Efficient clinical data analysis for prediction of coal workers' pneumoconiosis using machine learning algorithms

Accurate diagnosis of coal workers' pneumoconiosis (CWP) remains a challenge in occupational health. Traditional methods rely on radiographic imaging and clinical symptoms, which may not detect early-stage disease effectively. Prior research has shown that pulmonary function indicators and blood oxygen parameters can reflect respiratory impairment. However, no prior work had resolved how to integrate these indicators with machine learning for early detection. This gap motivated the development of a predictive model that could identify CWP at an earlier stage than conventional methods. Existing studies have used radiological data and dust exposure history, but they lack integration with machine learning techniques. The need for a more efficient and accurate clinical prediction system for CWP is clear. No prior work had combined feature selection approaches with machine learning algorithms to improve CWP diagnosis. This paper's contribution lies in its novel approach to integrating these methods for early-stage detection.

The aim of this study is to develop a clinical prediction system for coal workers' pneumoconiosis (CWP) using machine learning algorithms. The specific problem is the lack of an efficient and accurate early detection system for CWP. The motivation is to improve clinical diagnosis by identifying key indicators that can predict CWP at an early stage. The study focuses on integrating machine learning with feature selection methods to enhance diagnostic accuracy. The researchers propose that combining these methods can improve diagnostic outcomes. The study also seeks to determine which machine learning algorithm performs best for CWP prediction. The goal is to create a model that can be used in clinical settings for early diagnosis. The researchers hope to demonstrate that machine learning can provide more reliable predictions than traditional methods.

The study used an embedded method for feature selection, applying three different approaches to analyze clinical data. The researchers selected patients with CWP and dust-exposed workers as the study population. They collected data from August 2021 to December 2021 to ensure recent and relevant information. The team applied three feature selection techniques to identify the most predictive variables. These methods included statistical analysis and machine learning-based selection. The researchers then tested various machine learning algorithms as the model backbone. They combined each algorithm with the three feature selection methods to evaluate performance. The study compared the results of different models to determine the most effective one for CWP prediction.

The study found that AaDO₂ and certain pulmonary function indicators were significant predictors of early-stage CWP. The support vector machine (SVM) algorithm achieved the highest predictive accuracy. The area under the curve (AUC) values for SVM using three feature selection methods were 97.78%, 93.7%, and 95.56%. These results suggest that SVM outperformed other machine learning models in predicting CWP. The researchers observed that the combination of SVM and feature selection methods improved diagnostic accuracy. The highest AUC value of 97.78% indicates strong predictive power for early detection. The study confirmed that the SVM model is the most effective for CWP prediction. These findings support the use of machine learning in clinical diagnosis for CWP.

The study concludes that the support vector machine (SVM) algorithm is the most effective model for predicting coal workers' pneumoconiosis (CWP). The researchers found that combining SVM with feature selection methods improved diagnostic accuracy. AaDO₂ and pulmonary function indicators were identified as key predictors for early-stage CWP. The study supports the use of machine learning in clinical settings for CWP diagnosis. The authors propose that the SVM-based model can be implemented for clinical use. The results suggest that machine learning can enhance early detection of CWP. The study does not claim that other models are ineffective, but that SVM performed best in this context. These findings may help improve clinical diagnosis and early intervention for CWP.