The exact origin of the phrase "seeing the forest through the trees" is not clear, but it is a metaphor that has been used for centuries to describe the ability to understand the bigger picture without getting lost in the details. In the context of life underwriting, this analogy refers to the process of analyzing the medical data from complex medical records to understand an individual's overall health and risk profile, rather than getting bogged down by the specific details of each medical issue or treatment. Essentially, it's about being able to step back and look at the overall picture to help the underwriter make the appropriate decision.
This metaphor really hit home as I sat down to think how far we have come over the last year with DigitalOwl’s release of V4.0 of the Digital Medical Abstract.
When it comes to underwriting and claims analysis for insurance, “seeing the forest through the trees” is a critical skill.
A typical 200-page medical record has 44,000 words and 3100 unique data points. Not all data points have the same importance for decision making. The truth is that many data points are irrelevant for decision making and they are the noise that still your attention from the critical data. In the end we need to boil it down only to the “Key” extractions to make the decision for a specific line of business. Combining these data points into the right concepts all while putting it in a timeline becomes quite challenging.
And furthermore, as EMR fuels ever more lengthy and complex medical records with duplication and more noise within the records, seeing the forest for the trees is becoming ever more challenging.
Many companies turn to traditional manual summaries of medical records because it allows the user to quickly and efficiently review the most relevant information about the applicant's or claimant's health and risk profile. A summary can highlight key issues or conditions that may be of particular concern , and it can provide an overview of the individual's medical history and current health status. Manual summaries can save users time but typically with a cost associated for the summary and a delay in processing which hits operational efficiency. It is also very difficult to scale human capital up and down during peaks and valleys leaving uneven capacity models and increasing expense as wages continue to increase. And probably the biggest challenge: the accuracy and level of expert of the manual summary team is not steady and your team often doesn’t trust that the summary is surfacing the most important medical conditions–or that it may overlook significant nuances in the history.
What if there were a better way to “fell the trees”? What if you could have your best experts summarize your cases every single time, 24 hours a day with no sick days? What if you could do this at half of the cost of traditional manual summaries? In addition to the summary, what if you could have access to the relevant data in a structured format to use for automation and better mortality models.
DigitalOwl has solved for this and a lot more with our proprietary Natural language processing (NLP) and artificial intelligence (AI) platform which can be used to help with this process by automating the summarization of medical records. Based on feedback from our customers, our Digital Medical Abstract is now organized so users can perform either a full review of the medical record or quickly find significant information for a specific medical condition or test result. We’ve enhanced our impairment run down to ensure we’re providing ALL the relevant details so the user can be sure they have the information they need to make an accurate decision. To understand the nuance and spirit of the case, a generative model can be used to automatically generate a shorter version of a longer text document, while retaining the most important information and ideas from the original text. We are now using these models take long, complex medical encounters and turn them into a succinct, easy-to-understand summary typical to how a human likes to consume data.
We’ve made great strides in making the abstract more succinct and focused on the most important information. In short, “we’ve come a long way baby.”