AI Improves Matching in Disability Medical Records (Full Transcript)
[00:00:00] Speaker 1: What you're trying to find is one regulation to establish disability requires them to meet a musculoskeletal listing for the back, but the word is limited range of motion. With regular NLP, it really had to be a one-for-one to match that from the medical record. AI changed that because it can distinguish between like words between the same thing and still get you a conclusion on whether these particular conditions meet these requirements.
[00:00:31] Speaker 2: For example, we're saying instead of it just looking for literally the word limited, it's maybe reduced. It puts that into the bucket of what we're going to flag.
- Traditional NLP often depends on exact one-to-one wording matches in medical records.
- Disability evaluations for back conditions may require evidence such as limited range of motion.
- AI can identify equivalent clinical terms rather than relying only on exact phrases.
- Terms such as “reduced” can be treated as relevant evidence for “limited” range of motion.