Results of ECA Survey about Artificial Intelligence in Equipment Qualification

Now, one of the hottest topics is the discussion on the use of artficial intelligence (AI) in the field of GMP. How is AI used in equipment qualification? This is a question the ECA wanted to clarify with a survey. The survey contained 8 questions. In total, 60 participants took part in the survey. In the following, you will find a summary of the questions asked and their answers. Not each participant has answered every question, and sometimes several answers were given. This means, that the percentages may add up to more or less than 100%. Here are the results.

The majority of the 60 participants came from the pharmaceutical industry (38%), 18% are involved in biologics, 8% work in the area of  APIs and 5% in the area of excipients.

The second biggest group are the "others" (20%). Here, the participants are coming from: 

• Training and consultancy for pharma industry
• Chemical APIs and Medicinal Products
• Formulation
• Engineering Consultant
• Cannabis
• Providing validation service for the pharmaceutical industry
• Inspection software for drugs (packaging and product integrity)
• Poultry Vaccines
• Quality control laboratory, R&D and Consultancy Services
• Medical devices and Medicinal Products
• Consultancy
• Software such us: Learning Managment System or developed solution to paperless validation in the GxP environment

43 % of the 60 participants come from companies with more than 500 employees, 25% of the participants are employed in companies with 101 -500 employees. The remaining participants are divided between companies with 1 - 50 (17%) and 51 - 100 employees (15%).

The ECA wanted to know, if the participants have anything to do with AI. The clear majority of 47% has nothing to do with AI. 38% have experimented with AI and 15% are already using AI.

The ECA wanted to go more in detail about AI and asked "Do you already use AI in the GMP environment?". 83% answered no. But 16% are using AI, e.g. for

• Batch record review and equipment qualification protocotol and report
• For input to e.g. deviation, conclusion and text (however still being verified by the employees for correct GMP)
• Writing protocol and analyzing validation data
• Within a structured reviewed approved and secure environment 
• High level requirements engineering  
• Support to line clearance

Next the ECA asked where AI is already being used in in the participants´ environment. Consistent with the answers to the previous question the majority is not using AI (53%). But 27% are using AI for writing texts. 13% use it as an advanced search tool and 5% use AI for evaluation purposes.

Very interesting are the answers to the following question "How do you process the qualification?" Still 60% of the participants answered that they process it in a purely paper-based way (word files). Hybrid systems are used by 31% and software-based solutions (IT-tools) are used by 9% of the participants.

The ECA asked, if the participants could "imagine a meaningful use of AI in qualification and if so, for what purposes?" The majority sees AI as a possibility that can be used for evaluation and reporting (38%). The next group says it can be used to create entire protocols (27%). And the third group says it can be used for the definition of test programs (20 %). But there is also sceptisism. 12 % don´t see any meaningful use of AI at all at the moment.

The estimations of the participants regarding the question "How long do you think it will take for AI to fully cover the area of qualification are also very intersting" 50% see a possibility for a short term implementaton within 3-5 years. 42% see a possibility for long term  implementation (10 years of the earliest) of AI. There are also some comments:

• 5-10 years (2x)
• Probably never, because AI will always lack the specific business context
• I don not see that AI could correctly FULLY cover qualifications
• It may take longer than 3-5 years for the benefits of the technology to be recogniszed and accepted by the majority of the organisation 

Conclusion: AI is not widely used yet. Nevertheless, there are first applications used,  but less in equipment qualification projects. But 50% of the participants expected that AI will fully cover the area of qualification within 3-5 years.

This year  ECA´s Equipment Qualification Forum takes place in Heidelberg, Germany, on 05/06 November  2024. As one topic, you will discuss and see in a workshop how to use AI for writing qualification protocols. An EU Inspector will give his comments regarding this.