Emergency Department Infection Prediction Model with Clinical Variables

Sara N. Søgaard's study presents a machine learning model for predicting infection at emergency department admission. The research evaluated four algorithms, with a Random Forest model achieving an accuracy of 80% and an AUC of 84%. The model uses clinical variables like C-reactive protein, leucocyte count, temperature, diastolic blood pressure, and heart rate.

Use Cases

• Training infection prediction models based on clinical variables like C-reactive protein and leucocyte count.
• Benchmarking machine learning algorithms (Logistic Regression, KNN, Decision Tree, Random Forest) for clinical classification tasks.
• Developing early warning systems for infection detection in emergency department settings based on the described methodology.

Strengths

• Model performance metrics are provided, including a Random Forest specificity of 75%, PPV of 94%, and accuracy of 80%.
• The study compares four distinct machine learning algorithms with clear evaluation metrics (AUC, AP, sensitivity, specificity).
• Key predictive clinical variables are explicitly named: C-reactive protein, leucocyte count, temperature, diastolic blood pressure, and heart rate.

Limitations

• The dataset is a 17.8 KB document describing a model; the underlying patient data used for training is not included.
• Row count and column-level documentation for the original training data are unknown.
• The description metadata is limited; actual data quality and model implementation details require inspection of the document.

Provenance

Source

Sara N. Søgaard via figshare

Collection Method

Research study applying machine learning to clinical data, likely from emergency department admissions.

Freshness

Last updated 2026-04-30 05:38:40; freshness should be verified.