Electronic Nose Sensor Drift Affects Diagnostic Reliability and Accuracy of Disease-Specific Algorithms

Area of Science:

• Biomedical Engineering
• Gastroenterology
• Sensor Technology

Background:

• Electronic nose (eNose) technology offers potential for disease diagnosis but is susceptible to sensor drift.
• Sensor drift, a known disadvantage, can compromise the accuracy of diagnostic algorithms.
• Routine correction for sensor drift is not standard practice in eNose applications.

Purpose of the Study:

• To investigate the impact of eNose sensor drift on developing a disease-specific algorithm for inflammatory bowel disease (IBD).
• To assess the influence of sensor drift on the accuracy of differentiating IBD patients from controls using fecal samples.

Main Methods:

• A multi-center cohort study involving 63 IBD patients and 63 controls.
• Fecal samples collected prior to bowel lavage during colonoscopy.
• Fecal eNose profiles measured using Cyranose 320® over four consecutive days.
• Disease activity assessed via endoscopy; controls had no endoscopic abnormalities.

Main Results:

• Sensor data showed reproducible associations with the date of measurement, independent of clinical factors.
• Correction for sensor drift significantly improved the accuracy of differentiating IBD patients from controls (accuracy of 0.68).
• Six specific sensors demonstrated significant differences after drift correction (p-values provided).

Conclusions:

• Short-term sensor drift profoundly affected fecal eNose profiles in this IBD cohort.
• Sensor drift correction is essential for enhancing the reliability and repeatability of eNose studies.
• Implementing drift correction protocols is vital for robust clinical applications of eNose technology.