Artificial intelligence (AI) and machine learning (ML) have changed countless industries over the past decade. However, there’s a big difference between an AI tool alerting consumers to the contents of their refrigerators and machine learning systems driving the future of clinical trial development.
Clinical research organizations (CROs) and sponsors have approached these tools with both optimism and caution. Researchers are starting to understand the power of large-scale data management and analysis. In this post we look at how AI and machine learning tools are optimizing the clinical trial process. If you would like to learn more, please register and attend the next ACDM Lunch & Learn on Thursday August 5, 2021.
Fewer Data Sets Go to Waste
Data collection has always been a key part of research. As technology grows more advanced, it’s easier to collect even more data points and harvest valuable information that can determine the effectiveness of a drug or treatment plan. Still, just because data is collected doesn’t necessarily mean it gets used. As a result, hundreds of key data points are often ignored or thrown out during the analysis and publishing processes.
AI and machine learning developers are working to fix this. They want researchers to use the data points they collect in the short run but also in future trials.
“Companies are using machine learning approaches to understand the data they already have in-house and how they can more effectively learn from it for future studies/trials,” writes the team at data-focused CRO Bioforum.
Additionally, these developers are training their systems to highlight data that could benefit from additional review and use. This means that researchers get more from the trial efforts, driving a higher ROI over time.
Machine learning is leading researchers into the future of data optimization, or getting more from the information collected from patients. These systems can also work to improve data cleaning, so fewer data sets are rejected because of collection errors and system hiccups.
“As clinical studies continue to become more complex… it is important that the data generated is used in the optimal way during the trial,” says Jennifer Bradford, director of data science at global biometrics CRO Phastar. “Powerful ML technologies have the power to monitor this data as it is generated; identifying issues and inconsistencies as trials are ongoing.”
AI and machine learning tools are also fast, which means CROs can catch problems early on, preventing the entire trial from getting corrupted by bad data.
Incorporating Multiple Data Sources
Not only can CROs do more with the data they collect, but machine learning and AI systems can also help them pull from existing data sources outside of the trial environment.
“Over the past few years, biopharma companies have been able to access increasing amounts of scientific and research data from a variety of sources, known collectively as real-world data (RWD),” write researchers Karen Taylor, Maria Joao Cruz and Francesca Properzi at Deloitte. “However, they have often lacked the skills and technologies to enable them to utilise this data effectively.”
Taylor, et al. explain that the clinical trial process is likely to become less linear, which will allow for greater flexibility and agility to focus on treatments that work.
Faster, Better Clinical Results
The goal of implementing AI and machine learning systems is to move pharmaceutical treatments to market faster. This in turn gets patients the medications they need with fewer side effects and costs. Future trials will be less wasteful and can be completed in shorter times while still completing each research step thoroughly.
“AI models can crunch through millions of pharmaceutical compounds and predict what might work best far faster than a human researcher could,” says Rob Karel, vice president of product marketing at identity platform Okta. “AI models require high-quality data sets to work with, so that the AI model can learn from the data, helping it make further predictions that develop from a foundation of sound science.”
CROs and sponsors are already looking to hire researchers with backgrounds in data science who can use these systems and optimize them. The future of clinical research is here, and it is machine learning.
“Everybody is scratching their heads on how we are going to cope with the complexity and the speed of change,” says Patrick Nadolny, innovation committee chair at the Society for Clinical Data Management and global head of clinical data management at Sanofi. “What we need to do is define a very clear roadmap of where we are going.”
Artificial intelligence and machine learning cannot solve every problem with clinical research. However, these tools can increase the amount of data collected, pull multiple data sources, and ensure the data is clean for researchers to properly analyze. In a few hours, they can accomplish tasks that would take human researchers months or even years. These efforts can benefit the researchers who develop the treatments while getting the right therapies into the hands of patients faster.