kdb Is Transforming The Use Of Data Analytics In The Pharmaceutical Industry

KX commissioned Forrester Consulting to conduct a Total Economic Impact™ (TEI) study. To better understand the benefits, costs, and risks associated with kdb, the vector database, time series, and real-time analytics engine, Forrester interviewed two decision-makers from a life sciences organization.¹ This abstract will focus on the use of kdb and advanced data analytics in the development of new drugs across different trial phases and how this is transforming the pharmaceutical industry.

Two interviewees from a life sciences organization participated in this study:

• The chief data and analytics officer discussed how kdb was integrated into their organization’s new data backbone. They detailed the clinical trial use case.
• The vice president of advanced analytics discussed how kdb was integrated into their platform and used for patient simulation use cases.

Pharmaceutical companies have to spend billions of dollars and many years in bringing a new drug to market successfully. The use of advanced data analytics in healthcare has lagged other industries; as a result, performing complex queries of databases with very high volumes of data took a long time and/or required costly and sub-optimal infrastructure.

The interviewees from the life sciences organization noted seeing the opportunity to use kdb to quickly and efficiently analyze very large volumes of data, including real-time sources for a number of different use cases. These use cases included patient simulation, whereby initial new drugs can be trialed without the need to test on humans as part of the first testing phase; and clinical trials, whereby optimal timing and location of clinical trials ensure they happen at the right place with the right investigator and with sufficient patient availability. These are typically in later testing phases. The benefits of these use cases not only include reduced time and lower costs in new drug development, but also reduced negative outcomes of drug trials and shorter time to market for new medications and treatments.

Investment Drivers FOr DATA ANALYTICS IN HEALTHCARE

The interviewees’ organization adopted kdb because it saw the opportunity to transform the way data analytics could be used in the pharmaceutical industry. Interviewees noted their organization wanted to accelerate and improve the value of its data processing capabilities to improve its customer offerings. This organization was looking to optimize its legacy environments, including:

• Reducing data processing times. Processing time with legacy systems could result in inefficient workflows as teams waited for data results. As a result, interviewees stated their organization was always looking for opportunities to improve processing times, maximize the efficiency of its teams, and make data available to customers for real-time decision-making.
• Improving scaling to high-volume data processing, including real-time sources. The interviewees’ organization was always looking to maximize the useful data that could be included in its analyses, especially streaming data to allow for up-to-date information and informed decision-making. At the same time, the interviewees’ organization wanted to continue streamlining queries and analysis of its data sources to make data more accessible to its employees and customers.
• Reducing costs and increasing performance of legacy technology infrastructure. The compute power required for the processing of very large volumes of data can be extensive. The interviewees noted their organization looked for ways to improve performance and reduce costs of legacy infrastructure and systems.

kdb Features

The interviewees’ organization chose to invest in kdb for the following reasons:

• Fast, flexible, and efficient time series database and analytics engine. The architecture and design of kdb provides faster and more efficient results and also supports more flexible queries and types of analysis. kdb has native support for time series operations, which accelerates performance and improves data results, particularly for very high-volume data queries and analyses.
• Integration of real-time data inputs. kdb incorporates live-streaming data with historical datasets to enable real-time, up-to-date analyses and results without the need for additional changes or coding.
• High performance with low infrastructure cost. The system is designed to provide high performance at low cost by optimally distributing database operations across compute power; better exploiting CPU cache memory; and leveraging in-memory workspace.

Key Results FOR PHARMACEUTICAL DATA ANALYTICS

While there are four important benefits detailed below, the overall, combined impact is transformational and will enable pharmaceutical companies to reduce the time and cost of bringing new drugs to market. New business models and ways of working are still being explored to further enhance the capabilities that kdb enables.

The results of the investment for the interviewees’ organization included:

A step change in information processing time compared to traditional database design and structure. The benefits of accelerated data processing were:

• A reduction of 90% or more in processing time. With kdb engine, it was possible to analyze and process large amounts of data much more quickly. This enabled decisions to be made sooner, and processes and workflows to become more streamlined and efficient. The chief of enterprise data and intelligence officer said: “It would take an hour and 5 minutes. We have the same process now running in a minute and 23 seconds.”
• The ability to demonstrate query results in front of the customer. Whereas previously the interviewees’ organization had to take questions and queries in a customer meeting, run them as a follow-up activity, and then share results in a follow-up meeting, kdb allowed it to now run a possible live demonstration. The chief of enterprise data and intelligence officer said: “There’s nothing more powerful than being able to be in front of a customer and run predictive simulations in a matter of seconds versus saying, ‘I’m sorry, we can’t answer your question, we’re going have to go back to the office and we’ll get back to you in two weeks after we get done crunching the data.’”
• Increased database size. In patient simulation use cases, the interviewees’ organization increased the database size from 5,000 patients to 100,000 patients, while still reducing the data processing time from four months to a matter of weeks and complying with all requirements.

Increased data value from the inclusion of real-time data sources. The ability to add real-time data streams greatly increased the value of the analysis.

• A reduction of six to eight months in clinical trial time. Not only could the interviewees’ organization meet the customer and regulatory criteria for the clinical trial of a drug, including the identification of the right investigator and an appropriate set of patients, it could also run a real-time situation assessment of trial sites and identify the ideal trial site and timing for that trial. This reduced the chances of failed and unsuccessful trials, not only reducing costs but also reducing the total clinical trial time by half a year or more, which potentially enabled new drugs to come to market much sooner. The chief of enterprise data and intelligence officer said:  “And so by including this type of intelligence into your clinical trial design, it can cut six to eight months out of a clinical trial timeline and be able to expedite the ability to get medications to patients.”

An agile and scalable approach keeps costs down. Achieving scalability with the capacity to pivot and expand allowed for cost efficiency and kept essential resources to a minimum:

• Reuse of code minimizes additional resource requirements. In the patient simulation use case, there was not a lot of incremental effort required when applying to new clinical areas as the bulk of the code can be reused. The vice president of advanced analytics said:  “Manpower wise, we don’t need a lot of people because we have the code already written up — it’s really just a matter of fine-tuning parts of the code to conform to any nuances of the new therapeutic area.”
• Use of automation and optimal architectural design limits development resource requirements. The new data backbone was built leveraging automation, and while the initial investment into training a team of 20 developers was not insignificant, there is no need for additional resources when it increased in scale, using the KX Data platform powered by kdb. In the clinical trial design use case, the interviewees’ organization created new use cases quickly without the need to expand the team or add resources.
• Reduced infrastructure costs. The implementation of kdb enabled lower compute power requirements, which increased the interviewees’ life sciences organization’s ability to be flexible in how it prices its services. It reduced customer costs by 50% to 60%, while still increasing its margins.