This study explores the potential of using sweat samples for the screening of breast cancer. The researchers employed 2-Dimensional Gas Chromatography-Mass Spectrometry to analyze the chemical compounds present in the sweat samples. Statistical analysis techniques were used to identify volatile organic compounds (VOCs) that could serve as potential cancer biomarkers. The study found promising results, as the analysis of sweat volatomics profiles showed distinct patterns in breast cancer patients. The researchers also investigated confounding and instrumental parameters that could affect the accuracy of the analysis. Overall, this preliminary study suggests that sweat analysis using 2-Dimensional Gas Chromatography-Mass Spectrometry has the potential to be a non-invasive and effective method for breast cancer screening.
| Signal | Change | 10y horizon | Driving force |
|---|---|---|---|
| Screening of Breast Cancer from Sweat Samples Analyzed by 2-Dimensional Gas Chromatography-Mass Spectrometry | From traditional breast cancer screening methods to sweat analysis | Sweat analysis becomes a common method for breast cancer screening | Advancements in technology and the need for non-invasive and early detection methods |
| Application of LOG chemical standardization | Standardization of chemical compounds in sweat samples | Improved accuracy and reliability of sweat analysis | Standardization ensures consistency and comparability of results |
| Principal component analysis: visualization of clusters of samples based on their similarity | Visualization of sample clusters based on similarity | Enhanced understanding of the patterns and relationships between sweat samples | Identifying distinct patterns can aid in the identification and classification of breast cancer |
| Split the data into two sets: a learning/validation set and a test set and rebalance the classes/labels on the learning set employing SMOTE | Use of machine learning techniques and rebalancing of data for analysis | Improved accuracy and generalization of the models | Balancing classes reduces bias and improves the performance of the models |
| Train models with TPOT | Use of automated machine learning for model training | Faster and more efficient development of accurate models | Automated machine learning reduces the need for manual model development and optimization |
| Evaluation of the performance of the selected models by computing their F1-weighted score on the test set | Assessment of model performance using F1-weighted score | Objective measure of model accuracy and effectiveness | F1-weighted score provides a comprehensive evaluation of model performance |
| Interpret the model using SHAP | Interpretation of model predictions using Shapley values | Increased transparency and understanding of the factors influencing model predictions | Shapley values help identify the contribution of each feature to the model’s output |
| Determine the relevance of VOCs using the probe variable method | Identification of relevant volatile organic compounds (VOCs) | Improved understanding of the specific VOCs associated with breast cancer | Identifying relevant VOCs can aid in the development of targeted screening and diagnostic methods |
| Pre-vs. Post-Surgery Status on Breast VOCs (GC Column 2) | Comparison of VOCs before and after breast surgery | Insight into the changes in VOC profiles due to surgery | Understanding the impact of surgery on VOC profiles can help refine screening methods |
| Pre-vs. Post-Surgery Status on Hand VOCs (GC Column 2) | Comparison of VOCs before and after hand surgery | Insight into the changes in VOC profiles due to surgery | Understanding the impact of surgery on VOC profiles can help refine screening methods |