Detection of COPD and Lung Cancer with electronic nose using ensemble learning methods

Ultra-sensitive analysis of exhaled biomarkers in ozone-exposed mice via PAI-TOFMS assisted with machine learning algorithms

Ground-level ozone ranks sixth among common air pollutants. It worsens lung diseases like asthma, emphysema, and chronic bronchitis. Despite recent attention from researchers, the link between exhaled breath and ozone-induced injury remains poorly understood. This study aimed to identify novel exhaled biomarkers in ozone-exposed mice using ultra-sensitive photoinduced associative ionization time-of-flight mass spectrometry and machine learning. Distinct ion peaks for acetonitrile (m/z 42, 60, and 78), butyronitrile (m/z 70, 88, and 106), and hydrogen sulfide (m/z 35) were detected. Integration of tissue characteristics, oxidative stress-related mRNA expression, and exhaled breath condensate free-radical analysis enabled a comprehensive exploration of the relationship between ozone-induced biological responses and potential biomarkers. Under similar exposure levels, C57BL/6 mice exhibited pulmonary injury characterized by significant inflammation, oxidative stress, and cardiac damage. Notably, C57BL/6 mice showed free radical signals, indicating a distinct susceptibility profile. Immunodeficient non-obese diabetic Prkdc-/-/Il2rg-/- (NPI) mice exhibited minimal biological responses to pulmonary injury, with little impact on the heart. These findings suggest a divergence in ozone-induced damage pathways in the two mouse types, leading to alterations in exhaled biomarkers. Integrating biomarker discovery with comprehensive biopathological analysis forms a robust foundation for targeted interventions to manage health risks posed by ozone exposure.

Cross-site validation of lung cancer diagnosis by electronic nose with deep learning: a multicenter prospective study

BACKGROUND

Although electronic nose (eNose) has been intensively investigated for diagnosing lung cancer, cross-site validation remains a major obstacle to be overcome and no studies have yet been performed.

METHODS

Patients with lung cancer, as well as healthy control and diseased control groups, were prospectively recruited from two referral centers between 2019 and 2022. Deep learning models for detecting lung cancer with eNose breathprint were developed using training cohort from one site and then tested on cohort from the other site. Semi-Supervised Domain-Generalized (Semi-DG) Augmentation (SDA) and Noise-Shift Augmentation (NSA) methods with or without fine-tuning was applied to improve performance.

RESULTS

In this study, 231 participants were enrolled, comprising a training/validation cohort of 168 individuals (90 with lung cancer, 16 healthy controls, and 62 diseased controls) and a test cohort of 63 individuals (28 with lung cancer, 10 healthy controls, and 25 diseased controls). The model has satisfactory results in the validation cohort from the same hospital while directly applying the trained model to the test cohort yielded suboptimal results (AUC, 0.61, 95% CI: 0.47─0.76). The performance improved after applying data augmentation methods in the training cohort (SDA, AUC: 0.89 [0.81─0.97]; NSA, AUC:0.90 [0.89─1.00]). Additionally, after applying fine-tuning methods, the performance further improved (SDA plus fine-tuning, AUC:0.95 [0.89─1.00]; NSA plus fine-tuning, AUC:0.95 [0.90─1.00]).

CONCLUSION

Our study revealed that deep learning models developed for eNose breathprint can achieve cross-site validation with data augmentation and fine-tuning. Accordingly, eNose breathprints emerge as a convenient, non-invasive, and potentially generalizable solution for lung cancer detection.

CLINICAL TRIAL REGISTRATION

This study is not a clinical trial and was therefore not registered.

