Everyone seems to be talking about integrating Artificial Intelligence (AI) into both personal and professional spheres. Smart tools are being developed to handle all kinds of tasks, especially where there’s access to substantial data. When processes become time-consuming and costly, the demand for AI solutions grows even stronger. In this article, we’ll explore how AI tools should be developed so that they could be beneficial specifically to the approval process under the European Food Safety Authority (EFSA) and how applicants might leverage AI tools to streamline and expedite the preparation of dossiers.
Experts widely recognise that obtaining authorisation for feed additives, novel foods, traditional foods, and similar so called regulated products is not only time-consuming but also costly. The reasons are clear: consumers and manufacturers alike demand food for both people and animals that is safe and risk-free. Additionally, laws require that products are produced in ways that pose no risk to manufacturers or the environment (including animals), with strict monitoring in place. To assess these factors, numerous studies must be conducted and then reviewed through multiple, lengthy stages – resulting in a significant workload for regulatory authorities and applicants.
For the applicant of a new regulated product (e.g. a feed additive), there has been further complications since 2021 in addition to the already very time-consuming preparation of a dossier. The reason for that, was that the Transparency Regulation (EU) 2019/1381 came into force on March 27th 2021. The aim of this new regulation was to give consumers, authorities, policymakers, scientists, non-governmental organisations, and industry the opportunity to access and review publicly available scientific data on ‘Open EFSA’. However, applicants for new regulated products have the right to keep certain parts of their dossier confidential. To exercise this right, each confidential section must be individually identified, with a justification for confidentiality provided for each request before submission. After submission, new discussions frequently emerge with the confidentiality assessment team regarding nearly every ‘blacked-out’ section. Ultimately, the final non-confidential version of the dossier is made publicly available, accessible to everyone – including competitors.
To reduce the significant workload, both applicants and the regulatory authority EFSA, would greatly benefit from smart solutions that streamline dossier preparation and review – while ensuring that all safety and confidential questions are thoroughly addressed. But for utilising any kinds of machine learning data needs to be provided. Looking at the huge amount of data sources (research studies, literature, consumer feedback) that is being presented to EFSA by every submitted dossier, it should be possible to use AI solutions, like EFSA stated in its theme (concept) paper about the usage of AI in context of evaluating the risk assessment: ‘AI will reduce workload, gather and analyse more evidence, and improve the quality of risk assessments’. As a possible impact for EFSA and its partners the following can be quoted: “Data is at the core of AI Applications: as such, any kind of labour where resources are spent to take in and process information to support decision making or recommendations can be replaced by AI”. Gathering and analysing information to support decision making is at the heart of EFSA’s work, as well as that of other regulatory agencies. Therefore, the impact of AI on EFSA’s activities will be substantial. But implementation of AI support into the authorisation work of EFSA needs to be well prepared.
With the integration of AI tools into the approval process, planned for 2027, questions arise about how the new system might accelerate approval times and benefit applicants. Alternatively, could other AI tools designed specifically for applicants further enhance the process’ efficiency?
