Examining Predictive Modeling–Based Approaches to Characterizing Health Care Fraud

This document presents a literature review on the application of predictive-model based approaches for characterization of healthcare frauds. It discusses the effectiveness of predictive modeling in identifying fraud patterns and taking efficient measures.

Header: FOR CONFERENCE-RELATED PAPERS, REPLACE THIS LINE WITH YOUR SESSION NUMBER, E.G., AB-02 (DOUBLE-CLICK HERE)
Examining Predictive Modeling–Based Approaches
to Characterizing Health Care Fraud
Authors Name/s per 1st Affiliation (Author)
line 1 (of Affiliation): dept. name of organization
line 2: name of organization, acronyms acceptable
line 3: City, Country
line 4: e-mail address if desired
Authors Name/s per 2nd Affiliation (Author)
line 1 (of Affiliation): dept. name of organization
line 2: name of organization, acronyms acceptable
line 3: City, Country
line 4: e-mail address if desired
Abstract— This document presents a literature review on the
application of predictive-model based approaches for
characterization of healthcare frauds. This is a major concern in
the modern healthcare industry, where a group people are using
genuine data of the patients in fraudulent activities to increase
their revenue and profit. Large number of false claims and false
billings as well as various other practices are used in the
fraudulent activities in the health sector. Predictive analysis
under big data science is found to be effective in other sectors like
the insurance sector for detecting frauds. Thus, a literature
review has been conducted to show that implementation of
predictive modelling can be useful to identify the fraud patterns
and take efficient measures.
Keywords-Predictive modeling, big data analytics, healthcare
fraud, insurance fraud
I. INTRODUCTION
ealthcare fraud is a major concern factor in today’s
world. It is happening in almost all over the world and
the healthcare fraud may result in spending of hundreds of
billions of dollar, which could have been better utilized on the
patient care [1]. It is often termed as a type of white-collar
crime involving dishonest health care claims for earning profit
[3]. Health care fraud has negative impact on the society as
well as on the economy, and hence, governments of all
countries across the world are taking rigorous measures to
prevent these fraud activities and mitigate the impact on the
people. In order to implement measures to detect healthcare
fraud and take measures, it is essential to analyze the
characteristics of the crime in a precise manner. This is
assumed to be beneficial to categorize the degree of fraudulent
activities in the health care and plan preventive measures and
penalizations accordingly. Data analytics is an effective way
to capture the incidences of health fraud and analyze those
information in a manner that helps to identify the pattern of
the crime and understand the characteristics [2]. Various
techniques under data analytics can be used for assessing the
characteristics of the health fraud. However, this paper will
discuss the effectiveness of Predictive Modeling–Based
Approaches for analyzing the features of the health care fraud
and taking efficient measures through a literature review. A
H
comprehensive discussion on the given topic is presented
below.
.
II. DISCUSSION
A. Healthcare fraud
Healthcare fraud is a social crime. Majority of this fraud is
usually committed by few dishonest healthcare providers.
Unfortunately, due to the fraudulent activities by a small
number of dishonest people affect the reputation of most
trusted as well as respected healthcare providers in the
community [4]. These people misuse their power as well as
people’s trust on them for committing the fraud on a large
scale. While conceiving the fraud schemes, these people often
bring on creative plans as they have access to a wide range of
variables, such as, the entire population of the patients;
complete range of the potential medical conditions as well as
treatment on the basis of which false claims are made; and the
scope of spreading the false billings among many insurers at
the same time. This includes the government’s medical
insurance or health care programs also [5]. This lowers their
chances of being identified by any single insurer. The most
common types of health care frauds include:
a) Billing for the services that were never provided and they
use genuine patient information, obtained through
identity theft, for fabricating the claims with the services
that did not take place;
b) False billing for more expensive procedures or services
than actually performed, known as ‘upcoding’ using
more critical disease and process code; performing
unnecessary medical services solely for increasing
revenue;
c) False diagnosis for justifying unnecessary tests or
surgeries, unbundling or billing each step of medical
process as those are separate process;
d) Billing a patient more than their co-pay amount which are
already prepaid by the plan and many more [5].
Thus, it can be understood that there are various ways that
healthcare fraud can take place and it is important to identify
0018-9464 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. (Inserted by IEEE.)

