Revolutionizing Deep Vein Thrombosis (DVT) Management: Machine Learning Unveils Precision in Early Detection
K. Malathi, E. D. Kanmani Ruby, Kapilya Gangadharan, Ram Kumar M
Abstract
The primary objective of this study was to create and validate a machine learning model that can detect Deep Vein Thrombosis (DVT) at a stage. Timely identification of DVT is crucial, for improving patient outcomes. To achieve this we utilized a dataset consisting of medical records, clinical notes and imaging studies. We went through extensive data preprocessing, feature engineering and distinguishing between cases that were positive for DVT and those that were negative. Various algorithms such as regression, random forests, support vector machines and neural networks were evaluated to find the effective one. The random forest model emerged as the performer with an 85% sensitivity and 90% specificity in identifying stages of DVT. These findings emphasize the impact of machine learning in reducing risks associated with DVT and enhancing patient care. One key aspect of our approach was conducting an assessment of the models accuracy, precision, recall, specificity, F1 score and AUC ROC. This provided insights into features and their influence on DVT detection. The research demonstrates how the random forest algorithm can significantly contribute to diagnosing DVT and suggests that integrating machine learning models, into workflows could greatly benefit DVT management while complying with regulations and continuously refining the model.