CHAPTER 1

INTRODUCTION

1.1 Introduction
Health insurance market is a crucial market because one-third of GDP [1] is spent on
health insurance in the United States, and everyone needs some level of health care.
Health insurance is one of the most significant investment an individual makes every
year. This study is an effort to find mathematical models to predict future premiums and
verify results using regression models. Medical costs that occur due to illness, accidents
or any other health reasons are considerably expensive, by having health insurance, an
individual is not liable for paying the entire medical costs of the procedure. According to
the Office of Health and Human Services (HHS) [2], the total health service budget for
the fiscal year 2015 is around 1100 billion dollars. There are several health care systems
around the world. For example, single payer system followed by Canada where premiums
are paid by taxes, government health care system followed by the United Kingdom where
healthcare is the responsibility of central government. In country like the United States
the insurance is funded through both government and private organizations. For example,
65% of insurance is through private insurances, 25 % through Medicare, Medicaid and
veterinary programs which are funded by government and the remaining population is
uninsured. Kaiser Family Foundation [3] analyzed health insurance premiums over the
years from 2007-2016 and found out that health care premiums have been going up by
17% each year. It is also predicted that premiums are further going to increase. The
question of finding the trends in the health care premiums is profoundly important as it
tends to be the core of all presidential campaigns and other social movements. The
healthcare premiums keep changing every year because of various factors [3] such as
medical trends, pharmaceutical trends, and political factors etc. over which the customer
has no control. The only option an individual can have been to plan carefully for future
expenses. There are no existing tools to the best of our knowledge that can predict future
premiums based on historic data. Therefore, there is a need to conduct research to find the

1
premiums across the United States. This research is focused on predicting the health
insurance premiums for 2017 based on the individual health insurance market place data
made public since 2014.

1.2 Survival of Health Care Insurance Market

This Although there is great deal of uncertainty around healthcare system in

recent years, health care system is not completely doomed because of the following

reasons:

1) Premiums fluctuate according to the market place demand. For example,

Aetna, United Health marketers lowered their premiums for the year 2015 as compared to

2014 and increased in 2016. Other Companies have affordable prices which target certain

section of the society.

2) The government provides protection from an increase of hikes. Consumers

receive a tax credit for their insurance premiums.

1.3 Advancements in Big data and Machine Learning algorithms

Big data [4] is being generated all around us by various digital, electronic devices such as
smart phones, IOT devices, and social media such as Facebook, Twitter, Instagram etc.
Every digital process and social media exchange produces it. Systems, sensors and
mobile devices transmit it. Big data is arriving from multiple sources at an alarming
velocity, volume, and variety. To extract meaningful value from big data, optimal
processing power, analytics capabilities, and skills are needed. Health insurance
marketplace data provides premiums of health plans of several insurance companies
participating in the marketplace. These rates are based on various complex factors such
as age, geographic location, the level of health coverage etc. and vary across states and
markets. Individual plans data (1 GB) considered for this research may not be
considered as big data by industrial standards but still, the conventional methods will
not be able to detect patterns in data. The new data arrives every year. The
2
methodology developed should be able to predict the trends in health care premiums
based on the data collected in future.

1.4 Importance of Research on Health Insurance Data:

Data from market place shows that 30 million plans are available each year for purchases
in different states across the United States. And these plans are selected according to
users anticipated service usage. From Figures 1.1, 1.2, 1.3 & 1.4, it can be seen prices of
premiums are affected by various factors such as age, plan type, geographic location etc.
Therefore, there is a need to develop a mathematical model to predict premiums based on
various parameters which impact the premiums. The premium rates are set by insurance
providers using complex algorithms based on previous years health care utilization and
total number of enrolments.

1.5 Health Insurance Market

In this Section we are exploring the overview of the health insurance market in the
Different Countries through the word was discussed. According to Kaiser Family
Foundation [8], health insurance coverage breakdown is as follows: 49% employee
provided insurance, 20% Medicaid (which includes low-income households and
Children’s Health Insurance Program (CHIP)), 14 % Medicare (over age of 65), 2% other
public (for veterans), 7% non-group such as individuals and finally 9% uninsured. Before
discussing the features of the actual dataset; in the next section, basic terminologies of
health insurance are defined. These definitions are taken from healthcare.gov official
website [9].

