A novel method for inferring RFID tag reader recordings into clinical events

Area of Science:

• Biomedical Engineering
• Health Informatics
• Infection Control

Background:

• Nosocomial infections (NIs) are critical indicators of patient safety and hospital performance.
• Higher NI rates in ICUs stem from complex patient care involving invasive procedures.
• Emerging
• superbugs
• necessitate enhanced infection control, with caregiver-patient contact being a major cross-infection route.

Purpose of the Study:

• To develop a Contact History Inferential Model (CHIM) using radio frequency identification (RFID) technology.
• To infer clinical events from RFID tag reader recordings.
• To validate the CHIM model through real-time observations in an ICU setting.

Main Methods:

• A pre-study involved testing RFID proximity sensing and deployment in a Clinical Skill Center.
• The CHIM was developed using variables like duration, frequency, and caregiver ID, classifying events into close-in, contact, and invasive.
• Validation involved real-time observations by recruited observers and participatory observation by the first author in an ICU.

Main Results:

• The CHIM demonstrated effective inference of proximity events, with varying accuracy for different event types.
• Sensitivity, specificity, and accuracy for close-in events were 73.8%, 83.8%, and 81.6%, respectively.
• Contact events achieved 81.4%, 78.8%, and 80.7% for sensitivity, specificity, and accuracy, while invasive events reached 90.9%, 98.0%, and 97.5%.

Conclusions:

• Proximity sensing via RFID technology effectively detects and infers proximity events, aiding in complete clinical contact history recording.
• The CHIM accurately infers ICU activities, suggesting its applicability in other hospital wards.
• This technology assists in tracing the causes of nosocomial infections and can be used for additional purposes.