Apoio á decisão na revisão da farmacoterapia: orientações para o aprazamento utilizando algoritmos genéticos

Introduction: Pharmacists encounter certain problems to carry out a pharmacotherapy review, mainly because of the time required to perform the activity and difficulty accessing quality information. Their job is to analyze the prescription to find possible inconsistencies and suggest a strategy for the patient to take the prescribed medicines and facilitate adherence to treatment. Objective: The aim of this study was to develop and validate an intelligent information system, by using genetic algorithms, to help scheduling how and when to take medicines appropriately. Methods: Consensus meetings were held to model, through variables and computational rules, the clinical reasoning used in the pharmacotherapy review process. The system was developed using genetic algorithms and hypothesis validation of hypothesized scheduling cases performed by the system and by human experts. The models for scheduling were evaluated qualitatively by pharmaceutical experts who had clinical and research experience in the pharmacotherapy review process. The degree of agreement between the assessments made by the human experts and the system were measured by the Kappa index. Results: The intelligent information system obtained a superior performance in all aspects as compared to that of the human experts. In the detection of purposive errors, the system was able to identify up to 80% of them, whereas the human experts identified between 20% and 70%. Regarding the general evaluation, the system achieved 87.3% of the evaluations considered adequate, whereas the human expert who achieved the highest score obtained 75.50% success. Conclusion: The intelligent information system we created, using genetic algorithms as the main resource, can help improve the quality of the pharmacotherapy revision process, being able to find prescription errors and establish schedules for medication use, according to the patient’s routine.