Computacional tool to aid in the interpretation of laboratory tests

Abstract

The interpretation of biochemical laboratory tests often presents significant challenges due to the diversity of formats and structures adopted by different laboratories. The variation in data presentation and lack of standardization often hinder the efficient analysis of these results, potentially leading to misinterpretation and imprecise diagnoses. Healthcare professionals face an obstacle when dealing with this heterogeneity, which can affect the quality and speed of the clinical decision-making process. In this context, this work focuses on developing an innovative tool to address this issue. An application was devised using advanced technologies such as Optical Character Recognition (OCR) combined with a Long Short-Term Memory (LSTM) Recurrent Neural Network. This approach enables precise reading of tests from different laboratories, regardless of the formats used. The application stands out by creating an interface that standardizes the results in a way that simplifies the interpretation of the tests, thereby facilitating the diagnostic process for these professionals. The developed model represents an effective solution to the exposed problems, overcoming the complexity arising from the diversity of formats. By automating and simplifying the analysis, it contributes to reducing interpretation errors and, consequently, improving the quality of diagnoses, positively impacting efficiency and precision in the healthcare field.