Developing a prediction model to predict the incidence of metastasis in lung cancer patients using artificial intelligence techniques.
- 1Department of Pulmonary Medicine, Faculty of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
- 2Health Information Management and Medical Informatics Department, School of Allied Medical Sciences, Tehran University of Medical Sciences, Tehran, Iran. [email protected].
- 3Thoracic Research Center, Imam Khomeini Hospital complex, Tehran University of Medical Sciences, Tehran, Iran. [email protected].
Abstract
Background
If metastasis can be predicted, appropriate management of lung cancer will be possible to improve prognosis. Therefore, the main objective of this study is to develop a prediction model with appropriate performance for predicting metastasis based on patient data.Methods
The study analyzed patient records from Kaggle, performing exploratory data analysis and preprocessing techniques such as missing values handling, encoding categorical variables, and outliers' removal. Next, Data normalization and feature engineering were applied. Imbalanced data was addressed using Synthetic Minority Over-sampling Technique (SMOTE). Supervised classification models, including boosting algorithms were developed and tuned via Grid Search algorithms. Performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC, with stratified k-fold cross-validation ensuring robustness. All processes were conducted using Python language on Google Collab.Results
After data preprocessing, normalization, and selecting more related features, the dataset was stratified into training and testing subsets using an 80:20 split to ensure representative sampling classification models trained on training data. Class imbalance in the target variable was addressed using the SMOTE on the training set to balance classes without biasing the test set. Various machine learning models, including Logistic Regression, KNN, Decision Tree, Random Forest, and boosting algorithms (AdaBoost, XGBoost, LightGBM, Gradient Boosting), were developed and evaluated using metrics like accuracy, precision, recall, F1-score, and AUC. SMOTE significantly improved model performance, with XGBoost emerging as the top performer, achieving the highest AUC (82%) and F1-score (74.83%). ROC curve analysis confirmed XGBoost's superior predictive power for metastasis.Conclusions
The development of a high-performing prediction model for lung cancer metastasis to other organs has significant implications for clinical practice. Early and accurate prediction of metastasis can enable timely interventions, potentially improve patient outcomes and reduce healthcare costs.
Related
- Low-Dose CT in the Age of Artificial Intelligence: Techniques, Evidence, and Clinical Impact
- Utility of Artificial Intelligence in Detecting Lymph Node Metastasis in Oral Cancer: A Systematic Review
- Artificial Intelligence for developing a personalised functional profile and supporting upper-limb rehabilitation decision-making in children with unilateral Cerebral Palsy (AInCP): protocol for an observational study
- Artificial intelligence for vascular minimally invasive surgery: techniques, clinical integration, and challenges
- Developing and evaluating patient information leaflets for artificial intelligence use in healthcare
BibTeX
@article{gharabaghi2026developing,
title = {{Developing a prediction model to predict the incidence of metastasis in lung cancer patients using artificial intelligence techniques.}},
author = {Mehrnaz Asadi Gharabaghi and Marsa Gholamzadeh},
journal = {BMC pulmonary medicine},
year = {2026},
month = jul,
doi = {10.1186/s12890-026-04530-6},
url = {https://europepmc.org/article/MED/42907991},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Developing a prediction model to predict the incidence of metastasis in lung cancer patients using artificial intelligence techniques.
AU - Mehrnaz Asadi Gharabaghi
AU - Marsa Gholamzadeh
PY - 2026
DA - 2026/07/01
JO - BMC pulmonary medicine
DO - 10.1186/s12890-026-04530-6
UR - https://europepmc.org/article/MED/42907991
AB - Background If metastasis can be predicted, appropriate management of lung cancer will be possible to improve prognosis. Therefore, the main objective of this study is to develop a prediction model with appropriate performance for predicting metastasis based on patient data.Methods The study analyzed patient records from Kaggle, performing exploratory data analysis and preprocessing techniques such as missing values handling, encoding categorical variables, and outliers' removal. Next, Data normalization and feature engineering were applied. Imbalanced data was addressed using Synthetic Minority Over-sampling Technique (SMOTE). Supervised classification models, including boosting algorithms were developed and tuned via Grid Search algorithms. Performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC, with stratified k-fold cross-validation ensuring robustness. All processes were conducted using Python language on Google Collab.Results After data preprocessing, normalization, and selecting more related features, the dataset was stratified into training and testing subsets using an 80:20 split to ensure representative sampling classification models trained on training data. Class imbalance in the target variable was addressed using the SMOTE on the training set to balance classes without biasing the test set. Various machine learning models, including Logistic Regression, KNN, Decision Tree, Random Forest, and boosting algorithms (AdaBoost, XGBoost, LightGBM, Gradient Boosting), were developed and evaluated using metrics like accuracy, precision, recall, F1-score, and AUC. SMOTE significantly improved model performance, with XGBoost emerging as the top performer, achieving the highest AUC (82%) and F1-score (74.83%). ROC curve analysis confirmed XGBoost's superior predictive power for metastasis.Conclusions The development of a high-performing prediction model for lung cancer metastasis to other organs has significant implications for clinical practice. Early and accurate prediction of metastasis can enable timely interventions, potentially improve patient outcomes and reduce healthcare costs.
SN - 1471-2466
ER -