Insuranceciooutlook

DigitalOwl: Taking the Herculean Tasks Out of Medical Record Reviewing

Amit Man, Co-founder and CTO and Yuval Man, Co-founder and CEO, DigitalOwlAmit Man, Co-founder and CTO and Yuval Man, Co-founder and CEO
It’s a no-brainer that time is money in the insurance industry. The need to comprehensively review every page in medical records is a mainstay of this sector, and for good reason. Especially for claims processing and underwriting, insurance companies have to review medical record data with an extremely high level of scrutiny to verify that every claim is backed up with all supporting information. Even a minute oversight can result in wrong pay-outs, costing millions of dollars in revenue loss down the line.

However, would such levels of scrutiny cost insurance carriers their business? If 1000s of man-hours are spent each day trying to review the records manually, then the ROI on the review may very well be diminished in contrast to the time spent to earn it. The average medical record holds more than 44,000 words, including more than 2900 medical extractions and data points such as procedures, conditions, and medications. Lawyers, underwriters, and other skilled professionals end up wasting their time on reading medical records. Alternatively, outsourcing the process of summarizing medical records is about $1 a page, which is quite substantial if we do the math, especially considering the average medical record is 200-300 pages long.

This was the very problem that bothered Yuval Man, a lawyer by training and currently the co-founder and CEO of DigitalOwl. As a personal injury lawyer who often had to sue insurance companies on behalf of his clients, Yuval used to spend hours manually reviewing medical records, to build his case. So when he communicated to his brother, Amit Man—who presently serves DigitalOwl as its co-founder and CTO—about just how hectic this was, Amit considered the possibility of leveraging AI to make the process faster and hassle-free. And with that mission, DigitalOwl was formed.

The Man brothers wanted to address this need in the market by creating a solution that dramatically improves the process of reviewing medical records for claims processing and underwriting. The company’s clients deal with highly unstructured medical records, consisting of scanned documents of varying quality and layouts. Most off-the-shelf products cannot handle this, generating gibberish instead of gleaning meaningful data points. To this end, DigitalOwl was careful not to base their solution on something so finicky, deciding to opt for a ‘from scratch’ approach instead of patching together existing solutions. The company spent a lot of time and effort building its proprietary engine that relies on natural language processing to analyze all types of scanned documents. The feat took some creative ideas regarding the architecture of DigitalOwl’s AI models and a multi-disciplinary team to tag and feed the AI system. DigitalOwl trains its taggers for 4-5 months to enable them to be as capable as a junior underwriter. With an army of doctors and medical experts, the company also ensures to feed the models with high-quality input data, which, in turn, ensures they deliver a true production solution that customers can depend on.


Ours is a solution that uses natural language processing, to extract information from hundreds to thousands of pages of scanned medical records and present that data chronologically, allowing users to search and filter by condition, date, body part, body system, and provider

“Ours is a solution that uses natural language processing, to extract information from hundreds to thousands of pages of scanned medical records and present that data chronologically, allowing users to search and filter by condition, date, body part, body system, and provider,” states Yuval. As a result, the DigitalOwl system can extract twice as many meaningful medical data points as any human-enabled solution at a fraction of the cost.

DigitalOwl currently works in the US with multiple companies to help them get through the mind-numbing hassle of manual underwriting and claims analysis. The company’s technology solution extracts over 17,000 medical data points from a variety of different conditions, procedures, and medications, including cancer, heart disease, accidents, orthopedic conditions, diabetes, arthritis, hypertension, and brain injuries, with greater than 95 percent accuracy. It is a highly versatile, smart, and dynamic tool that generates a focused data set with multiple filtering options and easy-to-use navigation. Every condition, date, and entry is clickable, redirecting users to the source of information in the record. The resulting dataset is formatted in the form of a meaningful summary that puts the most relevant data at the user’s fingertips, all in a fraction of the time. The complete history of the case/incident is transformed into this meaningful summary and packaged within a “smart” and “dynamic” PDF file, which can be easily reviewed.

Their client success stories demonstrate the value proposition of DigitalOwl. Yuval recalls an instance where one of their clients was handling a case where the insured fell down the stairs, hurting their back. They went on to file a claim for 30 percent disability. With $1.5 million of coverage, the claim would be about $500,000. Long story short, the person who fell down the stairs was diagnosed with a condition called kyphosis. The insurance company took all the medical records and sent it to a 3rd party doctor who went over the medical records to verify that the patient indeed had kephosis and that a 30 percent pay-out would be fair.

This was, however, the first claim that was analyzed by the client using the DigitalOwl engine. The system extracted the relevant data and deduced that the patient already had kyphosis before buying the policy and never initially declared. Within 2 minutes, the claim was declined on the grounds of pre-existing conditions, something that the human reviewer missed.

Having carved a unique niche in the insurance space by scripting several such success stories, DigitalOwl, looking ahead, aims to broaden the scope of their offering. After all, there is still more for their NLP-powered solution to address. Most medical record data, nearly 97 percent is just images in the form of printed pages.“We set out initially to summarize medical records for the underwriting and claims process, and now, our goal is to leverage the enormous amount of extracted medical data points. In the future, we will be able to refine and enhance current underwriting models for risk selection and pricing, and bring even greater value to our customers—both current and upcoming,” concludes Yuval.

Share this Article:
Top 10 InsurTech Startups - 2021
DigitalOwl

Company
DigitalOwl

Headquarters
Tel-Aviv, Israel and Portland, ME

Management
Amit Man, Co-founder and CTO and Yuval Man, Co-founder and CEO

Description
The company offers advanced AI-based technology for faster medical records review, enabling clients to attain increased productivity and a competitive edge. DigitalOwl currently works in the US with multiple companies to help them get through the mind-numbing hassle of manual underwriting and claims analysis. The company’s technology solution extracts over 17,000 medical data points from a variety of different conditions, procedures, and medications, including cancer, heart disease, accidents, orthopedic conditions, diabetes, arthritis, hypertension, and brain injuries, with greater than 95 percent accuracy. It is a highly versatile, smart, and dynamic tool that generates a focused data set with multiple filtering options and easy-to-use navigation