Neural Network Classification Method for Solution of the Problem of Monitoring Theremoval of the Theranostics Nanocomposites from an Organism

A4 Conference proceedings

Internal Authors/Editors

Publication Details

List of Authors: Sarmanova O, Burikov S, Dolenko S, von Haartman E, Sen Karaman D, Isaev I, Laptinskiy K, Rosenholm JM, Dolenko T
Editors: Samsonovich Alexei V, Klimov Valentin V.
Publisher: Springer, Cham
Publication year: 2018
Publisher: Springer
Book title: First International Early Research Career Enhancement School on Biologically Inspired Cognitive Architectures
Title of series: Advances in Intelligent Systems and Computing
Volume number: 636
Start page: 173
End page: 179
ISBN: 978-3-319-63939-0
eISBN: 978-3-319-63940-6


In this study artificial neural networks were used for elaboration of
the new method of monitoring of excreted nanocomposites-drug carriers
and their components in human urine by their fluorescence spectra. The
problem of classification of nanocomposites consisting of fluorescence
carbon dots covered by copolymers and ligands of folic acid in urine was
solved. A set of different architectures of neural networks and 4
alternative procedures of the selection of significant input features:
by cross-correlation, cross-entropy, standard deviation and by analysis
of weights of a neural network were used. The best solution of the
problem of classification of nanocomposites and their components in
urine provides the perceptron with 8 neurons in a single hidden layer,
trained on a set of significant input features selected using
cross-correlation. The percentage of correct recognition averaged over
all five classes, is 72.3%.

Last updated on 2020-17-02 at 04:14