Due to the complexity of medical procedures and the variety of goods and services available today, medical billing is prone to errors and waste. Mistakes in electronic medical records are common, and many patients don’t examine their monthly insurance Explanation of Benefits closely enough to catch them.

Fraud, waste, and abuse (FWA) is the term used for both intentional and unintentional billing errors that end up costing insurers money. Although deliberate acts often target Medicare and Medicaid recipients, they are sometimes aimed at private insurers. Sometimes, the patients themselves are in on the fraud.

Fraud in the healthcare industry is not a new problem and eliminating it may be impossible. The National Healthcare Anti-Fraud Association estimates the cost of healthcare fraud to be $80 billion annually in the U.S. Other sources determine it is closer to $200 billion, or 3-10% of the total spend on healthcare.

However, big data can tackle not only deliberate fraud, but also unintentional input errors that staff make. Machine learning powered by big data can identify abnormal patterns and outliers from individual providers based on historical data that is continuously updated to provide more accurate results.

This can help insurers recover more losses, which research suggests is only 5% annually.


Unintentional Waste

Medical billing codes are now seven digits long, increasing the likelihood of input errors. These errors are not only bad for insurers but can also result in inflated bills for patients. The resulting loss of patient trust and money can negatively impact their willingness or ability to seek future care.

Streamlining databases and designing artificial intelligence systems to detect input errors can help reduce unintentional waste and overcharging. By combining patient data systems and making CMS more efficient, medical providers can reduce the number of times staff must input data, therefore reducing the number of chances for errors to occur.

However, these errors are not nearly as draining as the money lost to deliberate fraud and abuse of insurance plans. Plus, insurance companies cannot force healthcare providers to implement better patient data systems. Insurance companies must make their own improvements to tackle every type of FWA and minimize their risks.


Healthcare Fraud and Abuse

All types of insurance have to be aware of fraud and abuse, and health insurance is no exception. Physicians or other healthcare staff can submit false claims to get more money from insurance companies than would be possible through honest means.

Sometimes the fraud is subtle, such as charging insurers for a more expensive version of a procedure than what was actually performed. Other times, the patient never received a service at all or is given a false diagnosis that a physician then uses to rack up charges.

Another example of fraud involves fake diagnoses in an attempt to get patients approved for government disability payments. Although this is not as common as fraud for the healthcare provider’s own gain, it can happen with severe mental illness diagnoses. One case in 2016 involved a psychiatrist in Miami, Florida, whose false diagnoses resulted in over $50 million in fraudulent insurance claims and undeserved disability benefits being paid out to patients over a decade.

That physician was caught thanks to an FBI informant. Improvements in big data in the past several years have made it easier for companies to detect, investigate, and recover losses from fraud.


Using Patterns to Uncover Waste and Fraud

Big data includes information ranging from patient zip code to physicians’ certifications. Insurers can examine and compare populations and providers to detect fraudulent trends manually, but modern data analysis programs can often identify fraud automatically.

Naturally, the larger a dataset is, the more useful it will be. Accurately detecting fraud and waste requires data that paints a full picture of a situation and includes all variables that may be influencing the data.

For example, simple numerical analysis of a surgeon’s costs may show that he is charging far more than his nearby peers, which could suggest that he is committing fraud. However, a more robust analysis may reveal that he is certified to do resource- and labor-intensive neurosurgeries that no one else in the state can perform.

Training big data to prevent false positives takes thousands of hours of work. Healthcare and billing practices are always changing, and historical data that worked a few months ago could suddenly become less useful if there are major changes in a particular specialty. However, experts in data analytics in healthcare fraud are more than up to the challenges of this ever-evolving field.

How Big Data Works

Crunching numbers and other data from healthcare providers nationwide requires massive amounts of computing power. Machine learning involves complex programs with AI that can use decision trees, outlier detection, data visualization, and other techniques to flag various forms of fraud and waste accurately.

These days, many big data analysis programs rely on data lakes stored in cloud computing for faster processing. Data lakes allow raw data to be stored and easily searched. The power of cloud computing enables insurers to access the computing power they need without having to invest in their own servers.

Because of the amount of computer power and programming complexity required, big data analysis and machine learning for data as complex as healthcare fraud simply weren’t possible in decades past. Computer data analytics programs have evolved significantly, and fraud detection programs have been put to work in many financial sectors.

To maximize profit margins and stay competitive, insurance providers must use every tool at their disposal. Health insurance companies can work with data analytics companies that specialize in healthcare to design cloud computing solutions as part of a broader digital transformation for the 21st century.


Tracing a Web of Bad Actors

Although most healthcare providers are honest and don’t abuse the system, providers who are engaging in deceptive billing practices often work together. Some doctors use fraudulent referrals to send patients to a doctor they don’t need to see, then receive kickbacks.

Other times, managers may train billing specialists to inflate patient bills by inputting codes for more expensive services. If all staff at a provider are submitting bills that are inflated, then it might not be noticeable to an insurer at first.

Big data analytics can uncover these broader schemes and connect the dots between fraudulent providers and staff. Even unintentional poor data management or inputting procedures at a particular facility could be revealed and reported when the right analysis methods are used.

If there is enough evidence to suggest deliberate fraud, criminal charges could be pursued, and insurance companies could recover some of the overpayments.

However, detecting fraud amidst millions of annual transactions can’t be performed without the right technology. Big data can revolutionize the healthcare industry and finally cut costs for insurance companies around the world.

Data Science Techniques for Fraud Detection

Data science-driven anomaly detection finds suspect behavior, even as fraudsters adjust their behavior. With data science techniques, the people working to prevent fraud have shifted the battlefield in the war against fraud. Read More

Multiple Skills for Fraud Prevention

Successful applications of big data to healthcare must use multiple levels of analysis to detect all types of fraud and waste. Since there are various ways providers may make errors, programs must be fine-tuned to respond to all possible scenarios appropriately.

Data must be able to be singled out individually with outlier detection and claim analytics. Following up on individual flagged transactions is relatively simple, especially when data is well-organized.

However, broader and more systematic attempts at fraud require deep AI machine learning.

Fraudulent referrals, for example, can be hard to catch among a sea of similar referrals between general practitioners and specialists. Analysis by zip code, income, patient diagnoses, provider type, and physician credentials requires complex text data to be read and sorted by a program that thinks similarly to a human fraud prevention specialist.

Both individual cases and broader trends require careful data management that keeps all data correct, accessible, and stored in a way that allows it to be analyzed in hundreds of different ways. Being able to accomplish both with the same data sets and systems is a unique challenge, but not an impossible one thanks to innovations in data science.


How to Prevent Fraud and Abuse in Healthcare

Data doesn’t exist in a vacuum, and it takes a large field of experts to put it all in context. Healthcare, in particular, is a complex field that requires specialized knowledge to analyze properly. Running data analysis, tweaking machine learning algorithms, and addressing false positives in fraud detection requires careful consideration of the thousands of variables affecting patients and providers.


At Syntelli Solutions, we’ve established ourselves as leaders in fraud prevention, digital transformation, and other key areas of innovation for the healthcare industry. We can use open-source programming codes and platforms to create data lakes and programs that can be continuously improved and updated as the healthcare industry changes. Contact us to learn more about what we can do for you. 






Achieving Smarter AI with NLP

Achieving Smarter AI with NLP

Experts predict that 30% of companies will base decisions on graph technologies by 2023. This change and a shift to operationalizing AI may cause an increase in streaming data and analytics infrastructures. Several AI techniques, including machine learning (ML),...

read more