Validating the Future through Computerised Systems Validation (CSV) in Life Sciences

Compliance — In the era of Manufacturing 4.0, the life sciences industry faces unprecedented challenges in validating computerised systems. Their validation is an essential process that ensures compliance with healthcare regulations, such as those enforced by the Food and Drug Administration (FDA) or the European Medicines Agency (EMA). This process is crucial for healthcare companies, as it guarantees that the software and hardware used in the manufacturing of pharmaceutical products, medical devices, and biotechnologies work as intended and present no risk to patient safety. 

“Compared to traditional systems, AI is like a
black box, and we have limited knowledge about its inner workings.”

LILIAN VEYSSEYRE /
COMPUTERISED SYSTEMS VALIDATION ENGINEER – CADUCEUM

What is your expertise? 

— My expertise is in Computerised System Validation (CSV). It is vital as it represents the digitalised version of the Good Documentation Practices which ensures traceability and reproducibility of production processes, analytical results, and more widely, all data that guarantees patient safety and drug quality. 

What are the main trends and challenges within that expertise? 

— Digitalised workflows come with many advantages: greater availability of data, easier inspections for authorities, and simplification of processing, all which contribute to the prevention of errors and the improvement of data processing time. The challenges usually lie in the transcription of the user requirements into logical rules and computer code. Data accessibility also implies that data centres have to be adapted, resilient, and penetration-proof as data security is as crucial as data accuracy. 

In terms of CSV, how do you see the life sciences industry evolving in the next 10 years? 

— I believe that the next challenge for CSV will involve handling Artificial Intelligence. AI generates a large amount of data based on specific prompts. As CSV specialists, our main challenge is to validate the AI, understand how it operates, and how it processes the data and information we provide. Compared to traditional systems, AI is like a black box, and we have limited knowledge about its inner workings. Some clients are interested in using AI, but there is still a lack of understanding when it comes to validation. Currently, AI is widely used for simple tasks based on prompts. However, as the prompts become more complex, it becomes increasingly difficult for experts to validate, as we currently lack an understanding of how these “black boxes” operate. 

Since we are talking about patients’ lives, how can we guarantee better traceability and reliability of their data? 

— In terms of data reliability, using open-source code is a good practice, but not always one chosen by companies. They often have proprietary codes and patents in place, in order to ensure patient data remains internal. Patient safety should always remain a top User Requirement Specifications (URS) priority. In the end, we must always think about how to consider patient data in our requirements. 

What can we expect from the Internet of Things and what is its impact on medical interventions? 

— It really depends on the devices. We have connected watches,for instance, that can pick up on abnormalities in cardiac rhythm at an early stage. We are seeing devices that are used to automatically dispense medicinal drugs. With the help of IoT, we can rely on connected devices that indicate whether a drug was dispensed correctly, if any problems were encountered during the intervention, and even the patient’s previous medical record when prescribing medicine. 

TO GO FURTHER