You are here

Improved Performance of Models Enables Innovation of New Methods for Analyzing Post-Market Drug Efficacy


Background

The testing of the safety and effectiveness of new drugs continues even after drugs are approved for sale. In the months and years after a drug goes to market, researchers can observe how well drugs perform and what side effects may occur, but valid assessment of drugs’ performance outside of randomized trials can be challenging. Development of methods for non-randomized research requires extensive simulations based on large and complex datasets, which in turn require extensive staff time and computing capabilities. The rapid development of research methodologies allows medical investigators to draw more accurate conclusions more quickly, thus helping improve patients’ lives.

Challenge

Very large simulation studies are often required to create and improve methods for assessment of drugs’ safety and effectiveness. Unlike randomized early-stage trials, the hallmark of these studies is that they work with data observed in the real world from patients who’ve been to their doctors, been prescribed a medication, taken the medication and experienced a qualitative health outcome, either positive or negative, as a result of the medication.

As a consequence, drawing conclusions from the data observed in the real world of medical care tends to be more complex and more problematic than the data analysis in the controlled environment of early-stage trials.
"This is a very difficult thing to simulate,” says Jeremy Rassen, ScD, Asst. Professor of Medicine, Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, at the Brigham & Women’s Hospital and Harvard Medical School in Boston. "We want to create new methods and we want to push the bounds and test the validity of these methods. We need to know whether our methods give us valid results that take into account all of the nuanced statistical issues about working with real patient data observed in a non-randomized environment.”

"The reason that a particular treatment is prescribed might be that the patient’s disease is very advanced. Someone with very advanced disease may also tend to have a negative outcome, as opposed to someone who is relatively healthy who may be more likely to have a positive outcome. The outcome we observe as medical researchers could be partly due to the positive or negative effect of the drug, but is also heavily influenced by the stage or severity of the disease,” says Rassen. "As a researcher, you have to untangle that in every study, using hundreds or thousands of variables.”

Teasing out meaningful information often requires multiple simulation studies. And that’s where the process can become extremely difficult.

"We can run one simulation relatively quickly. But if we’re trying to determine whether a new method works, then we’re doing it hundreds, thousands, tens of thousands of times,” says Rassen. "If a simulation takes 20 minutes, we would be able to wait 20 minutes. But we’re just not able to wait 20 minutes times 10,000 runs to do all the simulations needed to prove a method.”

Solution

Rassen’s team uses a variety of approaches for analyzing data, including a solution developed jointly by IBM Netezza and Revolution Analytics, which distributes an enhanced distribution of R, an Open Source language designed specifically for data analysis. He has also used SAS and several programs written in open-source R.

The advantage of the joint Netezza-Revolution approach is two-fold: From a user’s perspective, it is faster and more flexible than the SAS solution. And unlike standard R with in-memory datasets, which have limited capabilities for analyzing large or complex data sets, the joint solution can handle the kind of “tangled” data used in post-approval drug testing scenarios.

The combination of Netezza and Revolution “has allowed us to do things we couldn’t do before. We can set up large matrix operations…big enough to cause trouble for standard R programs… and we run them in a very reasonable amount of time,” says Rassen.

Results

Once deploying Netezza as their research analytics infrastructure along with Revolution R Enterprise for IBM So far, the results have been positive. Rassen and his research team can run large-scale simulation studies in faster timeframes. As a consequence, the team’s productivity and efficiency have increased.

"At the end of the day, our job isn’t about the programming and it isn’t about the system. It’s about getting the work done. I’m all in favor of the solution that lets me get the work done quickly and easily,” says Rassen. “We’ve been able to do something that we can’t really do otherwise. The speed is pretty remarkable. And I’m always pleased when technology products work the way they’re supposed to."

User

About Revolution Analytics

Revolution Analytics was founded in 2007 to foster the R community, as well as support the growing needs of commercial users. Our name derives from combining the letter "R" with the word "evolution." It speaks to the ongoing development of the R language from an open-source academic research tool into commercial applications for industrial use.

Though our Revolution R products, we aim to make the power of predictive analytics accessible to every type of user & budget. We provide free and premium software and services that bring high-performance, productivity and ease-of-use to R – enabling statisticians and scientists to derive greater meaning from large sets of critical data in record time.  

We also offer our full-featured production-grade software to the academic community for FREE, in order to support the continued spread of R's popularity to the next generation of analysts. 

For customers such as Pfizer, Novartis, Yale Cancer Center, Bank of America and others, our flagship Revolution R Enterprise product stands for faster drug development, reduced time of data analysis, and more powerful and efficient financial models.

About Revolution Analytics

Revolution Analytics was founded in 2007 to foster the R community, as well as support the growing needs of commercial users. Our name derives from combining the letter "R" with the word "evolution." It speaks to the ongoing development of the R language from an open-source academic research tool into commercial applications for industrial use. 

Though our Revolution R products, we aim to make the power of predictive analytics accessible to every type of user & budget. We provide free and premium software and services that bring high-performance, productivity and ease-of-use to R – enabling statisticians and scientists to derive greater meaning from large sets of critical data in record time.  

We also offer our full-featured production-grade software to the academic community for FREE, in order to support the continued spread of R's popularity to the next generation of analysts. 

For customers such as Pfizer, Novartis, Yale Cancer Center, Bank of America and others, our flagship Revolution R Enterprise product stands for faster drug development, reduced time of data analysis, and more powerful and efficient financial models.