidemiology analysts built a profile of the undiagnosed patient by identifying disease marker combinations, such as comorbidities (visible in claims data) and specific symptoms (encoded in EHR data). We then validated and strengthened the profiles through application of machine learning to identify relationships within the data. This enabled us to build an algorithm that would find potentially undiagnosed patients across disparate data sets, showing the true size of the estimated patient population.
As a result, the company was able to confirm its projection of the patient population and to identify a key dynamic driving underdiagnosis – while primary care physicians seldom correctly diagnosed patients, endocrinologists nearly always did. This finding allowed the company to develop highly targeted sales and marketing strategies for the different physician audiences, serving primary care physicians general education while focusing on drug benefit communications with endos, and could help many patients who would have otherwise gone undiagnosed and untreated.
The challenges we face in achieving data maturity are great, but they are greatly eclipsed in scale by the potential of these techniques to transform our business and drive better care and improved outcomes. Read our recent white paper, How to Make Health Data Work for Life Sciences Companies, for further insights into these topics, and stay tuned for our fourth and final post in this series, which will look at organizational factors in achieving data success.