Current Edition

The Trial Master File (TMF)

IPI speaks with Aaron Grant, VP of Solutions Consulting at Phlexglobal, (a PharmaLex Company)
on how managing the TMF requires following best practice and taking advantage of digital
innovation such as AI

The Trial Master File (TMF) is the key to being able to evaluate the conduct of a clinical trial. It documents that the trial was run in a way that met all necessary regulatory requirements and conformed with Good Clinical Practices (GCP). Managing it requires following best practice and taking advantage of digital innovation such as AI, as Aaron Grant of PharmaLex explains.

Q: Can we start with an overview of PharmaLex’s best practices approach to the TMF?

A: Having a high-quality TMF is a priority since poor documentation can result in critical inspection findings.1 A best-practice approach to the TMF means carrying out quality control of documents, conducting oversight of clinical research organisations, and doing TMF completeness checks. The challenge is these are resource-intensive activities and because few companies have the resources to QC 100% of TMF documents, the practice has been to take a risk-based approach in accordance with regulatory guidelines.2

Q: Recently artificial intelligence has become a more widely used tool in industry. How can it help to enhance the management of the TMF?

A: Primarily, AI is becoming more available and accepted as a potential solution for some of these problems. People in industry are getting more familiar with the terminology around AI, what it does and why. At a high level, what AI does is remove the challenge around resources and allows companies to conduct a more thorough QC of the TMF. AI makes it possible to do complete review, and much quicker than a human can. You still need people to do those completeness and quality checks, but AI means you don’t have to rely on a risk-based approach and instead can carry out a thorough assessment without the need for additional human resources.

Q: There are obviously many ways in which AI can be leveraged to support the TMF. Can you take us through some of those key steps.

A: There are four key takeaways here – document classification, document quality checks, metadata extraction, and risk-based QC.

First, the TMF Reference Model established a common standard for classifying and organising documents.1 The challenge is that there are, on average, more than 400 sub-artifacts within the TMF, making it very difficult for document owners to know where to put the document in the classification. AI, in essence, acts like an expert looking over your shoulder and offering prompts on document classification.

During document quality checks, a lot of time is spent looking through the content for blank or missing pages, legibility issues, inconsistent rotation of the document pages and other such issues. AI can be used to scan the documents and find these issues, which is a huge time saver.