Modelling the influence of pH and temperature on the response of an acetylcholinesterase biosensor using Machine Learning Methods

Edwin R. Garcia Curiel, Larysa Burtseva, Margarita Stoytcheva, Felix F. Gonzalez-Navarro, Ana S. Estrella Sato


This work resolves the doubts found in the literature about learning of biological functions provided by the use of electrochemical sensors. Keeping the virtues of these, a detailed process is presented in the development of different learning models, and the procedure for the evaluation of results. Experimental data are analyzed through different perspectives (Cs, pH, T), also their behavior and location of the possible areas of the extreme values in the resulting current, with the aim of finding the combination of parameters that maximize the sensitivity of the determination of acetylcholine. According to the analysis of the state of the art, the Artificial Neural Networks are the most suitable regression models to the task at hand, i.e. to predict and analyze the parameters of a biosensor. In the same way, the Support Vector Machines offer a solid and competitive performance with the possible advantage that it has less tuning parameters than a Neural Network. Although the results obtained using Neural Networks were satisfactory, the Support Vector Machines show the best performance, i.e. function approximation, both in testing and simulation modeling process.

Palabras clave

acetylcholinesterase biosensor; machine learning; neural network; support vector machine; model; optimization

Texto completo: PDF


  • No hay Refbacks actualmente.

OmniaScience, 2011-2018 -