No abstract

No abstract

This work presents ave used of machine learning algorithms for data analysis and generation of intelligent response of a sensor based on surface plasmon resonance (SPR). The sensorgrams obtained by the real-time response of an sensor SPR, were used as a database. The sensorgrams were initially described with the discrete cosine transform; and then classified with the k-nearest neighbor (k-NN) algorithm; and the identification of the substances/areas of interest of sensorgram were performed by applying linear regression. The results presented are satisfactory, attesting that the sequence of steps chosen makes it possible to classify the sensorgrams, identify and analyze areas of interest in the sensorgram, therefore creating, a SPR sensor with intelligent response. Resumo: Este trabalho apresenta o uso de algoritmos de aprendizagem de máquina para análise de dados e geração de resposta inteligente de um sensor baseado na ressonância de plasmons de superfície (SPR). Os sensorgramas obtidos pela resposta em tempo-real de um sensor SPR, foram utilizados como base de dados. Os sensogramas foram inicialmente descritos com a transformada discreta do cosseno; em seguida classificados com o algoritmo de vizinho mais próximo (k-NN); e a identificação das substâncias/áreas de interesse do sensorgrama foram realizadas com o uso de regressão linear. Os resultados apresentados são satisfatórios, atestando que a sequência de etapas escolhidas possibilita classificar os sensorgramas, identificar e analisaráreas de interesse no sensorgrama, criando assim, um sensor SPR com resposta inteligente.

Surface plasmon resonance (SPR) sensor is a technology for analysis of biomolecular interaction, largely applied in biology and pharmaceutical research. The simulation of the surface plasmon optical excitation is an important steps of the development process for sensors based on this phenomenon. The structure, design and configuration of the desired SPR sensor benefits from previously simulated, analyzing the responses generated and check the feasibility of the materials selected for the optical coupling. Here a online web-based SPR sensor's simulator is presented. With visual-oriented interface, enable drag \& drop actions to easily and quickly modeling a variety of sensors arrangement. The embedded materials database for metals, glasses, polymers and custom substance enable flexible configuration for sensors operating in angular and spectral modes. The light propagation through the multilayer of materials is presented in terms of Fresnel coefficients, which are graphically displayed. The so-called SPR morphology parameters can be visualized. Sensor dynamic behavior could be wispy knowledge by the Sensorgram simulation. Simulation scenario's in various configurations, designs and excitation were performed and compare with other simulator. The proposed simulator guarantee comparable results with an agile and intuitive flow of execution.