Machine learning-based clinical decision support using laboratory data

Artificial intelligence (AI) and machine learning (ML) are becoming vital in laboratory medicine and the broader context of healthcare. In this review article, we summarized the development of ML models and how they contribute to clinical laboratory workflow and improve patient outcomes. The process of ML model development involves data collection, data cleansing, feature engineering, model development, and optimization. These models, once finalized, are subjected to thorough performance assessments and validations. Recently, due to the complexity inherent in model development, automated ML tools were also introduced to streamline the process, enabling non-experts to create models. Clinical Decision Support Systems (CDSS) use ML techniques on large datasets to aid healthcare professionals in test result interpretation. They are revolutionizing laboratory medicine, enabling labs to work more efficiently with less human supervision across pre-analytical, analytical, and post-analytical phases. Despite contributions of the ML tools at all analytical phases, their integration presents challenges like potential model uncertainties, black-box algorithms, and deskilling of professionals. Additionally, acquiring diverse datasets is hard, and models' complexity can limit clinical use. In conclusion, ML-based CDSS in healthcare can greatly enhance clinical decision-making. However, successful adoption demands collaboration among professionals and stakeholders, utilizing hybrid intelligence, external validation, and performance assessments.

Detection of lung cancer and stages via breath analysis using a self-made electronic nose device

BACKGROUND

Breathomics is an emerging area focusing on monitoring and diagnosing pulmonary diseases, especially lung cancer. This research aims to employ metabolomic methods to create a breathprint in human-expelled air to rapidly identify lung cancer and its stages.

RESEARCH DESIGN AND METHODS

An electronic nose (e-nose) system with five metal oxide semiconductor (MOS) gas sensors, a microcontroller, and machine learning algorithms was designed and developed for this application. The volunteers in this study include 114 patients with lung cancer and 147 healthy controls to understand the clinical potential of the e-nose system to detect lung cancer and its stages.

RESULTS

In the training phase, in discriminating lung cancer from controls, the XGBoost classifier model with 10-fold cross-validation gave an accuracy of 91.67%. In the validation phase, the XGBoost classifier model correctly identified 35 out of 42 patients with lung cancer samples and 44 out of 51 healthy control samples providing an overall sensitivity of 83.33% and specificity of 86.27%.

CONCLUSIONS

These results indicate that the exhaled breath VOC analysis method may be developed as a new diagnostic tool for lung cancer detection. The advantages of e-nose based diagnostics, such as an easy and painless method of sampling, and low-cost procedures, will make it an excellent diagnostic method in the future.

Exhaled breath and urinary volatile organic compounds (VOCs) for cancer diagnoses, and microbial-related VOC metabolic pathway analysis: a systematic review and meta-analysis

BACKGROUND

The gradual evolution of the detection and quantification of volatile organic compounds (VOCs) has been instrumental in cancer diagnosis. The primary objective of this study was to assess the diagnostic potential of exhaled breath and urinary VOCs in cancer detection. As VOCs are indicative of tumor and human metabolism, our work also sought to investigate the metabolic pathways linked to the development of cancerous tumors.

MATERIALS AND METHODS

An electronic search was performed in the PubMed database. Original studies on VOCs within exhaled breath and urine for cancer detection with a control group were included. A meta-analysis was conducted using a bivariate model to assess the sensitivity and specificity of the VOCs for cancer detection. Fagan's nomogram was designed to leverage the findings from our diagnostic analysis for the purpose of estimating the likelihood of cancer in patients. Ultimately, MetOrigin was employed to conduct an analysis of the metabolic pathways associated with VOCs in relation to both human and/or microbiota.

RESULTS

The pooled sensitivity, specificity and the area under the curve for cancer screening utilizing exhaled breath and urinary VOCs were determined to be 0.89, 0.88, and 0.95, respectively. A pretest probability of 51% can be considered as the threshold for diagnosing cancers with VOCs. As the estimated pretest probability of cancer exceeds 51%, it becomes more appropriate to emphasize the 'ruling in' approach. Conversely, when the estimated pretest probability of cancer falls below 51%, it is more suitable to emphasize the 'ruling out' approach. A total of 14, 14, 6, and 7 microbiota-related VOCs were identified in relation to lung, colorectal, breast, and liver cancers, respectively. The enrichment analysis of volatile metabolites revealed a significant enrichment of butanoate metabolism in the aforementioned tumor types.

CONCLUSIONS

The analysis of exhaled breath and urinary VOCs showed promise for cancer screening. In addition, the enrichment analysis of volatile metabolites revealed a significant enrichment of butanoate metabolism in four tumor types, namely lung, colorectum, breast and liver. These findings hold significant implications for the prospective clinical application of multiomics correlation in disease management and the exploration of potential therapeutic targets.

