Contributed Commentary by Andrew MacGarvey, Phastar 

October 20, 2023 | The importance of data has become apparent to everybody in recent years. We can now interrogate massive datasets to add value to clinical trials and utilize technology more efficiently to speed up time to market. 

At the DIA 2023 Global Annual Conference, I was particularly interested in how global industry leaders discussed exploring the opportunities offered by big data analytics, the tensions facing the industry, and how tools like machine learning and AI can be used to empower everyone involved in the development of clinical trials. 

Big Data Analytics: An Opportunity for Healthcare and Pharma 

Big datasets, combined with statistical knowledge and technology, are creating new opportunities for the pharmaceutical and healthcare industries. For example, the FDA requires diversity of patient population. Analytics can be used to interrogate historical metadata from all sites and offer insight into which should be used because they offer the diversity of population needed. 

Visualizations allow sponsors to see if a site is recruiting correctly, analyze how it is performing and enable patient profiling without the need for a programmer. While visualization products have been around for a long time, the difference now is we have access to these big datasets. Everything is lined up – the techniques, the right people trained in the right way and the data. 

This does present challenges in terms of dealing with high-volume disparate data from a variety of sources. But, of course, it is that volume of data which is giving us the insight. 

Now we need to decide: What is the question? What is possible? What can we ask now that we could not ask five years ago? That is where we will find new opportunities. 

A Holistic Approach to AI and Machine Learning 

Historically, a programmer would program 95-99% of checks but there would always be some things which were not recoverable and there would always be somebody who needed to review the data. Now, we can train a natural language program to read the data and determine whether a query should be raised. There are both supervised and unsupervised applications but, from a data perspective, it can act as an extra quality control step.  

This is not about replacing people but adapting roles and empowering them to use these tools. There are two cohorts most affected by AI adoption. The first is the human-in-the-loop. They are there to make sure what is being created by ML programs is correct, that important data points are not being missed. They are human pilots waiting to take control of the plane if the autopilot fails.  

The second cohort is people who can be empowered to deliver even more against the trial because they are plugged into these new tools. For example, we have the workflow now to guide monitors to sites that most need it. By monitoring more effectively, the industry can save money because we are catching potential issues early before they become problems. 

We need to look at whether we are being efficient and appropriate in clinical trial design and the regulatory environment. America might drive development but there are important markets elsewhere in the world that do not want to wait six months to launch successful treatments.  

The need to get treatments to market more quickly is a pressure point across the industry. Automation and technology can help ease that pressure by offering quality controls at speed for increasingly complex studies. The key is a holistic approach rather than making it the be-all and end-all.  

Analytics Versus Privacy 

There is tension between the opportunities offered by analytics and the need to protect participant privacy. Currently, a lot of time we are looking at metadata, rather than patient data. You can anonymize and you can make it generic. But to get the full value of data, there is going to have to be some sort of permissions-based approach.  

People’s attitudes to data are also changing. In the future we will see people being the controllers of their own data and choosing to either monetize that data or share it for the greater good. But we still need to be extremely careful that we have the correct permissions for who can see data, when they can see it and how it can be used.  

Another emerging area where we need to look carefully at how data is prepared, and people’s rights is the use of synthetic data. One use case for this is the replacement of subjects in a control arm for a study, using existing data collected across hundreds of trials. The FDA has provided regulatory support for this approach. Ethically it is supported because you are using data that already exists rather than bringing patients into the trial environment. 

Changing Demand for CRO Expertise 

The modality of trials is evolving and getting more complicated, which is increasing the need for specialists. The CRO industry is estimated to be worth $76.6bn this year and is projected to reach $127.3bn by 2028. 

However, a lot of sponsors want to fully leverage the complex data being generated and keep it in house. This means rather than just outsourcing, they are increasingly asking CROs to provide their expertise, login to their systems and do the work there. 

This move towards the provision of expertise as well as systems is likely to continue to drive expansion of the CRO industry, particularly for data-centric and biometric organizations. 

 
 
Andrew MacGarvey, Chief Operating Officer of Phastar, began his career in statistical programming 25 years ago. He has held various technical roles in contract research organizations, served on several boards, and has worked in both the UK and the USA in leadership roles with a focus on growing businesses. Andrew is committed to delivering the best in customer service, creating an excellent working environment, and ensuring PHASTAR gives back to the wider community through initiatives such as PHASTAR Cares. Andrew holds a Law degree from the Open University and an MBA from Newcastle University. He can be reached at andrew.macgarvey@phastar.com.