Predictive Model for In-Hospital Stroke in Gastrointestinal Bleeding Patients

A retrospective cohort study of gastroenterology inpatients at Nanchong Central Hospital between September 2020 and December 2025. The dataset, created by Yiqian Liang, compares the incidence of in-hospital acute cerebral infarction between patients with and without acute gastrointestinal bleeding and develops a six-factor predictive nomogram.

Use Cases

• Training a binary classification model to predict in-hospital stroke based on patient history and clinical factors.
• Validating the association between gastrointestinal bleeding and stroke incidence in an inpatient population.
• Developing a clinical decision support tool using the described six-factor nomogram for risk assessment.

Strengths

• The predictive nomogram demonstrated good discrimination with an AUC of 0.864.
• The study includes a substantial cohort with 2,380 matched pairs after propensity score matching.
• Analysis identified six independent predictors, including previous cerebral infarction history (OR 13.47) and sepsis (OR 2.43).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying data is unknown, which may limit suitability assessment.
• The dataset is from a single-center study, which may limit generalizability.

Provenance

Source

figshare

Collection Method

Retrospective cohort study using hospital records.

Time Range

September 2020 to December 2025

Freshness

Last updated 2026-05-13 04:41:24.

Geography

Nanchong Central Hospital, China