Abstract:
Community-based health insurance schemes (CBHIs) apply the principles of insurance to the
social context of communities, guided by their preferences and based on their structures and
arrangements. CBHIs can help communities manage health care costs and provide access to
basic health care for the poor and other vulnerable groups.
Typically, CBHIs are organized and managed by a local community organization. The CBHI
plan establishes agreements with various health providers, thereby forming a network of
facilities. Most schemes cover basic health care services (e.g., antenatal care, deliveries, and
child health care) and family planning services, while some schemes may also cover costs of
hospital treatment.
Insurance fraud is an act that can be seen in different insurance types including CBHI insurance.
Fraud in the case of CBHI is done by misrepresenting facts to get unauthorized benefit from the
expenses covered under CBHI in the society. Globally companies are spending high amount of
claim costs due to insurance fraud. It is a concern for companies to have a system that could
differentiate frauds from incoming claims. Data mining tools and techniques can be applied in
different fields one of which is fraud detection. This research is conducted for the purpose of
testing the applicability of data mining techniques in detecting fraud suspected CBHI claims in
the case of west gojjam zone. A six step hybrid process model is used to guide the entire
knowledge discovery process. J48 decision tree and Naive Bayes classification algorithms are
used to build predictive model. Several experiments are conducted and the resulting models
show that the J48 decision tree is found to work well in detecting fraud with 93.06%
c l a s s i f i c a t i o n accuracy. A prototype is developed based on the rules extracted from the J48
decision tree model. Finally recommendations and future research directions are forwarded
based on the results achieved.