Breast cancer detection using a realtime breath analyzer: A pilot study.
Bibliographic record
Abstract
e13040 Background: Despite advances in mammography, limitations persist related to accuracy among challenging cases (e.g. dense breast tissue) and screening adherence. Volatile organic compounds (VOCs) associated with breast cancer have been identified in the exhaled breath, however, analytical tools such as gas chromatography-mass spectrometry for VOC analysis, are neither generalizable nor scalable. A breathomics device, called DiagNoze, that leverages a novel digital olfaction platform to fingerprint complex mixtures of VOCs, may offer a portable and non-invasive option for breast cancer detection. Methods: Patients with suspicious findings on a breast image or examination, presenting to the McGill University Hospital Centre (MUHC) Breast Center for diagnostic testing, were recruited into the study. Patients with a history of asthma, COPD, diabetes mellitus, cigarette smoking, or those with concurrent cancer or who were actively receiving chemotherapy, were excluded from the study. Participants were excluded if they consumed alcohol or recreational drugs within 8 hours of recruitment or food or liquids, other than water, within one hour. The DiagNoze device captured up to 5 alveolar breath samples per study participant. Digitized breath fingerprints represented by a 32 dimensional (D) time series were collected for each breath sample. Samples were labelled as either positive or negative for breast cancer based on biopsy results. A t-distributed stochastic neighbor embedding (T-SNE) dimensional reduction method was applied to convert the 32D datasets into 2D latent space plots, and a fitted model was applied to determine data clustering accuracy. The fitted model performance was evaluated for all patients, and a patient subgroup with high breast density. Results: A total of 182 patients were recruited, with 156 meeting study inclusion criteria, with biopsy results and with at least one breath sample meeting data curation criteria (56 positive, 100 negative, average number of breath samples of 3.4). Of those, 125 (41 positive, 84 negative) had highly dense breast parenchyma (ACR C or D). The positive cases had a cancer stage distribution, from 0 to 3 of: 7, 25, 20, and 1, with 3 not reported. The data shows clear separation between positive cases and controls using a T-SNE clustering method, with clustering performance shown. Conclusions: This study demonstrates that classification of breast cancer status from alveolar breath samples is possible using the DiagNoze device, independent of breast density. DiagNoze has the potential to diagnose the presence of breast cancer and could be used for diagnosis and for followup of patients post-treatment. Model performance. Population Sensitivity Specificity PPV NPV All Patients 84% 89% 81% 91% Patients with dense breast parenchyma 78% 90% 79% 89%
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".