Artificial Intelligence (AI) can augment productivity in clinical trial delivery and more, says Piotr Maslak, Director of Operational Data Intelligence at AstraZeneca.
Maslak was speaking on the first day of the 15th Annual Outsourcing in Clinical Trials Europe conference, where he delved into ways to implement AI in clinical trials, and then measure success, using practical examples.
Go deeper with GlobalData
Maslak outlined five practical applications of AI in clinical trials; patient recruitment and retention, protocol optimisation, real-time data monitoring, predictive analytics as well as operational efficiency and resource optimisation. He stressed that the quality of data in AI implementation is the most important thing, as without it, the AI tools prove to be ineffective.
Patient recruitment is an evergreen topic in the industry and using AI, sponsors are able to reach “patients faster, more effectively with better quality” said Maslak. In the context of clinical trial protocol optimisation, sponsors can utilise AI to create trial simulations using protocols and data from past trials to influence trial designs. For instance, adding a certain endpoint to a trial may delay recruitment and ultimately the completion of the trial. AI can be leveraged in the analysis of such historical trial data to identify trends and operational success factors as well as to tailor the design to reach the right demographic or to tackle the unique heterogeneity of a disease. With real-time data monitoring, companies can—without regulatory or quality risk— delegate some of those data monitoring tasks to AI.
AI can also be used to enhance the patient experience, and even predict retention risks and dropout rates by analysing the incoming data. According to Maslak, predictive analytics can be used for safety signal detection or to forecast treatment response rates and potential adverse events. He went on to say that while AI can currently help on a statistical level to predict treatment response, it cannot do so on an individual level.
The feasibility process is—and for foreseeable future—will be human-driven, but AI has the potential to provide support, through augmenting site selection, deploying resource allocation forecasting, measuring performance, controlling progress and even monitoring and managing supply chain inventory, which is very difficult, said Maslak.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataAI implementation projects in clinical trials
In order to navigate the uncertainty in AI implementation and keep up with the continuously evolving AI advancements, Maslak advised to show the value of AI quickly by building a minimum viable product and engaging subject matter experts early to receive feedback from them. He also recommended a phased rollout to a target audience with a smaller scope in the beginning to ensure easier product refinement and continuous improvement. Maslak also emphasised the importance of introducing agile management principles, such as the need to be driven by defined priorities and objectives based on stakeholder input. AI implementations need to be rapidly developed and improved through continuous integration and testing, he said.
Lastly, taking regulations around AI into consideration for planning is also key. “Regulatory agencies have still not fully embraced the new technology that is coming,” leading to gray areas, said Maslak. According to him, regulators do not explicitly state what applications of AI can and cannot be used, and it is important to align with the relevant data privacy laws and the enterprise’s AI governance stance to safeguard ethical technology use. Additionally, employee and target audience literacy and upskilling will be key to overcoming hurdles and developing universal skills.
The 15th Annual Outsourcing in Clinical Trials Europe took place from 29-30 April, in in Barcelona, Spain.
