AI in hospital pharmacy: an overview

Artificial intelligence (AI) has been revolutionizing the way we work and entertain ourselves for several years now. Unsurprisingly, applications of these new data analysis and production technologies are widespread in hospital pharmacies. The objectives of our presentation will first be to describe the uses of AI in hospital pharmacies by providing an overview with a focus on production applications.
Different types of AI (analytical, discriminative, predictive, prescriptive, and generative) have applications in hospital pharmacies. AI can learn to detect medical devices, read bottle labels, analyze prescriptions, identify target patient profiles for clinical pharmacy action, and analyze scientific articles. Some open-access AI models can now be installed in a private context to be trained and queried on sensitive hospital data. Among these, large language models (LLMs) are widely used by the general public with well-known interfaces (e.g., ChatGPT®, Mistral AI)®. LLMs that are more specialized in science or trained on specific datasets can assist in pharmaceutical information research, particularly during feasibility studies (e.g., Consensus.ai®, Perplexity®, Ellicit®).

The use of AI has many advantages, primarily saving time, but also limitations that must be anticipated, such as hallucinations (the creation of false information, e.g., a false scientific reference).
In conclusion, we will attempt to demonstrate the usefulness of AI in hospital pharmacy through a few examples from the perspective of pharmaceutical technology. We believe that these new technologies represent a turning point in many of our pharmaceutical activities, which will enable the profession to refocus on its added value.