Quantificação da glicemia através de análises de imagens da íris humana utilizando redes neurais

Alves, Deise Mota

doi:10.11606/d.18.2007.tde-20112007-104513

Cited by 4 publications

(7 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It is highly recommended that an ANN be trained with large experimental databases [16,17], so that it can lead to a proper convergence at its output. Once extensively trained and calibrated, ANN can then be used to estimate glucose levels with a given input data from outside the training set [18].…”

Section: Introductionmentioning

confidence: 99%

Analytical Model for Blood Glucose Detection Using Electrical Impedance Spectroscopy

Pedro

Marcôndes

Bertemes-Filho

2020

Sensors

View full text Add to dashboard Cite

Pathogens and adulterants in human feeding consumables can be readily identified according to their electrical properties. Electrical bioimpedance analysis (BIA) has been widely used for body contents characterization, such as blood, urine, lactate, and sweat. If the concentration of glucose in blood alters the electrical properties of the blood medium, then the impedance spectrum obtained by BIA can be used to measure glycemia. For some applications, artificial neural networks allow the correlation of these parameters both impedance and concentration of glucose by means of symbolic and statistical rules. According to our literature review, there is not any physical model that allows the interpretation of the relationship between blood’s electrical properties from impedance spectra and the concentration of glucose in blood plasma. This article proposes a simplified physical model for blood electrical conductivity as a function of concentration of glucose, based on Bruggeman’s effective medium theory. The equations of this model were obtained considering an insulating phase distribution diffused in a conductive matrix, in which red blood cells are represented by macroscopic insulating nuclei and glucose molecules by microscopic insulating particles. The impedance spectrum for different glucose concentrations (4.0 to 6.8 mmol/L) in a blood sample, published by Kamat Bagul (2014), were compared to the proposed model. The results showed a significant correlation with the experimental data, showing a maximum error of 5.2%. The proposed model might be useful in the design of noninvasive blood glucose monitoring systems.

show abstract

Section: Introductionmentioning

confidence: 99%

Analytical Model for Blood Glucose Detection Using Electrical Impedance Spectroscopy

Pedro

Marcôndes

Bertemes-Filho

2020

Sensors

View full text Add to dashboard Cite

show abstract

“…It is highly recommended that a ANN be trained with large experimental databases, [12][13] so it can lead to a proper convergence at its output. Once extensively trained and calibrated, ANN can be applied to estimate glucose levels with a given input data from outside the training set [14].…”

Section: Introductionmentioning

confidence: 99%

Analytical Model for Blood Glucose Detection using Electrical Impedance Spectroscopy

Pedro¹,

Marcôndes²,

Bertemes-Filho³

2020

Preprint

View full text Add to dashboard Cite

Pathogens and adulterants in human feeding consumables can be readily identified according to their electrical properties. Electrical bioimpedance analysis (BIA) has been widely used for body contents characterization, such as blood, urine, lactate and sweat. If the blood glucose concentration alters the electrical properties of the blood medium, then the impedance spectrum obtained by BIA can be used to measure glycemia. In some applications, artificial neural networks allows the correlation of these parameters (impedance and glucose concentration) by means of symbolic and statistical rules. According to our literature review, there is any physical model that allows the interpretation of the relationship between blood’s electrical properties, obtained by BIA, and the concentration of glucose in the blood plasma. This article proposes a simplified physical model for blood electrical conductivity as a function of glucose concentration, based on Bruggeman’s effective medium theory. The equations of this model were obtained considering an insulating phase distribution diffused in a conductive matrix, in which red blood cells are represented by macroscopic insulating nuclei and glucose molecules by microscopic insulating particles. The impedance spectrum for different glucose concentrations (4.0 to 6.8 mmol/L) in a blood sample, published by Kamat Bagul (2014), were compared with the proposed model. The results showed a significant correlation with the experimental data, showing a maximum error of 5.2%. The proposed model might be useful in the design of noninvasive blood glucose monitoring systems.

show abstract

“…Segundo Alves (2007) No passo forward, valores são aplicados à camada de entrada, são multiplicados e propagados para a frente, camada por camada, até a camada de saída, produzindo um conjunto de valores de saídas como resposta da rede. Após a propagação (passo forward), a resposta é comparada ao valor desejado pelo usuário da rede.…”

Section: Treinamento Das Redes Pmcunclassified

“…Este ajuste dos pesos é executado de acordo com uma taxa de aprendizado, de tal modo que se for muito pequena irá convergir muito devagar; se for muito grande, irá saltar e nunca irá convergir (ALVES, 2007). • Momentum Esses parâmetros são ajustados empiricamente, ou seja, a cada simulação podem-se testar novos valores.…”

Section: Treinamento Das Redes Pmcunclassified

“…Isso limita as utilizações práticas deste algoritmo, permitindo apenas o treinamento de pequenas redes, com poucos milhares de pesos ajustáveis. Embora alguns problemas do mundo real (problemas práticos) possam ser tratados utilizando redes desse tamanho, a maioria dos problemas para os quais a tecnologia neural poderia apresentar uma solução adequada, demandariam redes maiores e mais complexas (ALVES, 2007).…”

Section: Normalização Dos Dadosunclassified

See 1 more Smart Citation

Rede neural artificial para monitoramento em tempo real da concentração de potássio na vinhaça in natura

Souza¹

View full text Add to dashboard Cite

Aos meus amados e queridos pais, Paulo e Beatriz, obrigado por todo o carinho, suporte, presença, paciência que tiveram não só no período do mestrado, mas também durante toda a minha vida, pois, sem tudo o que vocês me proporcionaram, eu não teria chegado até aqui. Muito obrigado! Ao meu irmão Luiz pela grande amizade, pelos momentos divertidos e também por estar sempre presente em minha vida. Ao Prof. Dr. Dennis Brandão, pelo apoio, orientação nesta pesquisa e amizade que perdura desde os tempos de graduação. Fica aqui um grande abraço e sinceros agradecimentos por tudo. Obrigado por tudo.

show abstract

Quantificação da glicemia através de análises de imagens da íris humana utilizando redes neurais

Cited by 4 publications

References 13 publications

Analytical Model for Blood Glucose Detection Using Electrical Impedance Spectroscopy

Analytical Model for Blood Glucose Detection Using Electrical Impedance Spectroscopy

Analytical Model for Blood Glucose Detection using Electrical Impedance Spectroscopy

Rede neural artificial para monitoramento em tempo real da concentração de potássio na vinhaça in natura

Contact Info

Product

Resources

About