Intelligent Information System for Extracting Knowledge from Pharmaceutical Package Inserts

Abstract

Pharmaceutical package inserts are a rich source of information about medicines. To guide the patient, health professionals need information about the appropriate medication for an illness. This information can be found in the pharmaceutical package inserts. Extracting information from package inserts manually is a challenging task, especially when information is needed quickly and efficiently. Doubts about adverse reactions or interactions with others drugs are common. Automatically extracting information from package inserts can help health professionals to make decisions about therapies and drug prescriptions. This article describes the creation of an Artificial Intelligence model, using a hybrid approach called CRF+LG, which recognizes named entities of medicines, diseases and people in package inserts. The model was tested on two sets of package inserts: for stomach pain and diabetes treatment. This work was developed under the aegis of Soft Systems Theory. This research has a prescriptive character and its evaluation was carried out through the execution of experiments. The analysis of the results was carried out with a quantitative approach. The experiments showed that the model obtained, of measure F, 82.08% in the recognition of entities related to diseases, 59.14% of medicines and 94.26% of people. The main contribution of the article is the creation of a model that automatically recognizes entities named in pharmaceutical package inserts. This model can integrate an Intelligent Information System to assist health professionals in making decisions about therapies and drug prescription.