Header: FOR CONFERENCE-RELATED PAPERS, REPLACE THIS LINE WITH YOUR SESSION NUMBER, E.G., AB-02 (DOUBLE-CLICK HERE)
and categorize the types of healthcare fraud to implement
efficient measures by using appropriate data prediction or
analysis methods..
B. Predictive Modeling–Based Approaches
Predictive Modeling–Based Approaches are a part of the
predictive analytics under data science. Predictive modelling is
defined as the process, which uses techniques like data mining
and probability for forecasting outcomes [8]. Each of the
predictive model consists of a number of predictors that can
potentially influence the future results and after the relevant
data is collected for the appropriate predictors, statistical model
is formulated for examining the data. Predictive modelling
based approaches can be applied in many cases, such as,
weather forecasting and meteorology, online advertising and
marketing and customer relationship management (CRM). For
example, the data modelers or scientists use the historical and
transactional data and apply predictive modeling algorithms for
identifying which types of products the customers might be
interested in and on what they are likely to click. The
predictive models are useful for exploiting patterns in the
historical and transactional information for identifying the risks
and opportunities [9]. It is also useful in the fields of capacity
planning, disaster recovery (DR), change management,
engineering, digital and physical security management.
Under predictive analytics and modeling, various
approaches are used, such as, decision tree, random forest,
support vector machines, neural network and XGBoost for
detecting fraudulent activities in insurance and other sectors
[7]. In this process, the most complex stage is the neural
network, which is a machine-learning model, independently
reviewing the large volumes of the labeled data to find out
correlations between the variables. This is effective in detecting
very subtle correlations emerging after the review of millions
of data points
Figure 1: Deep neural network in predictive modelling
(Source: Nagabandi et al. 2018)
In the predictive modelling, it has been found that brilliant
sample complexity can be achieved by combining the neural
network dynamics model with the model predictive control
(MPC) in a model based reinforcement learning algorithm,
which produces stable and plausible gaits, accomplishing
several complex locomotion activities [10]. It was also
proposed that the deep neural network dynamics model could
be used efficiently to initialize a model-free learner for
combining the sample efficiency of the predictive model-based
approaches with the high performance of the model-free
processes. Thus, the neural network dynamics is a more
efficient function under predictive modeling to find out the
subtle correlations among the variables and hence, since the
healthcare fraud has various aspects and characteristics, it can
be assumed to be effective to identify the influential factors.
In this respect, another study by Ji et al. (2016) can be
mentioned. Multiconstrained model predictive control
(MMPC) has been applied to design a framework for path
planning and tracking, which could be used for maintaining a
collision free path for the autonomous vehicles. It has been
found that the MMPC has been effective identifying various
risk situations for different driving scenarios and it also
provided a path-tracking controller for dynamic tracking
performance and improved maneuverability [14].
C. Predictive Modeling–Based Approaches to
Characterizing HealthCare Fraud
Predictive modeling-based approaches are found to be
effective in the healthcare industry for determining the patients
who have risks of developing certain diseases, such as, asthma,
diabetes, heart diseases and various other lifetime diseases [9].
However, the systematic application of the predictive modeling
is relatively new in the healthcare industry, yet, in a short span
of time, the predictive modeling has proved to be quite
effective. Similarly, this technique of data analysis and
prediction has been proved to be efficient for fraud detection in
various industries, especially in the financial sector, such as,
credit card activities, tax returns, insurance claims, invoices,
online transaction activities and telecom call activities. These
industries are applying predictive modelling to detect fraud,
prevent and monitor those at a reasonable cost. The financial
statement fraud can also be detected through the predictive
modeling [12].
Effective fraud detection in the healthcare sector requires
highly advanced equipment for running the data science
technology. These technologies must be able to handle large
amount of different types of data and identify the relationships
among the influential factors through analysis. The more
number of patterns in the data is identified, the more accurate
would be the prediction of fraud, which would be beneficial for
implementing preventive measures [13]. As the healthcare
industry handles overwhelming amount of data, the chances of
healthcare fraud is very high and hence, big data analytics is
required, which includes predictive modeling for fraud
detection from false claims [11]. The major steps of predictive
modeling are shown in the diagram below.

