Home / Current Issue / Paper 1706213
Using Artificial Intelligence to Improve Insurance Claim Evaluation
Subject area: Science,Engineering and Technology · Area of research: Machine Learning
Abstract
Machine learning is used to improve vehicle insurance claim analysis in this study. The increased number of claims and efficiency requirements for claims processing have shown that manual methods cannot handle the load, resulting in delays, inaccuracies, and inefficiencies. The research uses Kaggle insurance claim data using OOAD and UML models to create a simple and reliable application architecture. Random Forest, known for its accuracy and diversity, produced a model with 99.5% precision compared to Decision Tree's 95.68%. Confusion matrix and ROC curve performance measurements showed machine learning algorithms' claim result prediction power. Results showed the intricacy of interactions between factors, including legal counsel and seatbelt usage, that affected claims results and scholarly literature on attorney and safety measures. Theoretical implications help build an overall evaluation approach to insurance claim analysis and allow incorporation of previously unimportant aspects impacting claim results. Practical implications include insurance claim processing transformation, which will systemize assessment processes, minimize error margins, and detect fraudulent claims. Expand data sources, incorporate live data, use advanced machine learning algorithms, and validate models in the real world. This ensures their efficacy and usability. Future research will address limitations like relying on historical data by integrating real-time data streams and using advanced predictive analytics methods like deep learning and NLP algorithms to analyse unstructured claim forms.
Keywords
Machine learning, insurance claim analysis, automotive insurance, predictive modelling, decision-making, fraud detection.
How to cite this paper
@article{1706213,
author = {Raphael Ibraimoh},
title = {Using Artificial Intelligence to Improve Insurance Claim Evaluation},
journal = {Iconic Research And Engineering Journals},
year = {2024},
volume = {8},
number = {2},
pages = {749-759},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1706213.pdf},
abstract = {Machine learning is used to improve vehicle insurance claim analysis in this study. The increased number of claims and efficiency requirements for claims processing have shown that manual methods cannot handle the load, resulting in delays, inaccuracies, and inefficiencies. The research uses Kaggle insurance claim data using OOAD and UML models to create a simple and reliable application architecture. Random Forest, known for its accuracy and diversity, produced a model with 99.5% precision compared to Decision Tree's 95.68%. Confusion matrix and ROC curve performance measurements showed machine learning algorithms' claim result prediction power. Results showed the intricacy of interactions between factors, including legal counsel and seatbelt usage, that affected claims results and scholarly literature on attorney and safety measures. Theoretical implications help build an overall evaluation approach to insurance claim analysis and allow incorporation of previously unimportant aspects impacting claim results. Practical implications include insurance claim processing transformation, which will systemize assessment processes, minimize error margins, and detect fraudulent claims. Expand data sources, incorporate live data, use advanced machine learning algorithms, and validate models in the real world. This ensures their efficacy and usability. Future research will address limitations like relying on historical data by integrating real-time data streams and using advanced predictive analytics methods like deep learning and NLP algorithms to analyse unstructured claim forms.},
keywords = {Machine learning, insurance claim analysis, automotive insurance, predictive modelling, decision-making, fraud detection.},
month = {August},
}