Biomedical Engineering Reference
In-Depth Information
5
Smart Biosensor Functions—A Machine Learning
Perspective
George K. Knopf
CONTENTS
5.1 Introduction ......................................................................................................................151
5.2 The Role of Machine Learning in Developing Smart Biosensor Functions ............154
5.2.1 Pattern Recognition ............................................................................................154
5.2.2 Artificial Intelligence and Neural Networks ..................................................155
5.3 Biosensor Data Analysis Using Artificial Neural Networks ......................................158
5.3.1 Sensor Calibration by Functional Approximation ..........................................158
5.3.1.1 Radial Basis Function Network ........................................................159
5.3.1.2 Multivariate Calibration Surface for a bR Photocell ......................161
5.3.2 Pattern Analysis ..................................................................................................163
5.3.2.1 Self-Organizing Feature Map ............................................................165
5.3.2.2 Pattern Classification ..........................................................................167
5.3.2.3 Pattern Association ..............................................................................168
5.3.2.4 Scientific Data Visualization ..............................................................170
5.4 Conclusions ......................................................................................................................172
References ....................................................................................................................................174
5.1
Introduction
From a historic perspective biosensors have been largely regarded as a subgroup of chemical
sensors. These devices utilize a highly selective biomolecular recognition component with a
high degree of affinity to the sample, or analyte, being investigated. The constituent compo-
nents of the sensor system transform one physical quantity as identified by the bioreceptor
element or biochemical reaction into another more easily observed secondary signal propor-
tional to a single analyte or a related group of analytes (1,2). A variety of signal parameters
such as changes in electrical resistance, conductance, potential difference, ion concentration,
current, optical properties, pH, and oxygen consumption can be measured using the appro-
priate transducer. The magnitude or frequency of the measured transducer signal will often
change with respect to the presence of the analyte or concentration of the substance being
measured. The basic stages of a biosensor system are summarized in Figure 5.1.
151
