Fraud Detection - A Real-Life Example of AI in Insurance: Workers' Compensation Focus

?

One of my favorite conversation starters when speaking to risk managers and claims professionals is asking them to share their most memorable fraud case.?? The stories are all over the place...? I recall a client, (it was a trucking company) who sent their private investigator into a strip club with a briefcase that hid an actual video camera - think 1988.? Another client (an airline) had an injured worker who was totally disabled from a shoulder industry, yet was working as a barber in 2015.? The investigator wore a camera on his tie and went into the shop to get a haircut.? Although these stories are fun conversation starters, they are also stories revolving around a serious (and costly) topic -- FRAUD.?

?

So, today I'm bringing you another example of how AI can be used in the insurance industry to detect fraud.

?

AI algorithms can analyze patterns and detect anomalies that may indicate fraudulent claims, helping adjusters prioritize and investigate suspicious cases.? For example, let's say you get a claim where the employee reports a severe back injury that allegedly occurred while lifting heavy equipment. The adjuster initiates the usual investigation process, but the AI system is in place to assist in fraud detection.

?

So, the AI-based fraud detection can analyze historical data of similar claims, looking for patterns and anomalies. It considers factors such as the nature of the injury, the circumstances surrounding the incident, and the claimant's history.

?

In this example, AI can also scan social media platforms for information related to the claimant. It looks for any posts, photos, or updates that may contradict the reported back injury. For example, if the claimant posted pictures of themselves engaging in physically demanding activities after the reported injury date, it could raise suspicions (insert eye rolls for all of the times you've been told by a judge "he/she was having a good day").

?

Natural Language Processing (NLP) can be used to analyze the text in the employee's statements and medical reports. The AI system identifies inconsistencies or changes in language that may indicate deception or exaggeration.

?

AI can? compare the employee's medical records with the reported injury. If there are discrepancies between the medical diagnosis and the claimed injury, the system can flag it (set a diary for the adjuster, team lead, etc.) for further investigation.

?

Likewise, AI can examine relationships between the claimant, healthcare providers, and witnesses. Unusual connections or patterns may be indicative of collusion or fraudulent activities.

?

Using historical data, the AI system creates a predictive model to estimate the likelihood of fraud based on various parameters. If the current claim exhibits a significantly higher probability of fraud compared to normal patterns, it raises a red flag.

?

Also, think about if the claim involved photo/video footage.? If the claim includes photographs of the workplace or the alleged injury, AI-based image analysis can be used to detect any signs of manipulation or inconsistency.

?

And of course, real-time monitoring.? AI can continuously monitor incoming claims in real-time. If it identifies patterns similar to known fraud cases or if it detects sudden spikes in claims with similar characteristics, it alerts the adjuster for immediate attention.

?

By combining these AI-driven techniques, workers' compensation adjusters can efficiently identify potential fraud cases, allowing them to prioritize investigations and allocate resources effectively. This not only helps in mitigating financial losses due to fraudulent claims but also ensures that legitimate claims (yes, there are many!) are processed more promptly.