Prediction of the benign and malignant nature of masses in COPD background based on Habitat-based enhanced CT radiomics modeling: A preliminary study

BACKGROUND

It is difficult to differentiate between chronic obstructive pulmonary disease (COPD)-peripheral bronchogenic carcinoma (COPD-PBC) and inflammatory masses.

OBJECTIVE

This study aims to predict COPD-PBC based on clinical data and preoperative Habitat-based enhanced CT radiomics (HECT radiomics) modeling.

METHODS

A retrospective analysis was conducted on clinical imaging data of 232 cases of postoperative pathological confirmed PBC or inflammatory masses. The PBC group consisted of 82 cases, while the non-PBC group consisted of 150 cases. A training set and a testing set were established using a 7:3 ratio and a time cutoff point. In the training set, multiple models were established using clinical data and radiomics texture changes within different enhanced areas of the CT mass (HECT radiomics). The AUC values of each model were compared using Delong's test, and the clinical net benefit of the models was tested using decision curve analysis (DCA). The models were then externally validated in the testing set, and a nomogram of predicting COPD-PBC was created.

RESULTS

Univariate analysis confirmed that female gender, tumor morphology, CEA, Cyfra21-1, CT enhancement pattern, and Habitat-Radscore B/C were predictive factors for COPD-PBC (P< 0.05). The combination model based on these factors had significantly higher predictive performance [AUC: 0.894, 95% CI (0.836-0.936)] than the clinical data model [AUC: 0.758, 95% CI (0.685-0.822)] and radiomics model [AUC: 0.828, 95% CI (0.761-0.882)]. DCA also confirmed the higher clinical net benefit of the combination model, which was validated in the testing set. The nomogram developed based on the combination model helped predict COPD-PBC.

CONCLUSION

The combination model based on clinical data and Habitat-based enhanced CT radiomics can help differentiate COPD-PBC, providing a new non-invasive and efficient method for its diagnosis, treatment, and clinical decision-making.

Molecular monitoring of lung allograft health: is it ready for routine clinical use?

Maintenance of long-term lung allograft health in lung transplant recipients (LTRs) requires a fine balancing act between providing sufficient immunosuppression to reduce the risk of rejection whilst at the same time not over-immunosuppressing individuals and exposing them to the myriad of immunosuppressant drug side-effects that can cause morbidity and mortality. At present, lung transplant physicians only have limited and rather blunt tools available to assist them with this task. Although therapeutic drug monitoring provides clinically useful information about single time point and longitudinal exposure of LTRs to immunosuppressants, it lacks precision in determining the functional level of immunosuppression that an individual is experiencing. There is a significant gap in our ability to monitor lung allograft health and therefore tailor optimal personalised immunosuppression regimens. Molecular diagnostics performed on blood, bronchoalveolar lavage or lung tissue that can detect early signs of subclinical allograft injury, differentiate rejection from infection or distinguish cellular from humoral rejection could offer clinicians powerful tools in protecting lung allograft health. In this review, we look at the current evidence behind molecular monitoring in lung transplantation and ask if it is ready for routine clinical use. Although donor-derived cell-free DNA and tissue transcriptomics appear to be the techniques with the most immediate clinical potential, more robust data are required on their performance and additional clinical value beyond standard of care.

Demystification of artificial intelligence for respiratory clinicians managing patients with obstructive lung diseases

INTRODUCTION

Asthma and chronic obstructive pulmonary disease (COPD) are leading causes of morbidity and mortality worldwide. Despite all available diagnostics and treatments, these conditions pose a significant individual, economic and social burden. Artificial intelligence (AI) promises to support clinical decision-making processes by optimizing diagnosis and treatment strategies of these heterogeneous and complex chronic respiratory diseases. Its capabilities extend to predicting exacerbation risk, disease progression and mortality, providing healthcare professionals with valuable insights for more effective care. Nevertheless, the knowledge gap between respiratory clinicians and data scientists remains a major constraint for wide application of AI and may hinder future progress. This narrative review aims to bridge this gap and encourage AI deployment by explaining its methodology and added value in asthma and COPD diagnosis and treatment.

