Islam, Nishat Margia and Yasin, Mohammad and Rahman, Rashedur and Rahman, Mahzabin Binte (2026) Explainable AI-Driven Business Intelligence for Detecting Fraud in Home Healthcare Claims. American Journal of Technology Advancement, 3 (7). ISSN 2997-9382

Text AJTA_4281_Nishat+Margia+Islam_Explainable+AI-Driven.pdf Download (1MB)

Abstract

Healthcare insurance fraud remains a major issue for insurers, healthcare providers, and insurance regulators, causing additional costs, lack of resource efficiency, and eroding trust in the healthcare system. Typical fraud detection methods involve manual audits and rule-based systems that are difficult to detect new and complex fraud trends in large volumes of health care claims. This study presents an Explainable Artificial Intelligence (XAI) based Business Intelligence (BI) system for fraudulent Healthcare Insurance Claims (HIC) detection with increased transparency and decision support. The data set for the research is a synthetic healthcare fraud dataset containing 10,000 insurance claim records that includes patient information, provider information, diagnosis and procedure codes, financial claim attributes, temporal variables, and a binary fraud indicator. To enhance data quality and model performance, data preprocessing techniques such as handling missing values, categorical encoding, feature engineering, and treating class imbalance are employed. Various machine learning models such as Logistic Regression, Decision Tree, Random Forest, Support Vector Machine and Extreme Gradient Boosting (XGBoost) are tested to find the best model for fraud detection. SHAP (Shapley Additive Explanations) is used to explain prediction results and to identify the most important factors that lead to fraudulent claims, to improve the interpretability of the model and to facilitate decision-making in practice. Business Intelligence dashboards are created to visualize fraud trends, provider risk profile, claim distributions and regional fraud patterns to help healthcare administrators and insurance investigators make informed decisions. The accuracy, precision, recall, F1 score, ROC-AUC, Precision–Recall Curve, and Confusion Matrix metrics are used to evaluate the model's performance. The suggested framework combines predictive analytics, Explainable AI, and Business Intelligence, offering a precise, transparent, and scalable approach to healthcare fraud detection. The results will be used to validate the effectiveness of explainable machine learning integrated with interactive business intelligence in improving the identification of frauds, gaining more stakeholder trust, and providing evidence-based information for better healthcare insurance fraud management and operational efficiency.

Item Type:	Article
Uncontrolled Keywords:	Explainable Artificial Intelligence (XAI), Business Intelligence, Healthcare Insurance Fraud Detection, Machine Learning, SHAP (Shapley Additive Explanations) and Predictive Analytics
Subjects:	H Social Sciences
Depositing User:	admin eprints
Date Deposited:	11 Aug 2026 23:37
Last Modified:	11 Aug 2026 23:37
URI:	http://eprints.umsida.ac.id/id/eprint/16917

Actions (login required)

View Item