The Future Of Early Disease Detection: Breath-based Screening & Diagnostics

Ancient physicians — as far back as Hippocrates and Galenus — used their sense of smell to diagnose disease. A sweet, fruity breath indicated diabetes. A fishy or ammonia-scented breath suggested kidney disease. Advanced liver disease produced a musty, sweet, or metallic breath. While the role of smell in diagnostics has faded since the era of these Greek physicians in favor of other modalities such as blood and sputum, my view is that smell is on the brink of an extraordinary comeback and will, indeed, be a central tool in the future of medicine.

‍

Long before physical symptoms of disease and illness are evident, they alter the body’s metabolism to produce a combination of volatile organic compounds (VOCs) that can be used as a biomarker. These biomarkers are released from the body each time a person exhales and can be captured and decoded to provide a real-time snapshot of someone’s health. This approach to capturing and interpreting biomarkers from the breath is known as breath-based diagnostics or breath biopsies.

Several studies have already identified early breath biomarkers for breast, lung, and gastric cancers, metabolic diseases such as diabetes, cardiorespiratory disease, infectious diseases, and even neurodegenerative conditions such as Parkinson’s and Alzheimer’s.

The Case For — And Challenges Of — Breath-based Diagnostics

Capturing these unique biomarkers and using them to detect disease in real time is both promising and exciting. The current state of screening and diagnostics is plagued with issues, posing significant barriers to early detection and timely intervention. Extended wait times lead to later testing of symptomatic individuals, while those that are asymptomatic or present with atypical symptoms often slip through the cracks all together. Moreover, conventional diagnostic procedures can be uncomfortable and intrusive, limiting patient compliance. The confinement of testing to traditional points of care adds an additional layer of difficulty, hindering accessibility. Finally, the costs associated with diagnostic tests can prove to be prohibitive to many.

The cumulative impact of these issues contributes to delayed diagnoses and escalating treatment costs within the healthcare system, ultimately heightening the prevalence of preventable illness, morbidity, and mortality worldwide.

Harnessing newer technologies presents an unprecedented opportunity to transform disease detection, paving the way for earlier and more accessible interventions that can dramatically improve patient outcomes while alleviating the global burden on healthcare systems. Breath-based diagnostics emerges as a particularly promising technology, offering a rapid, painless, and non-invasive alternative to conventional methods. However, there are no doubt certain challenges that need to be addressed to unlock its full potential at scale. Some of these challenges include:

• Low Concentrations of VOCs: VOCs, the key building blocks of the biomarkers, exist in very low concentrations, often measured in parts per million or less.
• Lack of Standardized Biomarker Libraries: The absence of a standardized database of breath biomarkers poses a significant obstacle in utilizing them at scale as reliable indicators of disease.
• Confounding Backgrounds: Exhaled breath includes the presence of VOCs from an external source, condensing humidity, and even respiratory droplets. Each of these can affect accuracy, which makes it particularly challenging to build a reliable and scalable solution.
• Generalizability: For widespread applicability, it must work irrespective of variations at the population or individual levels (such as environmental, geographical, or dietary differences).
• Reliable Performance: The breath-based platform must meet rigorous standards, ensuring high accuracy, reproducibility, and verifiability against gold-standard methods.

​​Even if all these challenges were to be resolved, a fundamental impediment persists — studies conducted to identify breath-based biomarkers have been predominantly reliant on highly specialized equipment that is rarely accessible outside academic or research labs.

Such equipment (gas chromatography and mass spectrometry, or GC/MS) is characterized by its substantial physical footprint, high procurement and maintenance costs, dependence on highly technical expertise for operation, and lack of scalability. In essence, this makes the practical deployment of breath-based diagnostic tools seem implausible.

But that is starting to change with a paradigm shift in disease screening and diagnostics.

‍

Novel Technologies And The Vital Role Of Machine Intelligence

The drive for rapid, low-cost, and non-invasive methods of disease screening and diagnosis has ignited a surge of innovation in novel technologies. Among them, digital olfaction technologies, often called "digital noses," stand out as a highly promising avenue for achieving portable, reliable, and scalable detection of breath biomarkers in disease screening.

That said, not all digital noses are created equal. The blueprint for a true digital nose needs to draw its inspiration from the gold standard in scent detection — the mammalian sense of smell. In its simplest form, a mammal's olfactory system is made up of three pivotal components that would need to be replicated. The first is a sophisticated front end capable of capturing odor signals and transforming them into distinctive “scent prints” — the digital equivalent of a nose complete with olfactory nerves and receptors. The second is an analysis engine with the capability to decode these scent prints, effectively discerning their meaning — the digital counterpart to the brain. However, this engine's functionality hinges on the third component — a repository of known odor scent prints to reference, which requires having the digital equivalent of a mammal’s ability to learn.

Accordingly, the foundation of building a high-performance digital olfaction platform lies in having a sensor chip equipped with a diverse array of chemical receptors. This chip serves as the digital nose, capturing odor information and producing digital scent prints. However, despite being an essential facet, developing a digital nose sensor chip is merely the starting point. While it addresses the first piece of the trifecta — capturing odors — the chip alone lacks the transformative power, or the brain, needed to decode the true meaning and significance of the detected odors.

This is precisely where machine intelligence takes center stage. Through sophisticated algorithms and advanced processing capabilities, machine intelligence acts as the digital brain, elevating the sensor chip's output to a level of understanding that brings meaning and elicits action. Artificial intelligence engines, finely tuned through training, can adeptly discern noise from signals, extracting relevant biomarkers from a breath sample while navigating through confounding background noise. Beyond this, the most sophisticated AI engines would also be capable of performing reliably across diverse devices, environments, and population groups — a key attribute known as generalizability.

This fusion of cutting-edge sensor technology and sophisticated machine intelligence propels us into uncharted territories, pushing the boundaries of medical innovation.

‍

A New Era In Healthcare

The emergence of breath-based screening and diagnostic devices, such as handheld breathalyzers, is poised to revolutionize point-of-care diagnostics. These portable devices would be cost-effective to operate and simple to use, allowing them to be seamlessly integrated into virtually any point-of-care setting, ranging from large urban hospitals to local clinics in rural environments.

A future marked by widespread availability of reliable, portable, and affordable breath-based diagnostic devices promises unprecedented access to disease testing, enabling earlier diagnoses. This in turn will enable a profound transformation in patient outcomes through more timely and effective medical interventions, particularly in the early stages of diseases when positive responses to treatment are most likely.