AREAS COVERED

This review offers an overview of the fundamental concepts of AI and machine learning, outlines the key steps in building a model, provides examples of their applicability in asthma and COPD care, and discusses barriers to their implementation.

EXPERT OPINION

Machine learning can advance our understanding of asthma and COPD, enabling personalized therapy and better outcomes. Further research and validation are needed to ensure the development of clinically meaningful and generalizable models.

Screening of Lipid Metabolism-Related Genes as Diagnostic Indicators in Chronic Obstructive Pulmonary Disease

Objective

It has been observed that local and systemic disorders of lipid metabolism occur during the development of chronic obstructive pulmonary disease (COPD), but no specific mechanism has yet been identified.

Methods

The mRNA microarray dataset GSE76925 of COPD patients was downloaded from the Gene Expression Omnibus database and screened for differentially expressed genes (DEGs). Lipid metabolism-related genes (LMRGs) were extracted from the Kyoto Encyclopedia of Genes and Genomes database and Molecular Signature Database. The DEGs were intersected with LMRGs to obtain differentially expressed lipid metabolism-related genes (DeLMRGs). GO enrichment analysis and KEGG pathway analysis were performed on DeLMRGs, and protein-protein interaction networks were constructed and screened to identify hub genes. The GSE8581 validation set and further ELISA experiments were used to validate key DeLMRG expression.

Results

Differential analysis of dataset GSE76925 identified 587 DEGs, of which 62 genes were up-regulated and 525 were down-regulated. Taking the intersection of 587 DEGs with 1102 LMRGs, 20 DeLMRGs were obtained, including 1 up-regulated gene and 19 down-regulated genes. 10 hub genes were screened by cytohubba plugin, including 9 down-regulated genes PLA2G4A, HPGDS, LEP, PTGES3, LEPR, PLA2G2D, MED21, SPTLC1 and BCHE, as well as the only up-regulated gene PLA2G7. Validation of the identified 10 DeLMRGs using the validation set GSE8581 revealed that BCHE and PLA2G7 expression levels differed between the two groups. We further constructed the ceRNA network of BCHE and PLA2G7. Cell experiments also showed that PLA2G7 expression was up-regulated and BCHE expression was down-regulated in CSE-treated RAW264.7 and THP-1 cells.

Conclusion

Based on a comprehensive bioinformatic analysis of lipid metabolism genes, we identified BCHE and PLA2G7 as potentially significant biomarkers of COPD. These biomarkers may represent promising targets for COPD diagnosis and treatment.

Automated detection of airflow obstructive diseases: A systematic review of the last decade (2013-2022)

BACKGROUND AND OBJECTIVE

Obstructive airway diseases, including asthma and Chronic Obstructive Pulmonary Disease (COPD), are two of the most common chronic respiratory health problems. Both of these conditions require health professional expertise in making a diagnosis. Hence, this process is time intensive for healthcare providers and the diagnostic quality is subject to intra- and inter- operator variability. In this study we investigate the role of automated detection of obstructive airway diseases to reduce cost and improve diagnostic quality.

METHODS

We investigated the existing body of evidence and applied Preferred Reporting Items for Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to search records in IEEE, Google scholar, and PubMed databases. We identified 65 papers that were published from 2013 to 2022 and these papers cover 67 different studies. The review process was structured according to the medical data that was used for disease detection. We identified six main categories, namely air flow, genetic, imaging, signals, and miscellaneous. For each of these categories, we report both disease detection methods and their performance.

RESULTS

We found that medical imaging was used in 14 of the reviewed studies as data for automated obstructive airway disease detection. Genetics and physiological signals were used in 13 studies. Medical records and air flow were used in 9 and 7 studies, respectively. Most papers were published in 2020 and we found three times more work on Machine Learning (ML) when compared to Deep Learning (DL). Statistical analysis shows that DL techniques achieve higher Accuracy (ACC) when compared to ML. Convolutional Neural Network (CNN) is the most common DL classifier and Support Vector Machine (SVM) is the most widely used ML classifier. During our review, we discovered only two publicly available asthma and COPD datasets. Most studies used private clinical datasets, so data size and data composition are inconsistent.