Header: FOR CONFERENCE-RELATED PAPERS, REPLACE THIS LINE WITH YOUR SESSION NUMBER, E.G., AB-02 (DOUBLE-CLICK HERE)
Figure 2: Major steps of predictive modeling
(Source: Raghupathi and Raghupathi 2014)
Predictive modeling was also found to be effective for
disease management programs and in the health insurance
business. It has been found to be beneficial for almost all
industries for risk evaluation or medical management through
the following steps, that is, risk stratification, simulation and
mapping [12]. These are the basic steps for a predictive
analysis and can be applied on the healthcare sector also for
identifying the fraud.
Figure 3: Foundation concept of predictive modeling
(Source: Bayerstadler et al. 2014)
III. CONCLUSION
From the above discussion, it can be said that, big data
analytics, especially, the predictive modelling approach is quite
useful for identifying the fraud in healthcare, as well as,
establishing the relationships between the variables. As it
consists of variety of statistical techniques, such as, modelling,
data mining, machine learning, game theory etc. for analyzing
a large amount of current as well as historical data, it would be
suitable to handle the fraud detection issue. However, it was
also observed that these models exploit the patterns found in
the transactional and historical data, allowing for the detection
of the risks, such as, fraudulent activities as well as
opportunities. Hence, it can be assumed that predictive
analytics can be useful for characterization of the healthcare
fraud and prevent the impact through relevant measures.
REFERENCES
[1] van Capelleveen, Guido, Mannes Poel, Roland M. Mueller, Dallas
Thornton, and Jos van Hillegersberg. "Outlier detection in healthcare
fraud: A case study in the Medicaid dental domain." International
journal of accounting information systems 21 (2016): 18-31.
[2] Gandomi, Amir, and Murtaza Haider. "Beyond the hype: Big data
concepts, methods, and analytics." International journal of information
management 35, no. 2 (2015): 137-144.
[3] Flasher, Renee, and Melvin A. Lamboy-Ruiz. "Impact of Enforcement
on Healthcare Billing Fraud: Evidence from the USA." Journal of
Business Ethics (2019): 1-13.
[4] Brooks, Graham, Martin Tunley, Mark Button, and Jim Gee. "Fraud,
error and corruption in healthcare: a contribution from criminology."
(2017): 25-44.
[5] Thornton, Dallas, Michel Brinkhuis, Chintan Amrit, and Robin Aly.
"Categorizing and describing the types of fraud in healthcare." Procedia
Computer Science 64 (2015): 713-720.
[6] Sullivan, Carol. "Eradicating Fraud in Healthcare: Possibly a Matter of
Life or Death." Journal of Modern Accounting and Auditing 13, no. 8
(2017): 345-349.
[7] Priya, K. Ulaga, and S. Pushpa. "A Survey on Fraud Analytics Using
Predictive Model in Insurance Claims." International Journal of Pure
and Applied Mathematics 114, no. 7 (2017): 755-767.
[8] Parish, Eric J., and Karthik Duraisamy. "A paradigm for data-driven
predictive modeling using field inversion and machine
learning." Journal of Computational Physics 305 (2016): 758-774.
[9] Mishra, Nishchol, and Sanjay Silakari. "Predictive analytics: A survey,
trends, applications, oppurtunities & challenges." International Journal
of Computer Science and Information Technologies 3, no. 3 (2012):
4434-4438.
[10] Nagabandi, Anusha, Gregory Kahn, Ronald S. Fearing, and Sergey
Levine. "Neural network dynamics for model-based deep reinforcement
learning with model-free fine-tuning." In 2018 IEEE International
Conference on Robotics and Automation (ICRA), pp. 7559-7566. IEEE,
2018.
[11] Raghupathi, Wullianallur, and Viju Raghupathi. "Big data analytics in
healthcare: promise and potential." Health information science and
systems 2, no. 1 (2014): 3.
[12] Bayerstadler, Andreas, Franz Benstetter, Christian Heumann, and Fabian
Winter. "A predictive modeling approach to increasing the economic
effectiveness of disease management programs." Health care
management science 17, no. 3 (2014): 284-301.
[13] Li, Jing, Kuei-Ying Huang, Jionghua Jin, and Jianjun Shi. "A survey on
statistical methods for health care fraud detection." Health care
management science 11, no. 3 (2008): 275-287.
[14] Ji, Jie, Amir Khajepour, Wael William Melek, and Yanjun Huang. "Path
planning and tracking for vehicle collision avoidance based on model
predictive control with multiconstraints." IEEE Transactions on
Vehicular Technology 66, no. 2 (2016): 952-964.