Related Experiment Video
Updated: Jan 4, 2026
06:49
Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
Published on: December 11, 2015
9.3K
Gold Standard Evaluation of an Automatic HAIs Surveillance System.
Beatriz Villamarín-Bello1, Berta Uriel-Latorre1, Florentino Fdez-Riverola2,3,4
1Preventive Medice Service, Complexo Hospitalario Universitario de Ourense, Rúa Ramón Puga 52-56, 32004 Ourense, Spain.
Biomed Research International
|October 31, 2019
Summary
The InNoCBR system enhances hospital-acquired infections (HAIs) surveillance using electronic health records. It shows promising accuracy in detecting HAIs, especially bloodstream infections, with potential for improvement in respiratory infection detection.
Area of Science:
- Healthcare Informatics
- Infectious Disease Epidemiology
- Clinical Informatics
Background:
- Hospital-acquired infections (HAIs) surveillance is crucial for effective prevention programs.
- Electronic health records (EHR) have enabled the development of automated HAI surveillance methods.
- Validating new automated systems against established diagnostic standards is essential.
Purpose of the Study:
- To validate the InNoCBR system for automated hospital-acquired infections (HAIs) surveillance.
- To assess the accuracy of the InNoCBR system against a gold standard for HAI diagnosis.
- To evaluate the performance of InNoCBR in both autonomous and semi-automatic modes.
Main Methods:
- Deployment of the InNoCBR system at Ourense University Hospital Complex.
- Validation against the gold standard of HAIs diagnosis.
- Performance metrics: sensitivity, specificity, positive predictive value, and kappa index.
- Analysis of performance variation by infection type.
Main Results:
- In autonomous mode, InNoCBR achieved 70.83% sensitivity, 97.76% specificity, and 77.24% positive predictive value.
- The kappa index for infection type classification was 0.67.
- Sensitivity varied by infection type, with bloodstream infections at 93.33% and respiratory infections at 53.33%.
- In semi-automatic mode, InNoCBR demonstrated higher performance: 81.73% sensitivity, 99.47% specificity, and 94.33% positive predictive value.
Conclusions:
- The InNoCBR system shows significant potential for automated HAI surveillance, particularly in autonomous mode.
- The system's performance varies by infection type, highlighting areas for future improvement, such as respiratory infections.
- Semi-automatic operation of InNoCBR yields high accuracy, offering a valuable tool for enhancing hospital infection control efforts.
