The identification of Emerging Risks using big data analytics can make the World a safer and healthier place to live. It also offers the potential to significantly reduce personal injury liability exposure for insurers. This week, Rick Huckstep talks with Bob Reville, the Los Angeles based CEO of InsurTech firm, Praedicat about giving insurers the ability to look forward and identify the ‘next Asbestos’.
Concussion is a 2015 film starring Will Smith. The movie dramatizes the efforts of Dr Bennet Omalu‘s fight against the NFL’s efforts to suppress his research on the brain damage suffered by professional football players. The film starts with Omalu’s interest in the unusual and unexplained behavior of professional footballer, Mike Webster prior to his suicide.
This resulted in Omalu identifying a condition suffered in high contact sports, known, in layman’s terms, as “punch-drunk”. In 2005, Omalu published a science paper and immediately faced a wall of objection from those with a vested interest to show his findings were wrong.
Omalu’s problem was simple.
He had an insufficient body of science backed evidence to prove his theory.
A decade later, and after a string of suicides by professional footballers who all ensured that their brain remained intact for further research, the NFL reached a $765m settlement. This was later challenged in court, but that’s another story.
The point is that the total payouts in personal injury claims in this case are expected to exceed $1bn. The evidence existed for action to be taken long before anything was actually done. But it took time for a sufficient body of scientific evidence to amass. As a result, players continued to be injured and insurers paid a higher price in the long run.
Although on a different scale, the same could be said with Asbestos. The cost in terms of health, livelihood and personal injury is an order of magnitude greater. Published medical research existed to anticipate the risk as early as the 1920s. This body of evidence slowly built over the next 50 years before the litigation emerged. For the U.S. insurance industry alone, asbestos-related losses could eventually reach as much as $85 billion.
Emerging Risk Groups
The issue for insurers and reinsurers is that whatever happened in the past isn’t going to tell them what’s going to happen in the future. Analyzing historical claims data and trend indicators is simply looking through the rear view mirror. It isn’t going to identify the ‘next Asbestos’.
Which is why the industry is looking to InsurTech for solutions that use new sources of data and a computing capability that didn’t exist until recently.
Within each insurer is an Emerging Risk Group (see example here from Munich Re). There are also industry bodies, such as the CRO Forum’s Emerging Risk Initiative, consisting of ten members from the reinsurance industry. And of course, the world’s specialist insurance
market, Lloyd’s has one (see here for a white paper published by Lloyd’s and Praedicat).
Up until now, these emerging risk groups have not been tech driven. Instead, they have relied on good old fashioned desk research, using detective work to identify the next big exposure.
Big data is the answer
To find out more about this subject, I Skyped recently with Bob Reville, the CEO of Praedicat. Like just about every other InsurTech founder and CEO that I have interviewed, Bob knows this subject incredibly well.
Over a decade ago, Bob was working for the RAND Institute for Civil Justice, a non-profit think-tank in Santa Monica. With a background in workers’ comp, victims’ comp and terrorist risk, Bob found himself working on casualty catastrophe modeling. Then in 2012, Bob and a team backed by RAND and RMS formed Praedicat to develop a platform to exploit machine learning, artificial intelligence (AI) and high performance computing.
Bob explained: “There’s a mass of peer reviewed scientific data out there. But it’s all held in scientific journals and academic research. Which makes it really hard to access, let alone understand and piece it together.
“So, we set about the task of mining the world’s community of scientists. Our goal is to identify emerging risks by identifying early signs from their published research and then using the platform we’ve developed to predict when the risk will occur.”
Bob is referring to CoMeta, their first product that is now joined by Oortfolio, a portfolio modeling platform. Together, these products “drawn from peer-reviewed science, provide an unprecedented platform for identifying, underwriting and managing systemic liability risk”.
Looking for the needle in a haystack
Part of the problem for insurers is that they don’t know what they’re looking for. It’s a needle in a haystack with no idea what the needle looks like. And since they cannot identify the risk (aka, the needle), insurers have become risk averse and focused on looking at where NOT to be exposed.
Bob explained; “Praedicat wants to change the mindset of insurers. There are so many risks where the science isn’t so strong that it will support litigation at the moment. So, they fall under the radar. But, by changing the way insurers manage their aggregations, focusing on science-based named-peril risks instead of industries, they can now price the risk properly when they couldn’t see it before.”
Risk is a two way street
The flip side of analyzing emerging risks is that insurers can also identify where risks are not as great as they are perceived to be.
Take all the controversy over power lines and cell phones. The science has never been strong enough to support litigation and as a result the story has run its course and coverage has shrunk for these risks.
But in the early days of this emerging risk, insurers ran away from it when the media were writing lots of doomsday headlines. With the benefit of hindsight (and now Praedicat), they didn’t need to. In fact, they could have done the complete opposite and covered the risk with a degree of assurance about the exposure they were writing.
Reading the cards
Praedicat’s algorithms are constantly mining scientific research, such as those exploring harms from cell phones, nanomaterials (such as carbon nanotubes), benzene and BPA. The result is dynamic metadata that is used to inform risk modeling.
The mining tech is centred on the search for saliency, which is defined as the state or quality by which harm hypothesis stands out relative to its neighbors in the scientific literature.
Using machine learning, AI and big data processing tech, Praedicat search through over 22 million pieces of peer reviewed scientific research. They then connect the risks that are uncovered by the algorithms to the industries exposed to the risk, including detailed analyses on over 1,000 companies.
Praedicat are constantly trawling for new articles. When they find related material from these disparate sources using algorithms, the hypothesis is deemed “salient” and the research is clustered together, the state of the science is analysed and the industrial footprint is described.
The sheer scale of this activity could only be achieved by developing computing capability and the tools to machine read and interpret the literature. Praedicat is constantly improving its technology to scale this process, including now developing an “automated abstract reader.”.
And whilst this is largely machine driven tech, Praedicat do have their own bio-scientists and economists to review the results and complete the analysis.
Praedicat make the impossible, possible
The tech that Bob and his team have built is pretty awesome.
It is simply inconceivable to image humans assimilating, cross-referencing and piecing together the multitude of disparate wisdom and learning that exists today.
By the application of InsurTech and the power of computing, Praedicat have made the impossible, possible.
Bob has the final word; “While commercial liability is one of the world’s most important insurance lines, over the last 30 years it has been a source of persistent problems for insurers. InsurTech advances the promise to revolutionize underwriting, drive product innovation, and usher in a golden age of Casualty over the next 30 years.”
The author, Rick Huckstep is an InsurTech thought leader and the Editor of InsurTech Weekly for The Digital Insurer.