Artificial intelligence in the clinical pharmacy service in a public hospital in Belo Horizonte/MG

Abstract

Objective: to evaluate aspects related to the analysis of prescriptions by clinical pharmacists and the rate of medication-related errors after the implementation of an AI tool for the analysis of medical prescriptions in a large public teaching hospital in the city of Belo Horizonte/MG. Method: This is an observational study in which the results of the analysis of medical prescriptions performed in two periods were verified: the first (denoted BEFORE), period previously to the use of the AI tool (NoHarm.ai), in the months of March to September 2021; the second (named AFTER), comprises the same period in 2022, already in use of the AI tool. Results: In the BEFORE period, it was found that the rate of prescriptions evaluated was 0.6%, with an error rate of 13% and an average of 85 pharmaceutical interventions/month, which resulted in average savings of direct medication costs of R$1020.76/month. In the AFTER period, there was a 49% evaluated prescription rate and a 0.3% error rate and an average of 239 pharmaceutical interventions/month, with an estimated savings of R$ 7848.39/month. Conclusion: The use of an AI tool contributed substantially to the pharmaceutical analysis of medical prescriptions with an average increase of 50% in the prescriptions evaluated, a 43-fold reduction in the number of errors and generated almost triple the number of pharmaceutical interventions after the implementation of the tool, in addition to the direct savings obtained with these interventions that increased sevenfold. The results of this study show that the use of an AI tool probable save of financial resources, increased productivity of the Clinical Pharmacy Service, and increased safety related to the medication use.