Using Natural Language Processing to identify text for qualitative research into the lived experience of people with dementia
Bibliographic record
Abstract
Abstract Background Medical records present a rich potential source of information on the lived experiences of people with dementia. These records are extensive and the work of extracting relevant data is labor‐intensive. We sought to determine whether we could use natural language processing (NLP) approaches to sift through medical records to prioritize an enriched subset of notes illuminating the lived experiences of people with dementia. Methods We identified an analytic cohort of adults aged ≥ 65 years from the Adult Changes in Thought (ACT) Study, a prospective cohort study of dementia in Seattle, WA. We collected a corpus of 254,532 notes from the electronic medical records (2003‐present) of these participants. The corpus included a range of clinical encounters from five years prior to dementia onset to most recent visit. We implemented an NLP algorithm, pytakes, which extracted concepts from notes in our corpus. The concepts were codes adapted from a prior qualitative study, which we further refined for this study. We used these concepts to curate smaller sub‐corpora for manual, qualitative review. Results Starting with 32 unique concepts, pytakes extracted over 1.7 million concepts. 178,042 notes (69.94% of the original corpus) had at least one associated concept. Preliminary review of notes extracted showed acceptable concurrence between the text retrieved and the target concepts. Leveraging the extracted concepts, we curated, then randomly sampled 500 notes for three sub‐corpora, each aimed at a different research question. The first investigates the topics and timings of patient‐ or caregiver‐initiated communications with the healthcare system. The second relates to caregiving needs and networks of people with dementia and how these shift over time. The third revolves around housing situations and transitions of people with dementia. Our qualitative team is examining these sub‐corpora to investigate the lived experience of people with dementia. Conclusions NLP tools may be a productive method for identifying materials suited to qualitative inquiry into the lives of people with dementia. This innovative approach can provide access to data that would otherwise be difficult to obtain, offering insight into the lived experiences, care needs, living situations, and social networks of people living with dementia.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.046 | 0.116 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.011 | 0.007 |
| Science and technology studies | 0.004 | 0.005 |
| Scholarly communication | 0.004 | 0.006 |
| Open science | 0.003 | 0.007 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".