Current breathomics--a review on data pre-processing techniques and machine learning in metabolomics breath analysis

Area of Science:

• Metabolomics
• Analytical Chemistry
• Computational Biology

Background:

• Breathomics studies exhaled volatile organic compounds (VOCs) to detect health status changes.
• Non-invasive diagnostic tools are in demand, and breathomics offers significant promise.
• Current analytical methods for VOCs are advanced, but machine learning applications are nascent.

Purpose of the Study:

• To describe the state-of-the-art in data pre-processing for breathomics.
• To review multivariate analysis and machine learning techniques for VOC profile fingerprinting.
• To highlight the potential of data fusion methods in breathomics.

Main Methods:

• Data pre-processing pipelines for gas-chromatography mass spectrometry and ion-mobility spectrometry.
• Application of machine learning techniques for identifying discriminatory VOCs.
• Discussion of advantages and drawbacks of various multivariate statistical methods.

Main Results:

• Established pre-processing pipelines yield matrices of VOC abundances.
• Machine learning methods can identify VOC patterns related to metabolic processes.
• Demonstration of the utility and limitations of different analytical techniques.

Conclusions:

• Accurate data pre-processing is crucial for effective breathomics analysis.
• Machine learning is essential for interpreting complex VOC profiles.
• Further research into data fusion can enhance breathomics diagnostic capabilities.