Clinical applications and challenges of wearable respiratory analysis

The presence of organic substances in exhaled breath was first demonstrated in the 1970s by the first analysis of breath using chromatography. Since then, more than 3,000 volatile organic compounds (VOCs) have been identified in breath that can be explored for medical applications. However, with the exception of breathalyzers that measure alcohol content, only a few sensing devices have reached a high technology maturity level, and diagnostic applications have been limited to chromatographic laboratory tests. Advances in materials science, synthetic biology, and engineering, as well as increased interest in diagnosing infectious diseases from exhaled breath, have recently driven the popularity of breath analysis.

According to MEMS Consulting, researchers from the University of Freiburg in Germany recently published an article titled "Wearable breath analysis" in Nature Reviews Bioengineering, highlighting the milestones and challenges faced by wearable devices for breath analysis in widespread clinical applications.

Breath analysis as a diagnostic tool

Breath analysis can be broadly interpreted as a two-step process: verifying the presence of diagnostic breath markers and establishing methods for their detection. In so-called “breath prints,” specific VOCs associated with certain diseases or metabolic activities have been identified through clinical trials. Breath analysis can therefore be used as a non-invasive, qualitative diagnostic tool. Breath sampling has a distinct advantage over other non-invasive sampling methods because the transfer of analytes from the blood to the lungs bypasses complex transport mechanisms.

Wearable Breathing Sensor

The transition from laboratory-based measurement techniques to wearable biosensors is necessary to accelerate the translation of breath analysis. Wearable devices in the form of masks can detect hydrogen peroxide (a biomarker for several respiratory diseases such as asthma, chronic obstructive pulmonary disease, and cancer) through electrochemical sensing and monitor respiratory conditions (such as respiratory rate, cough, and breath holding) by integrating self-powered pressure sensors. The COVID-19 pandemic has further brought breath analysis to the forefront as a method for detecting airborne infectious diseases from virus-containing aerosols. For example, masks with freeze-dried CRISPR sensors can detect nucleic acids from the new coronavirus (SARS-CoV-2) using traditional lateral flow analysis; optical sniffers (colorimetric sensor arrays) can semi-quantitatively analyze the severity of COVID-19 (from very mild to severe) by leveraging the relationship between color patterns and patient medical reports and real-time polymerase chain reaction (RT-PCR) results.

To realize the clinical potential of wearable breath analysis, several challenges need to be addressed.

Sensitivity

A wide range of breath analytes (approximately 3,500) have been identified, including non-physiologically produced exogenous markers; however, their concentrations in breath are often very low (1,000-10,000 times lower than in blood), depending on age, diet, smoking and medications, and expiratory flow rate. Therefore, wearable breath analyzers must be highly sensitive to detect breath analytes. The target molecules need to be concentrated either on the wearable device or at the source (e.g., lungs and/or airways).

In wearable devices, breath analytes can be separated in microfluidic systems as a pre-concentration step (e.g., μ-GC). In addition, the sensing element can be enhanced in sensitivity by using synthetic biology-based bioassays (e.g., CRISPR–Cas-based systems, microbial enzymes, and proteins), or nanostructures can be constructed to increase the effective sensing area. Target molecules can also be concentrated at the source by stimulating metabolic activity; alternatively, exogenous biomarkers can be expressed at the source by exploiting metabolic activity associated with a disease (e.g., lung cancer) or by applying engineered living organisms. The reactivity of the analyte can also be increased by chemical pretreatment or by forming analyte-nanoparticle conjugates.

Selectivity

To measure specific analytes in a mixture of substances with similar chemical and/or physical properties, incompatible design approaches can be adopted. In the top-down approach, sensor arrays are used to improve selectivity. Such sensing units need to be robust and apply a series of chemical interactions to capture different combinations of potential information compounds. The interpretation of the results relies on the similarity or difference of the measurements. Therefore, large amounts of data must be compiled and processed with the help of pattern extraction models to associate physiological states with sensory fingerprints. However, the same subset of sensors can be excited by different combinations of analytes, which makes it difficult to associate patterns with specific conditions. This dilemma can be overcome by balancing the complexity between sensory problems, hardware, and model architecture. Breath has a rich analyte chemical composition, and a priori predictions of relevant subsets of analytes remain challenging. Therefore, multi-array sensing technologies should be complex enough to respond to changes in composition. The only known sensing unit that reflects this complexity is the olfactory receptor, which can be integrated into the biohybrid sensor of a wearable breath analyzer. In addition, the data must be processed with black box models of similar complexity, such as deep learning methods, which can only be trained with large amounts of data.

sampling

Breath composition is strongly influenced by the sample collection method, which has not been standardized. The relative ratios of analytes vary with the choice of breath fraction (end-tidal breath), breathing pattern, sample contamination (e.g., saliva), sample collection mode (online continuous or offline discrete), and sample phase (vapor (gaseous) or condensate). Therefore, end-to-end wearable design studies should choose a sample collection strategy that allows continuous access to physiological states. For example, disposable masks can integrate sampling, sample preparation (if needed), and sensing modules. Alternatively, subnasal or intranasal patches, or implants mounted on the respiratory tract (potentially self-powered using metabolites such as lactate or glucose) can be used. Sensor-integrated nebulizers can achieve closed-loop drug delivery for the treatment of diseases.

Wearable devices for breath sampling and analysis

Ethical considerations

Ethical considerations associated with wearable respiratory analysis include patient data collection, patient safety, liability, legal responsibility, user compliance, accessibility, and fairness, which all need to be considered in the design and clinical application of wearable respiratory sensors. Patient safety and liability are particular concerns for wearable devices in therapeutic applications. Legal liability arises from the integration of patient data into the data pipeline, as well as errors or biases in the selection of models and training databases, which in turn determine the contribution of the wearable device's output to the final diagnostic judgment. In addition, given the programming languages, databases, and medical clues for certain diseases, misuse of the device and lack of software updates can lead to liability issues.

By addressing the technical and ethical challenges faced, wearable breath analysis can become a complementary tool for distributed preventive healthcare monitoring and transform our understanding of diagnostics.