12.07.2015 Views

Predicting Cardiovascular Risks using Pattern Recognition and Data ...

Predicting Cardiovascular Risks using Pattern Recognition and Data ...

Predicting Cardiovascular Risks using Pattern Recognition and Data ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

More detail about these applications can be seen in Ripley (1996); <strong>and</strong> Lisboa (2002). The use ofpattern recognition techniques is to somehow mimic the decision making of humans. According to Tou<strong>and</strong> Gonzalez (1974), the fundamental problems in building a pattern recognition system are: The sensing problem; The pre-processing <strong>and</strong> feature extraction problem; And the determination of optimum decision procedures, which are needed in the identification<strong>and</strong> classification process.In the sensing problem, the pattern recognition system is concerned with the representation of inputdata. This representation should be chosen to aid the measurement of similarity between the currentobject <strong>and</strong> previously recognized classes. In the second problem, the system is concerned with theextraction of characteristic features or attributes from input data <strong>and</strong> the reduction of the dimension ofpattern vectors. In the third problem, the data will be formed as pattern points or measurement vectorsin feature space. By <strong>using</strong> alternative classified techniques, the system will decide to which classesthese data belong. The architecture of an archetypal pattern recognition system can be seen in Figure3.1.Possible algorithm feedback or interaction(Statistic)Observedworldpatterndata piSensor/transducerPreprocessing<strong>and</strong>enhancementFeature/primitiveextractionalgorithmClassificationalgorithm(Syntactic)DescriptionalgorithmClassificationDescriptionMeasurement, miFigure 3.1: Typical pattern recognition system architecture (Schalkoff, 1992).As an example of this pattern recognition process, cardiovascular diagnosis can be seen as follows: Theoriginal collected data is represented in the form of database files; data might be then reduced orcleaned by <strong>using</strong> alternative data mining techniques; by <strong>using</strong> feature exaction algorithms, the featurespace dimension might be reduced; the system then uses classification algorithms or description21

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!