Processing bodily injury claims isn’t just a matter of crunching numbers and assessing a transcription of events. Because one particular claim can be open for 10 years and have hundreds of data points and medical notes, adjusters have to be de facto medical experts to make informed decisions.
“It’s not like transactional data on a spreadsheet,” said Tomas Vykruta, CEO of claims guidance platform company EvolutionIQ. “Bodily injury claims can be like a medical book to read through.”
Insurance is a massive industry — it’s worth an estimated $1.3 trillion in the U.S. alone. And insurance companies pay out roughly $230 billion annually in claims payments just across disability, worker’s compensation and complex commercial casualty lines.
But to the benefit of insurance companies, adjusters and the general public alike, that amount can – and should – be dramatically reduced, according to Vykruta. To this end, EvolutionIQ has developed what it calls the first human-in-the-loop artificial intelligence (AI) claims guidance technology for the insurance industry.
EvolutionIQ leads with ML
The three-year-old company, which today announced a $21 million series A funding round, created its system based on deep learning, an advanced branch of machine learning (ML). This leverages AI recurrent neural networks (RNN), which are based on time-series data or data involving sequencing.
This allows the system to actively monitor every open short-term and long-term group and individual disability, worker’s compensation and property and casualty claim under an examiner’s purview to guide them to those that require more attention, new actions or complex decision-making. It will generate a list of the handful that are most actionable, along with a “deep explanation” as to why and the outcome they should be aiming for.
Concerning bodily injury, the system can read through an entire sequence of events that describe the claim and rely on RNN data to simulate sequences of injuries, comorbidities (the presence of multiple diseases or medical conditions), demographics, conditions and other factors. These can then provide projections on when a claimant might recover, to what extent they might recover and what work conditions and responsibilities they can return to.
For example, with a short-term disability claim, a worker may have been out on leave for a week and will be in a queue with several similar workers. The system will pick up on that and determine that they could return to work within, say, 45 days, so long as they receive certain vocational training.
“It will put the information right in front of the examiner,” Vykruta said. “It’s a glass ball where they can see, ‘This is where you should spend your time.’”
As he noted, adjusters can be overwhelmed by dozens of complex claims that can last for years and often be worth hundreds of thousands of dollars each. These can be documented in hundreds of disparate pages and in many structured and unstructured formats.
Furthermore, “these are impossibly complex problems because there’s bodily injury,” he said. “You have to be a doctor in many cases to understand cases of comorbidities. There are way too many complex problems and far too few people to be able to sift through them.”
That said, the deep learning, human-in-the-loop AI system must have people plugged in, he said. Examiners are not eliminated; rather, they contribute to the system as it constantly learns, evolves and recalibrates based on new data and events. “Dealing with bodily injury is a truly complex task and a huge data problem,” Vykruta said. “The system has to partner with human experts.”
Modernizing insurance claims management
Working with customers including Reliance Standard, Principal and Sun Life, EvolutionIQ has processed millions of claims. Carriers and third-party administrators using its software for more than a year have seen claim flow-through reductions of up to 45%, according to Vykruta and the incidence rate of workers moving from short-term to long-term disability has been reduced by roughly 50%.
“Claims management is ripe for modernization,” said Vykruta, a former AI technical leader at Google. “It’s the biggest operational issue plaguing carriers because it’s a huge human effort that can be greatly improved using data. Tens of thousands of claims are open at any given time and there is a significant opportunity to impact them now with the right information.”
Vykruta explained that funding will be invested in R&D and the development of new AI modules. It will also help to build out the company’s team of engineers, data scientists and product and customer service experts. The company currently has 45 employees – many of them coming from Google, Facebook, Amazon and Bloomberg – and plans to grow that base to 85 by the end of the year. Vykruta pointed out that more than 25% of EvolutionIQ’s technical employees come from Google, a rarity for companies in the insurance industry.
As he noted, insurance is a huge industry and a necessary one for the modern world, yet he also underscored the fact that EvolutionIQ’s long-term goal is to reduce industry operational costs and premiums – which benefits everyone, from companies, to adjusters, to claimants and policyholders alike.
“We are very focused on making that claims process more efficient and affordable for everybody,” he said. “We consider this the inevitable future. In the next five years, every carrier will have to have a system like this, or they won’t be able to keep up.”
EvolutionIQ’s series A round was led by Brewer Lane Ventures. Seed investors FirstRound Capital, FirstMark Capital and Foundation Capital also participated, along with Altai Ventures, Asymmetric Ventures, Reliance Standard Life, New York Life Ventures, Guardian Life and Sedgwick.
VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.
Source: Read Full Article