November 29, 2022
Resolution Life is using AI software to determine whether insurance claims are "easy" or "complex" within 15 seconds of them being lodged. The insurance company, which manages around $30 billion in assets under management in Australasia, has upgraded the model used to make this determination, from an in-house buiilt model, to one run out of…

Resolution Life is using AI software to determine whether insurance claims are “easy” or “complex” within 15 seconds of them being lodged.

The insurance company, which manages around $30 billion in assets under management in Australasia, has upgraded the model used to make this determination, from an in-house buiilt model, to one run out of the H2O.ai platform.

At an H2O World Sydney event, Resolution Life’s head of data engineering Rajesh Malla said claims processing is a multi-stage process.

“When a claim is lodged, we would like to assess the claim at the right time and provide a customer with the right outcome,” Malla said.

“There are multiple steps before we [even] process a claim. The first step is to identify the segment of that particular claim, so that means is it an easy claim for us to move forward? Do I really need to spend any time on it? Or is it a complex claim [where] we really need to talk to our customers to understand their situation – was there a claim earlier made before, is it an extension to that claim and whatnot.

“So the segmentation of the claim makes a critical part of the claims journey.”

In the past, segmentation was a largely manual exercise. 

“Previously, when a claim was submitted, a claims manager needed to be assigned to the claim, open up a claim, look into it manually, [and] spend about a day or two – or a week in some cases – to identify the segment of that particular claim before we can start processing it,” Malla said.

Resolution Life had already tried to automate the segmentation process, and had created an in-house claims triage model that could split claims into “easy” and “complex” buckets with a 71 percent accuracy.

Six months ago, it decided to redo the model in the H2O.ai platform instead. Already, that revamped model is capable of correctly triaging a claim 77 percent of the time, an improvement on the in-house model.

“We spent quite a bit of time and the accuracy pre-H2O.ai was 71 percent, whereas when we moved across onto H2O.ai, the straight accuracy was 77 percent,” Malla said.

“Now, within 15 seconds of the claim being lodged, we can segment that claim into a bucket.”

He added: “We’re happy with that 77 percent for the time being, [but] we are looking for a 100 percent [accuracy] outcome going forward.”

On the architecture side, a lodged claim triggers an API call to the company’s Snowflake environment, where the H2O.ai-based model is deployed and engaged to provide the “predictive outcome”.

“The outcome is sent back across to CMS, which is our claims management solution, and also the outcome is actually stored in our container storage, like [Azure] Blob storage, where we can use it for our future purposes,” Malla said.

As well as streamlining the claims process and getting an outcome for customers sooner, Malla said the triage model also saved money for the insurer.

More H2O.ai use cases

Malla said Resolution Life is now looking to implement H2O.ai in its call centre operations to assess why customers are calling in.

He said the split of customer support is currently 40 percent self-service and 60 percent call centre.

Malla said that a customer may have multiple reasons to pick up the phone, but its existing call centre software only allowed it to capture a single reason for a call being made.

“A customer may call for various different reasons,” he said.

“There could be a primary reason, there could be a secondary reason or the secondary reason could be more problematic rather than the primary reason for them.

“We are looking at H2O.ai to really understand our queues, the call generation: why is the customer calling us and what is the predominant reason for a customer to call?

“Based on that, we can improve our business.”

The company is also looking to machine learning to better analyse call transcripts for insights leading to greater self-service capabilities and to prioritise calls to reduce call volumes and wait times.

Another H2O.ai-related use case being explored is with the vendor’s Hydrogen Torch tool, which aims to make deep learning more accessible.

“The plan is to build a product library that contains information about insurance products,” he said.

Other use cases the company is looking into involve extending usage of H2O.ai to claims processing itself, not just triage. Use cases are also possible in the marketing, actuarial and finance teams. 

Malla added that H2O.ai is not the only tool in use in the organisation; Azure Machine Learning is also in use.

“We have a data sciences practice and use multiple products within the organisation. It’s about the right tool for the right use case,” he said.

Source