Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1776
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-18T01:10:31.178880Z
status: published
lens: g_space
sector: health
headline: Dynamic Pressure Ulcer Risk Predictions for Hospitalized Patients: Development and Validation of a Machine Learning Model With Nurses
dek: Aim To develop and validate a dynamic prediction model for daily pressure ulcer risk predictions together with an expert group of nurses. Design Diagnostic study. Methods All admissions to general wards in a single-center tertiary university hospital in the Netherlands were included. A retrospective dataset with candidate predictors was collected from the electronic health records and split into a training set (December 2021-November 2022, N = 19,931) and a validation set (January 2023-June 2023, N = 11,387). Th…
gain_title: A nurse-co-designed machine learning model for daily pressure ulcer risk discriminated better than the Waterlow score on a held-out validation set of general ward admissions at one Dutch hospital.
problem_title: (none)
trace_subject: (none)
gain_reading: A nurse-co-designed machine learning model for daily pressure ulcer risk discriminated better than the Waterlow score on a held-out validation set of general ward admissions at one Dutch hospital.
gain_evidence: The model clearly outperformed the Waterlow score | The AUROC for the final model was 0.790 (≤ 72 h) and 0.795 (> 72 h) on the validation dataset.
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built and validated a dynamic model that produces daily pressure ulcer risk predictions for patients on general wards of a tertiary university hospital in the Netherlands. They used electronic health record data and worked with an expert group of nurses to define predictors and modelling choices.

The result suggests that nurse-informed machine learning could support more targeted prevention and lower registration burden than a static score. The evidence comes from retrospective data at a single center, so how it performs in routine care and whether it reduces ulcers remain untested here.
limitation: The model was developed and validated using retrospective data from general wards at one center, so performance elsewhere and its effect on patient outcomes are not established in this study.
tag: Evidence-backed gain
key_points: The training set covered December 2021 to November 2022 with 19,931 admissions; the validation set covered January to June 2023 with 11,387 admissions. | Separate models were built for the first 72 hours of admission and for later than 72 hours. | The final model was logistic regression with an L2 penalty and spline transformation on a limited predictor set. | Expected remaining length of stay, age and Activities of Daily Living score contributed most strongly. | The model is named DRAAI, Decubitus Risk Alert based on AI, where the Dutch word means 'turn'.
rundown: The study was a diagnostic study that included all admissions to general wards in a single tertiary university hospital. The pressure ulcer outcome was identified from both structured and free text registration in the health records.

Candidate models performed similarly on the training set, and the team chose a relatively simple penalized logistic regression. Predictor definition and selection were discussed with nurses, with attention to implementation and user acceptance.

On validation data, AUROC was 0.790 for the first 72 hours and 0.795 afterward. The reported Waterlow comparison figures were 0.816 vs. 0.702 for the first 72 hours and 0.800 vs. 0.677 after 72 hours. The authors state the model facilitates targeted prevention and reduces registration burden.
sources:
- peer_reviewed | Journal of Clinical Nursing | https://doi.org/10.1111/jocn.70574 | 2026-10-07
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

9b540fc5ec9196b221213077838414bdbd4be9c20bf4cc82495640df0b13ee69

previous

0000000000000000000000000000000000000000000000000000000000000000