Data-Driven Insights into AIFA’s Drug Innovativeness Evaluations Using Text Mining

Since 2017, the Italian Medicines Agency (AIFA) has published innovation assessment reports for newly authorised drugs, limited to those for which companies formally request recognition of innovativeness. These assessments provide evaluation on three items: therapeutic need, added clinical value, and quality of evidence. Beyond their regulatory role, they constitute a valuable source of information on how pharmaceutical innovation is conceptualised and communicated. However, to date, these documents have not been analysed using text mining approaches. The complete corpus of AIFA innovation assessment reports released between 2017 and 2024 has been compiled and a pre-processing phase has been applied to clean the reports. The proposed approach provides two principal tools to analyse the corpus. Firstly, a dictionary containing the most frequent words has been obtained and represented by wordclouds and networks in order to detect which are terms and topics of interest in compiling these reports. Secondly, splitting the corpus in two sub-corpora on the basis of the outcome of the assessment (approved or not approved innovativeness), some statistical models have been applied to verify the association between the final outcome and the presence (or absence) of specified terms. Finally, since the reports are available for multiple years, the evolution of the most influential terms and topics has been explored. The project is being carried out in collaboration with ISPOR Chapter Rome, providing methodological and scientific support.