The problem Healthcare Claims Fraud Detection Application solves

Healthcare Claims is an AI-based Android Application tool that enables people to flag the claims as fraud or not.

What is Healthcare Fraud?

A healthcare or a medical insurance fraud is more commonly defined as knowingly executing treatments to render medically unnecessary or over-utilizing services that result in useless costs to the healthcare system, including healthcare insurance providers. Potential offenders may include patients, hospitals, doctors, vendors, suppliers or even pharmacists.

Impact of Healthcare frauds:

Here are some facts related to healthcare claims:

• Each year out of the total claims, medical and healthcare industry claims alone account for more than 15 percent of the totally false claims.
• Reports suggest that the healthcare industry in India is losing approximately Rs.600 - Rs.800 crores incurred on fraudulent claims annually.

Issues with current methods and need for AI:

Up till now, healthcare fraud claims have involved manual work to investigate and identify frauds which have been time consuming and inefficient. The more effective way is to identify frauds in real-time before the claims are paid. Hence there is a need to embrace predictive analysis often used in other industries. This is necessary to prevent scamming, identify inconsistencies, and flag them appropriately.

Proposed Solution:

The solution to the problem stated before is to create an AI-driven FRAUD DETECTOR Application that would provide protection to the payer by:

• Identifying inconsistencies and potential rule-breaking and hence prevent med-care scamming.
• Providing real-time safety feature to flag out fraud transactions and block them.
• Application can be customized according to the needs and data provided by the organization.

Challenges I ran into

We faced certain issues while developing our project. The key issue, however, was the difficulty encountered in retrieving data for unique users from Firebase. In order to overcome this hurdle, we had to create a single Firebase instance and then call that instance repeatedly at several places.

Also, the dataset we worked on incorporated attributes in medical terminology which we weren't attuned to. So, we had to do a lot of digging to get accustomed to the features and then identify the necessary ones alongside the way they could be possibly used.