Ant-inspired Algorithms in Health Information System Data Mining, Classification and Visualization

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21460%2F16%3A00307020" target="_blank" >RIV/68407700:21460/16:00307020 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/68407700:21730/16:00307020

Výsledek na webu

<a href="http://80.link.springer.com.dialog.cvut.cz/chapter/10.1007/978-3-319-32703-7_171" target="_blank" >http://80.link.springer.com.dialog.cvut.cz/chapter/10.1007/978-3-319-32703-7_171</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-319-32703-7_170" target="_blank" >10.1007/978-3-319-32703-7_170</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Ant-inspired Algorithms in Health Information System Data Mining, Classification and Visualization

Popis výsledku v původním jazyce

In the paper we describe the process of cardiotocography (CTG) record clustering and the effort that was taken to retrieve information from an old (approx. 15 yrs) hospital information system. It consisted of retrieving data from DB tables in the form of unstructured text attributes. Information retrieval had to be performed for complementing cardiotocography signals with additional information. This was needed for further rule discovery mining and automated processing as the main goal was asphyxia prediction during delivery. A graph-based visualization technique has been designed and developed. Together with experts it helped to handle and efficiently process great amount of unstructured (or semi-structured data). Of course, we have used and tested several approaches: automated, semi-automated and manual clustering of the records. In the (semi-)automated experiments we have used k-means, self-organizing map and self-organizing approach inspired by ant-colonies. The overview obtained was consulted with medical experts and served for further mining and information retrieval from the database. Furthermore, an evaluation of CTG signal classification was carried out using multiple methods. These methods used different classification approaches (Naive Bayes, rule and tree based classifiers, etc.). We conducted ten-fold crossvalidation and for each experiment we have gathered multiple quantitative objective measures that have been statistically evaluated. As the best-performing method we have identified the ant-inspired ACO_DTree algorithm that performed significantly better and provided comprehensible results.

Název v anglickém jazyce

Ant-inspired Algorithms in Health Information System Data Mining, Classification and Visualization

Popis výsledku anglicky

In the paper we describe the process of cardiotocography (CTG) record clustering and the effort that was taken to retrieve information from an old (approx. 15 yrs) hospital information system. It consisted of retrieving data from DB tables in the form of unstructured text attributes. Information retrieval had to be performed for complementing cardiotocography signals with additional information. This was needed for further rule discovery mining and automated processing as the main goal was asphyxia prediction during delivery. A graph-based visualization technique has been designed and developed. Together with experts it helped to handle and efficiently process great amount of unstructured (or semi-structured data). Of course, we have used and tested several approaches: automated, semi-automated and manual clustering of the records. In the (semi-)automated experiments we have used k-means, self-organizing map and self-organizing approach inspired by ant-colonies. The overview obtained was consulted with medical experts and served for further mining and information retrieval from the database. Furthermore, an evaluation of CTG signal classification was carried out using multiple methods. These methods used different classification approaches (Naive Bayes, rule and tree based classifiers, etc.). We conducted ten-fold crossvalidation and for each experiment we have gathered multiple quantitative objective measures that have been statistically evaluated. As the best-performing method we have identified the ant-inspired ACO_DTree algorithm that performed significantly better and provided comprehensible results.

Druh

D - Stať ve sborníku

CEP obor

JD - Využití počítačů, robotika a její aplikace

OECD FORD obor

—

Projekt

<a href="/cs/project/NV15-31398A" target="_blank" >NV15-31398A: Charakteristiky elektromechanické dyssynchronie predikující efekt srdeční resynchronizační terapie</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2016

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

XIV MEDITERRANEAN CONFERENCE ON MEDICAL AND BIOLOGICAL ENGINEERING AND COMPUTING 2016

ISBN

978-3-319-32701-3

ISSN

1680-0737

e-ISSN

—

Počet stran výsledku

6

Strana od-do

868-873

Název nakladatele

Springer

Místo vydání

New York

Místo konání akce

Paphos

Datum konání akce

31. 3. 2016

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000376283000170