Insurers Are Trying To Reduce Hospital Discharge, Settlement Times For Life Insurance Using AI — Here's How

Last month saw Sampat Jain's family get a health scare when his sodium levels dropped, requiring him to be hospitalised at Apollo Hospitals in Kankurgachi, Kolkata. While he was ready to be discharged by about 8:30 a.m., he was eventually released nearly 12 hours later post the completion of discharge formalities and paperwork.

Insurance companies are attempting to bring down discharge and settlement time using artificial intelligence. They are using AI to further automate claims assessment, reduce human intervention, analyse medical records, and determine claim eligibility.

Leveraging tech and AI means a claim could be processed in minutes, said Mayank Bathwal, chief executive officer at Aditya Birla Health Insurance. Given that such data is not standardised in India, AI is being used to read documents quickly and convert that to data, he said.

Until now, procedures such as preparing investigation reports, discharge summary, and medical bills have been manual, said Nikhil Jha, founder at Hercules Insurance Advisors. What was once a seven to eight hour process can now take half the time or less, he said. Hospitals now have portals that need you to upload everything versus earlier, when there might have been an insurance company designate. Assuming there are 10 people at a ward ready for discharge, the hospital will use the first-in-first-out method and discharge them in the order of documentation completion.

Not just health insurers, life insurers such as Axis Max, too, are actively relying on AI for speeding life insurance claims settlements in the event of deaths. Manu Lavanya, chief operations officer at Axis Max Life Insurance, explains that the process, in fact, starts as early as when the company reaches out to prospective clients shortlisted by analytics for a life insurance product.

From the moment the onboarding journey starts, AI is used to calculate the risk propensity of the customer, Lavanya said. "We have to identify the risks from a financial and medical perspective." The information provided by the customer, along with their financial and medical history, provides the company with a view of the customer's propensity to pay up, or the likelihood of a customer to continue next year etc., he explained.

"With deep data integration and deriving data from account aggregators, if the customer gives consent to track their data you don’t need to ask for very much beyond that," he said.

The second part of the process of onboarding incorporates validating the authenticity of the customer—for instance, to ensure the same person who came for the insurance went to the medical centre.

Post issuance, the underwriting risk management practice can also flag fraud profiles basis a combination of parameters such as occupation, pin code, etc. Technology such as photography capabilities can determine by skin tone, capillaries or veins if a person is a smoker or a non smoker.

Almost a third of the cases flagged end up eventually getting declined or cancelled, Lavanya said.

When it comes to customer service too, analytics and AI is used to predict customer intent.

Every early claim is investigated, said Lavanya. "For non-early claims, we have built AI analytics to process that claim," he said. Nearly two-thirds of the portfolio is in non-early claims. "Within that, we were able to clear about 70% of the claims on the very same day," Lavanya said, adding the 24-hour claim paid ratio would be almost about 48% or there about. "By the end of this year, we target settling about 60% of all claims in 24 hours," he said. The company is now trying to ensure settling claims within three hours.

The benefits of AI in healthcare extend beyond operational efficiency, extending to assisting in early disease detection and personalised treatment plans, according to a blog post by Niva Bupa. AI tools like telemedicine platforms and virtual consultations are also making healthcare services accessible to a broader audience, it said. Still, issues with data privacy, compliance and integration remain.