CONCLUSIONS

Our review results indicate that Artificial Intelligence (AI) can improve both decision quality and efficiency of health professionals during COPD and asthma diagnosis. However, we found several limitations in this review, such as a lack of dataset consistency, a limited dataset and remote monitoring was not sufficiently explored. We appeal to society to accept and trust computer aided airflow obstructive diseases diagnosis and we encourage health professionals to work closely with AI scientists to promote automated detection in clinical practice and hospital settings.

Breath volatile organic compounds (VOCs) as biomarkers for the diagnosis of pathological conditions: A review

Normal and abnormal/pathological status of physiological processes in the human organism can be characterized through Volatile Organic Compounds (VOCs) emitted in breath. Recently, a wide range of volatile analytes has risen as biomarkers. These compounds have been addressed in the scientific and medical communities as an extremely valuable metabolic window. Once collected and analysed, VOCs can represent a tool for a rapid, accurate, non-invasive, and painless diagnosis of several diseases and health conditions. These biomarkers are released by exhaled breath, urine, faeces, skin, and several other ways, at trace concentration levels, usually in the ppbv (μg/L) range. For this reason, the analytical techniques applied for detecting and clinically exploiting the VOCs are extremely important. The present work reviews the most promising results in the field of breath biomarkers and the most common methods of detection of VOCs. A total of 16 pathologies and the respective database of compounds are addressed. An updated version of the VOCs biomarkers database can be consulted at: https://neomeditec.com/VOCdatabase/.

Early Detection of Prostate Cancer: The Role of Scent

Prostate cancer (PCa) represents the cause of the second highest number of cancer-related deaths worldwide, and its clinical presentation can range from slow-growing to rapidly spreading metastatic disease. As the characteristics of most cases of PCa remains incompletely understood, it is crucial to identify new biomarkers that can aid in early detection. Despite the prostate-specific antigen serum (PSA) levels, prostate biopsy, and imaging representing the actual gold-standard for diagnosing PCa, analyzing volatile organic compounds (VOCs) has emerged as a promising new frontier. We and other authors have reported that highly trained dogs can recognize specific VOCs associated with PCa with high accuracy. However, using dogs in clinical practice has several limitations. To exploit the potential of VOCs, an electronic nose (eNose) that mimics the dog olfactory system and can potentially be used in clinical practice was designed. To explore the eNose as an alternative to dogs in diagnosing PCa, we conducted a systematic literature review and meta-analysis of available studies. PRISMA guidelines were used for the identification, screening, eligibility, and selection process. We included six studies that employed trained dogs and found that the pooled diagnostic sensitivity was 0.87 (95% CI 0.86–0.89; I2, 98.6%), the diagnostic specificity was 0.83 (95% CI 0.80–0.85; I2, 98.1%), and the area under the summary receiver operating characteristic curve (sROC) was 0.64 (standard error, 0.25). We also analyzed five studies that used an eNose to diagnose PCa and found that the pooled diagnostic sensitivity was 0.84 (95% CI, 0.80–0.88; I2, 57.1%), the diagnostic specificity was 0.88 (95% CI, 0.84–0.91; I2, 66%), and the area under the sROC was 0.93 (standard error, 0.03). These pooled results suggest that while highly trained dogs have the potentiality to diagnose PCa, the ability is primarily related to olfactory physiology and training methodology. The adoption of advanced analytical techniques, such as eNose, poses a significant challenge in the field of clinical practice due to their growing effectiveness. Nevertheless, the presence of limitations and the requirement for meticulous study design continue to present challenges when employing eNoses for the diagnosis of PCa.

