A Comparison of Machine Learning Methods Applicable to Healthcare Claims Fraud Detection

Nnaemeka Obodoekwe, Dustin Terence van der Haar

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Citations (Scopus)

Abstract

The healthcare industry has become a very important pillar in the modern society but has witnessed an increase in fraudulent activities. Traditional fraud detection methods have been used to detect potential fraud, but for certain cases they have been insufficient and time consuming. Data mining which has emerged as a very important process in knowledge discovery has been successfully applied in the health insurance claims fraud detection. We implemented a prototype that comprised different methods and a comparison of each of the methods was carried out to determine which method is most suited for the Medicare dataset. We found that while ensemble methods and neural net performed, the logistic regression and the naive bayes model did not perform well as depicted in the result.

Original languageEnglish
Title of host publicationInformation Technology and Systems - Proceedings of ICITS 2019
EditorsManolo Paredes, Carlos Ferrás, Álvaro Rocha
PublisherSpringer Verlag
Pages548-557
Number of pages10
ISBN (Print)9783030118891
DOIs
Publication statusPublished - 2019
EventInternational Conference on Information Technology and Systems, ICITS 2019 - Quito, Ecuador
Duration: 6 Feb 20198 Feb 2019

Publication series

NameAdvances in Intelligent Systems and Computing
Volume918
ISSN (Print)2194-5357

Conference

ConferenceInternational Conference on Information Technology and Systems, ICITS 2019
Country/TerritoryEcuador
CityQuito
Period6/02/198/02/19

Keywords

  • Fraud detection
  • Healthcare
  • Machine learning

ASJC Scopus subject areas

  • Control and Systems Engineering
  • General Computer Science

Fingerprint

Dive into the research topics of 'A Comparison of Machine Learning Methods Applicable to Healthcare Claims Fraud Detection'. Together they form a unique fingerprint.

Cite this