Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2269
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-26T18:34:06.678106Z
status: published
lens: g_space
sector: health
headline: Improving Haemorrhage Prediction in General Ward Patients by Incorporating Patterns of Change Derived From Routine Ward Measurements
dek: To evaluate whether patterns of change derived from routine ward measurements provide additional predictive value beyond the most recent clinical measurements for haemorrhage prediction in general ward patients. A retrospective cohort study. Among 16,197 adult patients admitted to general wards in three university-affiliated hospitals in South Korea, 1152 (7.1%) experienced haemorrhage, whereas 15,045 (92.9%) did not. Three machine-learning algorithms evaluated three feature sets based on the most recent clinica…
gain_title: Machine-learning models that add patterns of change in routine ward measurements to the latest values predicted haemorrhage slightly better than models using the latest values alone.
problem_title: (none)
trace_subject: (none)
gain_reading: Machine-learning models that add patterns of change in routine ward measurements to the latest values predicted haemorrhage slightly better than models using the latest values alone.
gain_evidence: Models using both recent measurements and patterns of change showed small but statistically supported improvements in the area under the receiver operating characteristic curve
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers analysed records of 16,197 adults admitted to general wards at three university-affiliated hospitals in South Korea, of whom 1152 had a haemorrhage. They trained three machine-learning algorithms on the most recent clinical measurements, on patterns of change over time, or on both combined.

The result suggests that routine ward data hold predictive signal beyond the latest values, which could eventually support nursing surveillance and earlier intervention. The improvement was small and tested only through internal validation, so whether it helps patients remains unknown until prospective evaluation.
limitation: The models were only internally validated in a retrospective design, and the gains were small, so the authors say prospective evaluation is needed to establish clinical utility.
tag: Evidence-backed gain
key_points: The cohort included 16,197 adult general ward patients, and 1152 (7.1%) experienced haemorrhage. | Three machine-learning algorithms were compared across three feature sets: recent measurements, patterns of change, or both. | Combining both feature sets gave small but statistically supported gains in area under the curve over recent measurements alone. | Interpretation with Shapley Additive Explanations named neutrophil count, platelet count, white blood cell count, urine output and vital signs as key predictors.
rundown: The study was a retrospective cohort analysis across three university-affiliated hospitals in South Korea. Patients who bled were compared with 15,045 who did not, using routinely collected ward data.

Models were trained and internally validated with stratified five-fold cross-validation. Performance was compared using area under the receiver operating characteristic curve, sensitivity, specificity and F1-score.

The authors say such models may support risk stratification in general ward patients. They also state that prospective evaluation is needed to determine clinical utility and the value for nursing surveillance.
sources:
- peer_reviewed | Journal of Clinical Nursing | https://doi.org/10.1111/jocn.70611 | 2026-10-25
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

b9a6a08432b692959f7c38b1877207fc2e8cddbfc1d81636b1f5e87ecca6ca5f

previous

0000000000000000000000000000000000000000000000000000000000000000