IQVIA delivered targeted recruitment data with intelligent automation.
IQVIA is a leader in Human Data Science, the integration of data science, and technology & human science to help customers address both current needs & future opportunities. Faced with an arduous and inefficient clinical trial physician recruitment process, IQVIA knew there had to be a better way that incorporated data science and intelligent automation.
Based on a referral from within the healthcare space, the IQVIA team came to Promptworks for help building an Intelligent Automation solution to replace their slow & costly manual processes. They also wanted to see whether our reputation for technical excellence and adherence to agile principles could provide a model for their entire enterprise.
Here was our challenge:
IQVIA has grown through strategic acquisition. When they approached PromptWorks, it was with the goal of bringing together their many disparate sources of data, taking the onus off the researchers for list-building, and significantly diminishing the amount of time it took to build a list. Their in-house data scientists were data experts but needed help automating the process and building an application that would streamline the process for end users.
On top of their internal needs, we had to be mindful of requirements around healthcare data - needs that ended up requiring us to use on-prem solutions, as well as the competing goals that sometimes arise between innovation and end user. With significant internal time and budget savings on the line, IQVIA needed a partner that could help deliver on their vision with a real world solution.
Here's how we solved it:
Data First Mentality
For IQVIA to streamline the clinical trial recruitment process, they had to effectively leverage multiple data sources that were locked in different systems and stored with different structures. Handling this meant creating an efficient data pipeline that could pull, clean, rationalize, match, and dedupe information across these data sources, and then store the result in a system designed to efficiently answer the questions their team needed to ask.
As trials progressed, the upstream data sources also evolved. We needed a solution that included a strategy for establishing data provenance as well as processing data efficiently.
On Prem Focus
Due to legal agreements surrounding the sensitive nature of the data, the project had to be developed and deployed on premises.
This presented us with our first challenge. There wasn't a stable deployment infrastructure available at the time. IQVIA had a new mandate to move toward Kubernetes, but it was a new technology. As a result, the Promptworks team helped guide IQVIA's deployment of On Prem bare metal Kubernetes. Once the platform stabilized, we helped them establish best practices for how to utilize it.
See the results:
Working closely with IQVIA's product and data expert, PromptWorks created a machine learning data pipeline, performed much of the machine learning and statistical calculations, and fronted the data with a powerful web application.
Along the way, we hit several scaling challenges having to do with everything from the Kubernetes/SAN configuration to the need for deep knowledge when it came to database query engines to the complexity of how numpy and pandas accelerate calculations below the python level. Additionally, the product required the development of efficient algorithms to help strategically process and serve results.
The resulting product helped to significantly minimize the amount of time that was needed by research in trial. What once took a month could now be completed in just a minute or two. Our technical breadth and depth, coupled with the unmatched pace of our agile process, also helped inform the overall pace of technical innovation at IQVIA.
Interested in more examples of our work?
See how we helped Drexel investigate new treatment protocols with a mobile app and integrated algorithm or how Picwell helped consumers navigate health insurance using machine learning.
Contact us to discuss your next project.Contact