3
1.6 Health Insurance Service Area
Qualified Health Plan (QHP)

Premium

Deductible

Co-Insurance

Copayment

Service Area of Health Out- of-Pocket maximum

Health Maintance Organization

Preferred Provider Organization

Exclusive Provider Organization

Point of Services

Metal Levels

Figure 1.1: Service Area of Health

4
Qualified Health Plan (QHP): QHP is defined as a government certified insurance plan
that provides essential benefits such as emergency services, maternity care etc. and
establishes limits on cost sharing (such as deductibles, co-payments, and out-of-pocket
maximum amounts.

Premium: The amount paid for the health insurance every month. In addition to the
premium, other costs are also paid for health care, including a deductible, co-payments,
and coinsurance. If users have a marketplace health plan, costs can be lowered with a
premium tax credit. While choosing a plan the lowest monthly premium may not be the
best match for a person. If a person requires a lot of medical assistance, a plan with a
slightly higher premium but a lower deductible might save a lot of money. After enrolling
in a plan, the first premium directly paid to the insurance company, not to the Health
Insurance Marketplace.

Deductible: The amount paid for covered health care services before the insurance plan
starts to pay. With a $2,000 deductible, for example, the first $2,000 of covered services
should be paid by the customer. After reaching the deductible amount specified in the
plan, usually, only co-payments or coinsurance was paid by the customer for covered
services. The rest will be paid by insurance company. Many plans pay for certain
services, like a check-up or disease management programs, before meeting the
deductible. All marketplace health plans pay the full cost of certain preventive benefits
even before meeting the deductible. Some plans have separate deductibles for certain
services, like prescription drugs. Family plans often have both an individual deductible,
which applies to each person, and a family deductible, which applies to all family
members. Generally, plans with lower monthly premiums have higher deductibles. Plans
with higher monthly premiums usually have lower deductibles.

Co-insurance: The percentage of costs of a covered health care service must be paid
(20%, for example) after paying the deductible.

5
Co-payment’s: Fixed amount ($20, for example) paid for a covered health care service
after paying the deductible. Co-payments (sometimes called co-pays") can vary for
different services within the same plan, such as drugs, lab tests, and visits to specialists.
Generally, plans with lower monthly premiums have higher co-payments. Plans with
higher monthly premiums usually have lower co-payments.

Out-of-pocket maximum: The maximum amount paid by the person for covered
services in a year. After reaching this amount, the insurance company pays 100% for
covered services. Plan type provides a shorthand way to determine what sort of access
members have with health providers outside a plan’s network, including cost-sharing for
treatments. But since there are no industry-wide definitions of plan types and state
standards vary, individual insurers often have the flexibility to market similar plans under
different names.

Health Maintenance organizations (HMOs): This cover only care provided by doctors
and hospitals inside the HMO’s network. HMOs often require members to get a referral
from their primary care physician in order to see a specialist.

Preferred Provider organizations (PPOs): These cover care provided both inside and
outside the plan’s provider network. Members typically pay a higher percentage of the
cost for out-of-network care.

Exclusive Provider Organizations (EPOs): These are a lot like HMOs: They generally
don’t cover care outside the plan’s provider network. Members, however, may not need a
referral to see a specialist.

The point of Service (POS): Plans vary, but they’re often a sort of hybrid HMO/PPO.
Members may need a referral to see a specialist, but they may also have coverage for out-
of-network care, though with higher cost sharing.

Medical Underwriting: A process used by insurance companies to try to figure out the
health status when applying for health insurance coverage to determine whether to offer
the coverage, premium price, and limitations of coverage.

6
Metal Levels: Plans in the Health Insurance Marketplace are presented in 5 “metal”
categories: Bronze, Silver, Gold, Platinum, and Catastrophic. Metal categories are based
on the cost split between the health plan and the person. For example, for bronze metal
category health plan pays 60% and the person pays 40%. For platinum metal category
health plan pays 90% and the individual pays 10%. Metal categories do not affect the
quality of care.

1.7 Motivation
The Motivation of our Dissertation is that In Today’s hectic life nobody has time to detect
the behaviour of their life style. Once it gets infected, we all come under threaten. So, we
come to conclusion if we detected our working behaviour through any mechanism that
will give us prior information about Diseases.

1.8 Dissertation Road Map

Chapter 01: In this chapter we explained the basic concept of Medical Expenses.
Chapter 02: Here we explained the Uses of Data Mining& Machine Learning.

Chapter 03: Here we explained the Literature Survey

Chapter 04: Here we explained Problem Definition & Proposed Method

Chapter 05: Here we explained the Experimental Setup

Chapter 06: Here we explained Result Analysis

Chapter 07: Here we explained Conclusion & Future Scope