Related Experiment Video
Updated: Jun 3, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.7K
Integrative Stacking Machine Learning Model for Small Cell Lung Cancer Prediction Using Metabolomics Profiling
Md Shaheenur Islam Sumon1, Marwan Malluhi2, Noushin Anan1
1Department of Electrical Engineering, Qatar University, Doha 2713, Qatar.
Cancers
|January 8, 2025
Summary
A new machine learning model accurately distinguishes small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) using metabolomics data. This approach offers a promising non-invasive method for early lung cancer detection.
Area of Science:
- Oncology
- Computational Biology
- Biochemistry
Background:
- Small cell lung cancer (SCLC) is highly aggressive with poor survival rates.
- Current diagnostic methods for SCLC and NSCLC are invasive and limited.
- Early detection is crucial but challenging due to current diagnostic limitations.
Purpose of the Study:
- To develop a novel machine learning approach for classifying SCLC and NSCLC.
- To utilize metabolomics data for non-invasive lung cancer subtype detection.
- To compare the performance of a stacking ensemble model against traditional methods.
Main Methods:
- A stacking-based ensemble machine learning model was developed.
- Metabolomics data from 191 SCLC, 173 NSCLC cases, and 97 healthy controls were analyzed.
- Feature selection identified significant metabolites, with positive ions being more relevant.
Main Results:
- The multi-class model achieved 85.03% accuracy and 92.47 AUC (SVM classifier).
- The binary classification (SCLC vs. NSCLC) model reached 88.19% accuracy and 92.65 AUC (ExtraTreesClassifier).
- SHAP analysis identified benzoic acid, DL-lactate, and L-arginine as key predictive metabolites.
Conclusions:
- The stacking ensemble effectively enhances predictive performance by combining multiple classifiers.
- The model demonstrates potential for non-invasive early detection of SCLC and NSCLC subtypes.
- This approach offers a viable alternative to conventional invasive biopsy techniques.
