AI and Machine Learning’s Role in Clinical Data Management

Data management is becoming much more complex for clinical research organizations (CROs), as trials become larger and data sources more diverse. Yet it is possible to eliminate 3,000 hours of manual review in a single trial through machine learning. As the industry moves to a single view of all clinical study data, will we get clean data ready faster? What does this mean for the role of the data manager? Trevor Griffiths, Senior Director, Clinical Data Management, Syneos Health and Richard Young, Vice President of Vault CDMS Strategy at Veeva sat down to answer these types of questions on the State of Digital Clinical Trials podcast. Here are some of their insights:

Electronic data capture (EDC) systems are now standard, in comparison to data entry capabilities of the last twenty years, allowing for the automation of processes, storage, and electronic data display, improving patient care while reducing processing times and costs. With an average of more than 12 different data types from various sources being collected for a trial, it is imperative to have a good decentralized clinical trial (DCT) strategy to capitalize on learnings from the captured data.

Using artificial intelligence (AI) and machine learning capabilities, research teams can move more quickly in the review of their captured data. Moving to clean data capture with AI and machine learning vs the original manual review of data, it has been reported from larger trails to reduce manual data cleaning by more than 3000 hours.

It is an exciting time for data managers who are being relabeled as data scientists, as they are getting involved earlier in clinical trial strategy development, helping to decide the best method of data capture for efficacy and efficiency. This leaves room for data scientists to analyze data vs. manually cleaning and processing data before it can be scrutinized.

Through providing data scientist trainings, it is imperative to evolve existing team member skills to make them as effective with different data types as possible and ensure they know how to work effectively with clinical teams. There will always be a need for on-site data management but training and transitioning data scientist roles to central monitoring capabilities where sourced data can be managed remotely makes the data scientists responsibilities more flexible.

When recruiting and looking for new data scientists, it is imperative each candidate have the skills to manage different data types and the ability to evolve as data analysis continues to evolve.

With the boom of different types of data, including the traditional data collected in central labs (ECG), interactive voice response systems (IVRS) and interactive web response systems (IWRS), we are also seeing the rise of data collection from wearables (heart monitors, smart watches, phones), and bedside monitors.

With these types of data sources becoming more commonplace in the industry means quicker access to data, which was revolutionary in the early days of digital trials but is now the foundation of how clinical trials are being done.

“I won’t be surprised if there will be another evolution that streamlines data collection even further” says Trevor Griffiths, Senior Director of Clinical Data Management at Syneos Health. With pressure to shorten DCT trials, “it is essential to have vendor management and a plan ahead of the trial start to ensure everyone is aligned on timelines.”Even with so many different sources of data to manage, “we can shorten DCT trial timelines with proper data analysis tools.”

Decentralized trials offer a new type of design flexibility – but how do we ensure that a DCT is operationally and scientifically optimized within this level of adaptability?

Data management success in clinical trials means having to work through solutions as trials progress. Being adaptive means bringing in new solutions and adjusting as needed while balancing a clear strategy and thinking to enable adaptive solutions. The diversity of data types requires a new level of coordination to organize and analyze the data efficiently and effectively. Time spent on coordinating the use of data replaces the older manual execution of data processing (i.e. less data cleaning leaves experts to be more involved in analytics and level-setting).

The key to coordinating data management while remaining flexible and adaptable during a trial requires a high-level touch with sponsors. Hiring a clinical project manager to work with DCT vendors to align expectations allows data scientists to come to the forefront. Being able to process all the relevant data during the set-up or closing of a trial with the wide range of data types and proper stakeholder management means the trial will have the appropriate amount of support for success.

Explore how Syneos Health problem solves for data management in decentralized trials and across the product development lifecycle.