In situ laser-assisted synthesis and patterning of graphene foam composites as a flexible gas sensing platform

Gas-sensitive semiconducting nanomaterials (e.g., metal oxides, graphene oxides, and transition metal dichalcogenides) and their heterojunctions hold great promise in chemiresistive gas sensors. However, they often require a separate synthesis method (e.g., hydrothermal, so-gel, and co-precipitation) and their integration on interdigitated electrodes (IDE) via casting is also associated with weak interfacial properties. This work demonstrates in situ laser-assisted synthesis and patterning of various sensing nanomaterials and their heterojunctions on laser-induced graphene (LIG) foam to form LIG composites as a flexible and stretchable gas sensing platform. The porous LIG line or pattern with nanomaterial precursors dispensed on top is scribed by laser to allow for in situ growth of corresponding nanomaterials. The versatility of the proposed method is highlighted through the creation of different types of gas-sensitive materials, including transition metal dichalcogenide (e.g., MoS2), metal oxide (e.g., CuO), noble metal-doped metal oxide (e.g., Ag/ZnO) and composite metal oxides (e.g., In2O3/Cr2O3). By eliminating the IDE and separate heaters, the LIG gas sensing platform with self-heating also decreases the device complexity. The limit of detection (LOD) of the LIG gas sensor with in situ synthesized MoS2, CuO, and Ag/ZnO to NO2, H2S, and trimethylamine (TMA) is 2.7, 9.8, and 5.6 ppb, respectively. Taken together with the high sensitivity, good selectivity, rapid response/recovery, and tunable operating temperature, the integrated LIG gas sensor array can identify multiple gas species in the environment or exhaled breath.

Diagnosis by Volatile Organic Compounds in Exhaled Breath in Exhaled Breath from Patients with Gastric and Colorectal Cancers

One in three cancer deaths worldwide are caused by gastric and colorectal cancer malignancies. Although the incidence and fatality rates differ significantly from country to country, the rates of these cancers in East Asian nations such as South Korea and Japan have been increasing each year. Above all, the biggest danger of this disease is how challenging it is to recognize in its early stages. Moreover, most patients with these cancers do not present with any disease symptoms before receiving a definitive diagnosis. Currently, volatile organic compounds (VOCs) are being used for the early prediction of several other diseases, and research has been carried out on these applications. Exhaled VOCs from patients possess remarkable potential as novel biomarkers, and their analysis could be transformative in the prevention and early diagnosis of colon and stomach cancers. VOCs have been spotlighted in recent studies due to their ease of use. Diagnosis on the basis of patient VOC analysis takes less time than methods using gas chromatography, and results in the literature demonstrate that it is possible to determine whether a patient has certain diseases by using organic compounds in their breath as indicators. This study describes how VOCs can be used to precisely detect cancers; as more data are accumulated, the accuracy of this method will increase, and it can be applied in more fields.

Induction mechanism of ferroptosis: A novel therapeutic target in lung disease

Ferroptosis is a newly discovered form of non-apoptotic regulatory cell death driven by iron-dependent lipid peroxidation. Ferroptosis significantly differs from other forms of cell death in terms of biochemistry, genetics, and morphology. Ferroptosis affects many metabolic processes in the body, resulting in disruption of homeostasis, and is related to many types of lung disease. Although current research on ferroptosis remains in the early stage, existing studies have confirmed that ferroptosis is regulated by a variety of genes, mainly involving changes in genes involved in iron homeostasis and lipid peroxidation metabolism. Furthermore, the mechanism of ferroptosis is complex. This review summarizes the confirmed mechanisms that can cause ferroptosis, including activation of glutathione peroxidase 4, synthesis of glutathione, accumulation of reactive oxygen species, and the influence of ferrous ions and p53 proteins. In recent years, the mechanism of ferroptosis in the occurrence and development of many diseases has been studied; the occurrence of ferroptosis will produce an inflammatory storm, and most of the inducing factors and pathological manifestations of lung diseases are also inflammatory reactions. Therefore, we believe that the association between ferroptosis and lung disease deserves further study. This article aims to help readers to better understand the mechanism of ferroptosis, provide new ideas and targets for the treatment of lung diseases, and point out the direction for the development of new targeted drugs for the clinical treatment of lung diseases.