PCA statistical method for classification of sensors

• Ano de defesa: 2022
• Autor(a) principal: Paula, Jessica Fernandes de [UNIFESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de São Paulo
• Palavras-chave em Português: PCA; HIV; HCV; sensor; antibody; antigen; statistics
• Link de acesso: https://repositorio.unifesp.br/handle/11600/67178

A large number of the global population suffers from infectious diseases, so studies in the health area aimed at identifying these diseases are of great importance. Some diseases have a long immunological window, where antibodies take a long time to be identified. Rapid detection tests are essential for disease control and eradication. A possible identification and classification method uses the statistical analysis performed by the Principal Component Analysis (PCA), through which we can reduce the number of variables and identify the presence of these antibodies. This work aims to classify immunosensors according to the antibody detected, analyzing their responses in relation to impedance and frequency using the PCA statistical method. The study was based on data collected from two immunosensors, HCV sensor and HIV sensor (Hepatitis C virus and Human Immunodeficiency Virus), analyzing their response as a function of frequency. For the PCA statistical method, an interactive laboratory was adopted with Jupyter Notebook, Python, using libraries known as Pandas, Plotly, NumPyand Scikit-learn. This study analyzed several data and variables from the dataset of both sensors to build models with the PCA statistical method, it was possible to separate and classify the HIV and HCV sensors at specific frequencies. The PCA analysis results for the selected datasets showed a relevant classification using PC1 and PC2, with a variance of the original data above 90%.