Contribuição para o desenvolvimento de um software de diagnóstico baseado em sensores magnetoelásticos

Abstract

In recent decades, there have been several epidemics of deadly viruses with significant impact on global public health. Examples include Ebola, Marburg, Nipah, Zika, Crimean-Congo hemorrhagic fever, Dengue, and more recently, the SARSCoV-2 coronavirus, which caused the COVID-19 pandemic. The rapid spread of these viruses and the lack of adequate healthcare infrastructure in some regions have made it difficult to contain the epidemics. Global mobilization was necessary to contain the spread and provide adequate medical assistance to affected individuals. Rapid detection and the development of effective drugs and vaccines are crucial to combating future epidemics. Sensors made from magnetoelastic materials and machine learning models can be useful in this endeavor. In this work, the analysis of a dataset collected from a network analyzer using magnetoelastic sensors was employed to train a machine learning algorithm. A classification model was developed to accurately diagnose the presence of pathogens in test sensors, with the capability of exporting the model for use with external data. The developed model was able to distinguish between control and test data, demonstrating precision in identifying the presence of pathogens. The application of machine learning enabled the detection of small variations in the sensor signals, facilitating its use for diagnostic purposes. The amount of data available for model training partially limited the depth of the analysis; however, it did not compromise the applied technique. [resumo fornecido pelo autor]