Invisible Exposure, Measurable Impact: How Genetics and AI Are Revealing the Hidden Health Risks of Thirdhand Smoke

Through initiatives such as Genesis AI, Berkeley Lab is helping to lead a new era of discovery in which advanced computational approaches can illuminate the biological consequences of environmental exposures and guide strategies to protect vulnerable populations.

Environmental exposures often leave traces long after the original source disappears. One such exposure: thirdhand smoke (THS), is increasingly recognized as a hidden but persistent threat to human health. Unlike secondhand smoke, which dissipates relatively quickly, THS consists of toxic chemical residues that accumulate on indoor surfaces, clothing, dust, and building materials after smoking has occurred. These residues can persist for months or even years, continuing to expose individuals long after the cigarette has been extinguished. Recent research is beginning to reveal that the health consequences of these lingering residues are more complex than previously understood, particularly when genetic susceptibility and advanced data science approaches are brought into the picture.

THS contains a mixture of harmful chemicals, including tobacco-specific nitrosamines, polycyclic aromatic hydrocarbons, heavy metals, and volatile organic compounds, many of which are known carcinogens capable of damaging DNA and disrupting cellular processes. These residues do not remain chemically static. Instead, they can react with common indoor pollutants such as ozone and nitrous acid, producing additional toxic compounds over time. As a result, indoor environments contaminated with THS may become dynamic chemical reservoirs that expose occupants through inhalation, ingestion of dust, or absorption through the skin. Evidence suggests that these exposures can influence cancer risk through multiple biological mechanisms, including direct DNA damage that may initiate tumors and chronic oxidative stress and inflammation that can promote tumor progression.

At the same time, scientists are increasingly recognizing that environmental exposures do not affect everyone equally. Genetic background plays a critical role in determining susceptibility to environmental toxicants. To better understand these interactions, researchers have turned to the Collaborative Cross (CC) mouse population, a genetically diverse model designed to capture much of the genetic variation found in human populations. By studying responses to THS exposure across many genetically distinct strains, researchers have been able to uncover striking differences in biological and behavioral outcomes that depend on genetic background.

New studies at Berkeley Lab using CC model revealed that early-life exposure to THS can alter neurological outcomes, including anxiety-related behaviors and memory formation, but the magnitude and direction of these effects vary dramatically depending on the host genotype and sex. Some genetic strains showed increased anxiety and impaired memory after exposure, while others showed minimal effects or even opposite responses. Genome-wide analyses identified thousands of genetic variants associated with behavioral responses to THS exposure, many of which are linked to biological pathways involved in axon development, synaptic organization, cognition, and neuronal signaling. These findings provide compelling evidence that environmental exposures interact with genetic variation to shape health outcomes, highlighting the importance of gene–environment interactions in both neurological function and disease susceptibility.

The nervous system may be particularly sensitive to environmental toxicants during early developmental stages. The observation that relatively short periods of THS exposure can alter anxiety-related behaviors and memory performance raises important questions about potential impacts in human populations, especially for children who experience higher exposure through contact with contaminated surfaces and dust. These findings also extend the scientific discussion of tobacco-related exposures beyond cancer to include potential effects on brain health, cognition, and mental well-being.

Understanding these complex interactions between genes and environment presents a major scientific challenge, as it requires integrating large and diverse datasets spanning genomics, environmental measurements, behavioral phenotypes, and molecular biology. Artificial intelligence and machine learning are increasingly playing a transformative role in addressing this challenge. In recent work, researchers developed machine-learning models that integrate genetic variants and environmental exposure information to predict behavioral outcomes following THS exposure. These models demonstrated strong predictive performance, illustrating how AI can uncover patterns that are difficult to detect using traditional analytical approaches.

The application of AI to environmental health research is closely aligned with the vision of Genesis AI at Berkeley Lab, an initiative aimed at harnessing advanced artificial intelligence to accelerate biological discovery. Environmental health studies generate highly complex datasets that include genomic variation, exposure profiles, physiological responses, and behavioral outcomes. AI-driven frameworks provide the ability to integrate these diverse layers of information, revealing previously hidden relationships between environmental exposures and biological systems.

By combining experimental models, genomics, and AI-enabled analytics, researchers at Berkeley Lab are beginning to construct predictive frameworks that explain how environmental exposures interact with genetic susceptibility to influence disease risk. This emerging approach, often referred to as precision environmental health, seeks to identify which individuals or populations are most vulnerable to specific environmental hazards and to design prevention strategies tailored to those risks.

THS represents only one example of the broader challenge posed by persistent environmental exposures whose health effects unfold through complex interactions among genetics, biology, and environment. As new technologies continue to expand our ability to measure exposures and analyze biological responses, the integration of environmental science, genomics, and artificial intelligence will become increasingly essential for understanding how these exposures influence human health.

Through initiatives such as Genesis AI, Berkeley Lab is helping to lead a new era of discovery in which advanced computational approaches can illuminate the biological consequences of environmental exposures and guide strategies to protect vulnerable populations. By revealing how hidden residues like THS interact with genetic susceptibility and brain function, this research highlights the growing importance of data-driven science in addressing some of the most pressing environmental health challenges of our time.

These studies, recently published in Environmental International , supported by Tobacco-Related Disease Research Program, were led by Dr. Hang Chang (Staff Scientist) and Dr. Jian-Hua Mao (Senior Scientist) at Berkeley Biomedical Data Science Center; Biological Systems and Engineering Division, Lawrence Berkeley National Lab.