Korva is a scroll of the internet’s best ideas.
AI sorts claims by medical history
Why claims differ
Take two people who each have a heart attack, even one caused by an outside substance like a drug. One has a lot of comorbidities. The other is extremely healthy, with no other factors, and then all of a sudden has a problem. Their claims potentially aren't valued the same.
Records come first
That is why records come first. When litigation starts, the firm collects each client's medical records and categorizes the claims as early as it can, to get a handle on what it has. AI helps with that analysis, and nothing in the work involves filing AI-written briefs.
Feeding the math
The tiering feeds the math. Before the firm knows what number settles a case, it has to look at thousands of claims and do modeling. The differences among plaintiffs, different injuries over different time periods in different ways, are also why mass torts don't fit neatly into class actions.
Why do pre-existing conditions change a claim's value?
Tort law makes a defendant responsible only for the harm its conduct caused, so the question is what the exposure added to the person's condition. A patient with many existing illnesses has more alternative explanations for a heart attack, which gives the defense room to dispute causation. Damages can also be lower when an existing condition was already shortening a life or limiting earnings. Courts also apply the eggshell plaintiff rule, which holds a defendant liable for the full extent of harm to an unusually vulnerable person.
How can software read thousands of medical records?
Medical records often arrive as scanned pages, faxes, and PDFs. Optical character recognition converts the images to text. Language models can then extract items such as diagnoses, dates, and medications from that text, and the results can be sorted into categories. Human reviewers typically check the output, because extraction errors can change how a claim is valued.