Objectives The increasing rate of publications in medicine is a current concern. Find relevant and precise information with impact in patient care decision became complicated, making the process of relevant knowledge extraction difficult and time-consuming. In the guideline creation process the information overload synthesis is the core step in the process. In Latin America countries there is need for more efficient tools which enable the quality-based discovery, the extraction, the integration and the use of hidden knowledge in the publications, due to the limited trained personal. Text mining tools helps to cope with the interpretation of these large volume of data. To include the raw data in the publications and convert them to a format feasible of data mining could decreased the analysis time. This strategy should improve the actualization of the recommendations with instant review of the data.

Method We conduct a pilot study comparation in a guideline develop (neonatal Sepsis) process for data extraction with the traditional strategy (using research algorithm, manual review and GRADE pro tool) versus Data mining strategy (PubMed Reminer, Trip database tool, GATE, Tabula 1.2.1, Meshy, Weka and free existing tools) and comparing resources, time (learning and implementing) and with the guideline recommendations as measurable results.

Results There was a consistency in results during the evidence extraction process, but increased discrepancy in the data extraction between the traditional strategy and the text and data mining strategy (Kappa < 0.20)

Conclusions Significant efforts are being made to personalize information retrieval and ranking process steps during a guideline creation. The use of text mining tools could reduce the costs and error in handling information. Nowadays the free tools are insufficiently precise in the extraction of text and data that does not replace the traditional method of conducting a clinical practice guide.