Plataforma portátil de bioensaio eletroquímico para rastreamento de hepatite delta baseada em aprendizagem de máquina e nuvem

• Ano de defesa: 2024
• Autor(a) principal: Souza, Pedro Victor Eugênio de
• Tipo de documento: Tese
• Tipo de acesso: Acesso embargado
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Ciências da Saúde
• Palavras-chave em Português: Biossensores Eletroquímicos; Potenciostato Portátil; Análise Eletroquímica; Aprendizado de Máquina; Detecção de Hepatite Delta; Electrochemical Biosensors; Portable Potentiostat; Electrochemical Analysis; Machine Learning; Delta Hepatitis Detection; CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA::ENGENHARIA MEDICA::TRANSDUTORES PARA APLICACOES BIOMEDICAS; Ciências Médicas; Algorítmos computacionais; Inteligência artificial - Aplicações médicas; Infecção delta; Hepatite; ODS::ODS 3. Saúde e bem-estar - Assegurar uma vida saudável e promover o bem-estar para todos, em todas as idades.
• Link de acesso: https://repositorio.ufu.br/handle/123456789/44058; http://doi.org/10.14393/ufu.te.2024.5052

Introduction: Electrochemical biosensors are proficient in detecting a wide range of analytes within various biofluids. Traditional benchtop potentiostat systems, however, are often costly and lack full integration. This research focuses on the development of a novel, fully integrated electrochemical platform that features a portable and cost-effective potentiostat. This device is designed to simultaneously analyze two electrodes in real time across five distinct electrochemical analyses, and includes communication interfaces such as USB, Bluetooth, and Wi-Fi. Such capabilities enable seamless interaction with computers, mobile devices, and cloud-based environments, facilitating the application of Artificial Intelligence algorithms via AWS cloud for electrochemical data analysis and sample classification. Materials and Methods: The platform, named Bioconnect, was validated against the benchtop potentiostat EMSTAT3+ Blue from PalmSens. The validation involved Cyclic Voltammetry using a [K3Fe(CN)6/K4Fe(CN)6] solution with both Rhodamine 6G and Carbon Nanotube-modified electrodes, as well as non-modified electrodes. Furthermore, Bioconnect's integration with an AWS-based machine learning application was explored for the detection of hepatitis delta, a severe form of hepatitis traditionally diagnosed through chemiluminescence and RT-PCR methods. The biosensor was evaluated using a cohort of 40 subjects (20 positive and 20 negative) for hepatitis delta. Results: The performance comparison, based on the anodic and cathodic current peak measurements, demonstrated that Bioconnect achieved smaller standard deviations compared to the EMSTAT3+ Blue, indicating superior precision in measurements. With respect to distinguishing hepatitis delta using a Univariate Logistic Regression analysis yielded a specificity of 89.5% and a sensitivity of 85.7%, while a machine learning algorithm achieved a specificity of 80% and a sensitivity of 100%. This underscores the potential of Bioconnect as a precise, integrated tool for advanced electrochemical analysis and disease detection. Conclusion: Results confirm the efficacy of the platform and using it not only biological applications, but also for other types of applications. The comparison shows that the platform hardware has more precise measurements, and its use to distinguish normal subjects from those who are positive for hepatitis delta proves its applicability to be used as a Point of Care device. Besides that, the use of machine learning algorithms also proved a valuable tool to enhance electrochemical analysis.