A Data-Driven Intelligent Methodology for Developing Explainable Diagnostic Model for Febrile Diseases
Preprint
- 17 January 2025
- preprint
- Published by MDPI AG in Preprints.org
Abstract
Febrile diseases such as typhoid fever, malaria, tuberculosis, and HIV/AIDS present diagnostic challenges in Low- and Middle-Income Countries (LMICs). Misdiagnosis leads to delayed treatment, higher healthcare costs, and increased mortality. This study proposes a data-driven diagnostic framework leveraging explainable artificial intelligence (XAI) and machine learning (ML) to enhance diagnostic accuracy and usability in resource-limited settings. To develop and evaluate a diagnostic system combining ML algorithms and XAI techniques for accurate and interpretable detection of febrile diseases, focusing on usability in clinical settings. A dataset of 3,914 patient records from secondary and tertiary hospitals was used to train and validate predictive models. Random Forest, Extreme Gradient Boost, and Multi-Layer Perceptron algorithms were employed, with hyperparameter tuning for optimization. XAI frameworks, including Large Language Models and Local Interpretable Model-agnostic Explanations, were integrated to enhance interpretability. A mobile-based platform was developed for real-time diagnostic use. The ML models demonstrated high diagnostic accuracy across multiple febrile diseases. Integrating XAI frameworks provided intuitive visualizations and textual explanations, enabling clinicians to understand the reasoning behind predictions better. The mobile-based implementation facilitated real-time decision-making, particularly in resource-constrained environments. The study highlights the potential of combining ML algorithms with explainable AI to improve diagnostic accuracy and usability in LMICs. While the system demonstrated effectiveness, limitations such as excluding pediatric data and certain febrile diseases were noted. Future research will expand the dataset, include additional diseases, and optimize system deployment to enhance adoption. This work establishes a foundation for transparent, AI-driven diagnostic tools in resource-poor settings.Keywords
All Related Versions
- Version 1, 2025-01-17, Preprints
- Published version: Algorithms, 18 (4), 190.
This publication has 24 references indexed in Scilit:
- Decoding AI Complexity: SHAP Textual Explanations via LLM for Improved Model Transparency2024
- AI-Enabled Data-Driven Approaches for Personalized Medicine and Healthcare Analytics2024
- Unlocking Transparency in Credit Scoring: Leveraging XGBoost with XAI for Informed Business Decision-Making2024
- Possibilities and Pitfalls of Generative Pre-Trained Transformers in Healthcare2023
- Medical Text Prediction and Suggestion Using Generative Pretrained Transformer Models with Dental Medical NotesMethods of Information in Medicine, 2022
- A novel approach to explain the black-box nature of machine learning in compressive strength predictions of concrete using Shapley additive explanations (SHAP)Case Studies in Construction Materials, 2022
- The three ghosts of medical AI: Can the black-box present deliver?Artificial Intelligence in Medicine, 2022
- An Overview of the Algorithmic Diagnostics Methodology: A Big Data Approach2018
- Data-Driven Diagnosis of Cervical Cancer With Support Vector Machine-Based ApproachesIEEE Access, 2017
- Data Driven Analytics for Personalized Healthcare2016