Improving Medication Identification Accuracy and Regulatory Compliance through NLP and Ontologies: An Analysis of Otorhinolaryngology Prescriptions

This research presents an innovative method to improve the identification and regulatory compliance of medications in free-text medical prescriptions in Brazil. By integrating Natural Language Processing (NLP) and ontological frameworks, the study aims to enhance the accuracy of medication identification and ensure alignment with the Brazilian National List of Essential Medicines (RENAME). The research employs a systematic ontology creation methodology, adopting the Web Ontology Language (OWL) and utilizing OpenRefine for data extraction, transformation, and loading. The ontology, named OntoDrug, integrates terms from Schema.org for better interoperability and includes detailed properties relevant to medications, such as active ingredients, dosages, and drug classes. The study achieved a high medication identification success rate of $\mathbf{9 4. 5 \%}$, with 61% complete recognition and 33.5% partial recognition. However, $\mathbf{5. 5 \%}$ of prescriptions were not recognized, indicating areas for improvement. The results demonstrate that the integration of NLP and ontologies significantly improves the linkage between unstructured prescription texts and structured compliance verification processes. The research shows significant advancements in handling medical prescriptions through the integration of NLP and ontological frameworks. It enhances the accuracy of medication identification, regulatory compliance, and overall efficiency of EHR systems. The study highlights the potential for broader application across various medical fields and underscores the need for continuous improvements in medical informatics to further enhance patient care and streamline medication management.