Dear Data Analysts, AI Is Not Replacing You

By Dan Schell, Chief Editor, Clinical Leader

At this point, what don’t you know about AI’s potential use in clinical trials? Can it help with site selection? Yep. How about patient recruitment? Absolutely. How about improving diversity in trials, accelerating enrollment, reducing trial design time, automating regulatory documents, collecting and analyzing RWD … well, you get the point. These days, it seems that if you have a problem related to clinical trials, some company out there has an AI solution for you.

More and more use cases are also starting to see the light of day, which is encouraging that this AI wave isn’t just hype. In January, the panelists for our Real Applications For AI In Clinical Trials Clinical Leader Live gave first-hand examples of AI use in applications such as documenting a TMF and its use in Alzheimer’s research. As you would expect, a plethora of AI-related presentations filled the agenda of this year’s SCOPE Summit, especially in the data track. As an example, in How Pfizer Uses AI to Transform Clinical Trial Data Oversight, Moe Alsumidaie does a nice job of summarizing some of the key points of a SCOPE presentation from Pfizer’s Demetris Zambas, VP and global head of data monitoring and management. Another good resource is the 2023 article Artificial Intelligence Applied To Clinical Trials: Opportunities And Challenges, which includes insights from “publications tackling the use of AI and Machine Learning (ML) in CTs from the past 5 years in the US and Europe, including Regulatory Authorities’ documents.”

What Does ChatGPT Have To Say?

My colleague, John Oncea, the editor of RF Globalnet and Photonics Online, recently wrote an article called Is Artificial Intelligence Coming For Your Job? In it, he lists the response from ChatGPT-3.5 to the question “Is artificial intelligence going to replace RF engineers.” I thought I’d ask a similar question for all the folks who deal with data analysis in clinical trials: “Is artificial intelligence going to replace data analysts in clinical trials?” Here’s the response it gave:

Artificial intelligence (AI) has the potential to significantly augment and streamline many aspects of clinical trials, including data analysis. However, it's unlikely that AI will completely replace data analysts in clinical trials anytime soon. Here are a few reasons why:

Complexity of Data: Clinical trial data can be highly complex, involving various types of data such as genomic data, imaging data, patient-reported outcomes, and more. While AI can automate certain aspects of data analysis, human data analysts are often needed to interpret results, ensure data quality, and make critical decisions based on the context of the study.

Regulatory Compliance: Clinical trials are subject to strict regulatory requirements, and data analysis processes must adhere to these regulations. Data analysts with expertise in regulatory compliance are essential for ensuring that analyses meet regulatory standards and that the results are valid and reliable.

Interpretation and Context: AI can be powerful for processing large datasets and identifying patterns, but it may struggle with understanding the broader context of the data or considering factors that are not explicitly present in the data itself. Human data analysts bring expertise in the field of medicine and can provide valuable insights that AI may overlook.

Ethical and Social Considerations: Clinical trials involve human subjects, and ethical considerations play a crucial role in the design, conduct, and analysis of these trials. Data analysts often need to consider ethical implications when interpreting data, which may require human judgment and empathy.

While AI can enhance the efficiency and effectiveness of data analysis in clinical trials, it is more likely to complement the work of data analysts rather than replace them entirely. Human oversight and expertise remain essential for ensuring the integrity, validity, and ethical conduct of clinical trials.

I’m guessing none of this response is surprising to anyone who has been following AI in clinical trials. But I felt it bears restating that this technology is far from a panacea, and should, instead, be viewed as a partner that complements some of the tasks any data analyst/manager is faced with.