Applying NLP methods to code functional performance in electronic health records using the international classification of functioning, disability, and health.
Elizabeth Marfeo, Maryanne Sacco, Jona Camacho Maldonado and 4 others
PMID 40442018WHAT IT FOUND
The ICF framework helped standardize how therapists document mobility, self-care, relationships, and communication in electronic records.
But codes often overlapped, lacked cultural inclusivity, and failed to capture modern technology use, requiring manual consensus to make them usable for automated data extraction.
Key findings
01ICF codes for mobility and self-care were useful for annotation but suffered from overlapping definitions and lack of specificity regarding equipment and housing situations.
02Codes for interpersonal interactions and communication failed to reflect cultural diversity, dynamic relationship changes, and modern technology use like smartphones.
03Successful automated extraction required iterative team consensus to resolve ambiguities and expand code definitions beyond their original scope.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTsWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study was conducted in the US healthcare context, which is multi-payer and insurance-driven; documentation practices and data availability may differ significantly in single-payer or universal healthcare systems. The primary data is not publicly available, limiting the ability for other researchers to replicate the specific NLP models. The analysis focused on activity-level functioning, excluding participation domains like major life areas and community life, which limits the scope of the functional profile generated.
Declared interests
The study involved collaboration with the US Social Security Administration. The text does not report other specific financial conflicts of interest or author disclosures.
The easy way to misread this
Do not assume that ICF codes are ready for automated, error-free extraction of functional data. The study shows that codes frequently overlap, lack specificity, and fail to capture modern or cultural contexts, requiring significant manual interpretation and consensus to be useful.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →