Statistical analysis of MCC-IMS data for two group comparisons-an exemplary study on two devices

Area of Science:

• Analytical Chemistry
• Biomedical Engineering
• Computational Biology

Background:

• Multi-capillary-column-Ion-mobility-spectrometry (MCC-IMS) is a promising technology for breath gas analysis.
• Accurate classification of breath samples is crucial for distinguishing between healthy and diseased individuals.
• Existing statistical methods often overlook confounding variables, limiting the generalizability of MCC-IMS results.

Purpose of the Study:

• To evaluate the impact of technical devices, sex, and smoking habits on MCC-IMS breath gas measurements.
• To develop and present statistical solutions for adjusting MCC-IMS data acquired on different devices.
• To improve the accuracy and generalizability of breath sample classification using MCC-IMS.

Main Methods:

• A controlled breath gas study involving 49 healthy volunteers.
• Each participant underwent two MCC-IMS measurements, before and after orange juice consumption, on two different devices.
• Data analysis involved comparing manual (VisualNow) and automated (SGLTR-DBSCAN) evaluation methods.
• Statistical techniques including peak alignment and scaling were applied to address device variability.

Main Results:

• The study identified significant influence of technical devices on MCC-IMS measurements.
• Statistical adjustments for device variability (peak alignment and scaling) were effective in harmonizing data.
• No significant influence of sex or smoking habits on the analyzed MCC-IMS data was observed in this cohort.
• Both manual and automated data evaluation methods were employed, with focus on automated approaches for future applications.

Conclusions:

• Technical device variability is a critical confounder in MCC-IMS breath gas analysis.
• The proposed statistical methods (peak alignment and scaling) effectively mitigate device-related biases.
• These adjustments enhance the reliability and comparability of MCC-IMS data across different technical setups.
• Future research should incorporate these statistical strategies for robust disease classification from breath samples.