Leveraging AI and digital in clinical development

Artificial intelligence (AI) is transforming drug development today by accelerating target discovery, optimizing molecules, and enabling better decision making through rapid examination of massive amounts of data. 

Recent technological advances, energized by the first foundation models including large language models (LLMs), have boosted interest, and made AI application development and implementation even more accessible.

At Parexel, we’ve been utilizing AI technologies to improve our processes on the behalf of our clients, building on our deep foundation in pharmacovigilance, we’ve expanded out AI applications into other areas. And we’re leading the way in how AI can be further developed responsibly to bring efficiencies to the drug development process. 

Parexel is committed to the development of innovative, high-value use cases for AI and digital in clinical trials. We believe AI will continue to provide tools to improve clinical trial protocols by considering comprehensive data to develop realistic eligibility criteria, enhance participant diversity, strengthen statistical power, mitigate risk, and ultimately deliver better outcomes for patients.

How we’re currently utilizing generative AI at Parexel

As part of our evolution in the utilization of AI, we’ve deployed Parexel AI Assistant, which allows us to confidentially utilize data and the cumulative knowledge at Parexel.

What areas of drug development will benefit from AI soonest?

• Study design optimization
• Patient identification and recruitment 
• Medical writing
• Medical affairs
• Data management
• Pharmacovigilance 

How is Parexel investing in AI for the future?

• Ethical frameworks and responsible AI
• Applying AI to our risk-informed approach to clinical trial planning and execution 

What efficiency gains is Parexel targeting through its AI initiatives? 

• Reduce cycle time 
• Increase accuracy in data review
• Reduce labor for highly repetitious work to redeploy valuable talent to critical task to drive better outcomes. Trend detection and correlation in data patterns. Enables earlier detection, prevention and informed decision making. 

Ethically leveraging AI in clinical trials to manage risks

Read the full PDF

1.  Thoughtful design and deployment
   
   We are guided by the need for fairness and equity, bias minimization, transparency, and a sound ethical approach. The AI solutions we design and deploy always respect these fundamental points, and we monitor them in use to ensure performance does not degrade over time.
2. Accountability and senior-level governance
   
   Given the strategic significance of AI technologies for our business and stakeholders, Parexel has created a robust system of senior-level accountability for our pipeline of developing and deployed AI solutions. 
3. Human oversight and control
   
   We understand the importance of skilled human oversight when using AI. AI is subject to certain types of errors, including LLM ‘hallucinations’ in which incorrect information is presented as fact. It is crucial, then, that application outputs are carefully reviewed, and corrected if needed, before they are accepted and used. 
4. Transparent AI
   
   When reviewing output generated by an AI application, users are made aware that the content comes from an AI system, ensuring they appreciate the need to confirm its accuracy, validity, and integrity. 
5. Regulatory conformance and legal compliance
   
   Clinical development is highly regulated, and regulatory agencies closely monitor the use of AI. While global health authorities have been receiving AI-aided submissions for years, this number has grown and will only accelerate further. 
6. Security and privacy
   
   Like all web-enabled technologies, AI applications create specific cyber and data-related risks. At Parexel, these risks are managed by our chief information and security officer as part of our overall cyber governance framework. Parexel fully complies with applicable legislation on handling such data and we ensure that our controls remain adequate in the context of our AI solutions.