[Paper Review] NeuroSVM: A Graphical User Interface for Identification of Liver Patients

Kalyan Nagaraj, Amulyashree Sridhar|arXiv (Cornell University)|Feb 19, 2015

Face and Expression Recognition17 references11 citations

TL;DR

This paper proposes NeuroSVM, a hybrid machine learning model combining support vector machines (SVM) and feed-forward artificial neural networks (ANN) for accurate identification of liver disease patients. Using R platform, the method achieves 98.83% prediction accuracy, and a GUI package is developed for clinical usability, enhancing diagnostic precision in early-stage liver disease detection.

ABSTRACT

Diagnosis of liver infection at preliminary stage is important for better treatment. In todays scenario devices like sensors are used for detection of infections. Accurate classification techniques are required for automatic identification of disease samples. In this context, this study utilizes data mining approaches for classification of liver patients from healthy individuals. Four algorithms (Naive Bayes, Bagging, Random forest and SVM) were implemented for classification using R platform. Further to improve the accuracy of classification a hybrid NeuroSVM model was developed using SVM and feed-forward artificial neural network (ANN). The hybrid model was tested for its performance using statistical parameters like root mean square error (RMSE) and mean absolute percentage error (MAPE). The model resulted in a prediction accuracy of 98.83%. The results suggested that development of hybrid model improved the accuracy of prediction. To serve the medicinal community for prediction of liver disease among patients, a graphical user interface (GUI) has been developed using R. The GUI is deployed as a package in local repository of R platform for users to perform prediction.

Motivation & Objective

• To improve the accuracy of liver disease classification using machine learning techniques.
• To address the challenge of early diagnosis of liver infections through automated, data-driven classification.
• To develop a user-friendly graphical interface for medical practitioners to access predictive models without machine learning expertise.
• To evaluate and compare multiple classification algorithms (Naive Bayes, Bagging, Random Forest, SVM) on liver disease data.
• To design and deploy a hybrid NeuroSVM model that enhances prediction performance over individual models.

Proposed method

• Four traditional machine learning algorithms—Naive Bayes, Bagging, Random Forest, and SVM—were implemented in the R programming platform for baseline classification.
• A hybrid NeuroSVM model was developed by integrating SVM with a feed-forward artificial neural network (ANN) to improve classification accuracy.
• The hybrid model was trained and validated using statistical metrics including root mean square error (RMSE) and mean absolute percentage error (MAPE).
• The final model was evaluated for performance using standard classification accuracy on a liver disease dataset.
• A graphical user interface (GUI) was developed using R and deployed as a local R package for clinical accessibility and ease of use.
• The GUI enables non-technical users to input patient features and receive disease prediction results without requiring coding or model expertise.

Experimental results

Research questions

• RQ1 Can a hybrid machine learning model combining SVM and ANN improve classification accuracy for liver disease detection compared to individual models?
• RQ2 How do traditional machine learning algorithms (Naive Bayes, Bagging, Random Forest, SVM) perform on liver disease classification tasks?
• RQ3 To what extent does the integration of SVM and ANN enhance prediction performance in terms of RMSE and MAPE?
• RQ4 Can a GUI-based implementation of the NeuroSVM model support practical clinical use by non-expert users?
• RQ5 What is the optimal balance between model accuracy and usability in a medical diagnostic tool for liver disease?

Key findings

• The hybrid NeuroSVM model achieved a prediction accuracy of 98.83%, significantly outperforming individual models.
• The integration of SVM and feed-forward ANN reduced both RMSE and MAPE, indicating improved prediction stability and precision.
• Among the baseline models, SVM demonstrated superior performance compared to Naive Bayes, Bagging, and Random Forest on the dataset.
• The GUI package was successfully deployed in the R local repository, enabling seamless access for medical professionals.
• The study confirms that hybrid models can significantly enhance diagnostic accuracy in liver disease classification.
• The developed system provides a practical, accessible tool for early liver disease detection in clinical settings.