Artificial intelligence applications in health insurances: a scoping review

Maryam Ramezani-a, Ahad Bakhtiari, Mohammadreza Mobinizadeh, Rajabali Daroudi, Hamid R. Rabiee, Alireza Olyaeemanesh, Ali Akbar Fazaeli, Hakimeh Mostafavi, Maryam Ramezani-b, Saharnaz Sazgarnejad, Sanaz Bordbar, Amirhossein TakianCost 2025 · DOI

Abstract

Policies for AI applications in health insurance should prioritize the protection of personal health and medical data, address ethical concerns, and ensure robust data privacy and security. Additionally, these policies should promote the use of AI to enhance customer experiences, optimize risk selection, and generate revenue for both insurers and policyholders.

Introduction

The health insurance industry is currently navigating a period of significant turbulence, caught between escalating healthcare costs and an increasing imbalance between revenues and expenditures. While Artificial Intelligence (AI) is frequently presented as a "silver bullet" for these systemic issues, research in this domain has historically been siloed. Most studies focus narrowly on specific tasks—such as fraud detection or cost reduction—without considering the broader implications for the entire insurance value chain.

This scoping review addresses this fragmentation by consolidating 24 years of research (2000–2024) into a cohesive framework. The authors explore how AI can be integrated across various levels, from national policy-making to individual patient diagnostics, mapping a transformative landscape that shifts the industry from a reactive model to a proactive paradigm.

Methodology

To ensure a comprehensive and rigorous analysis, the researchers employed the Arksey and O’Malley scoping review framework. This five-stage process involved formulating the research question, identifying relevant studies through systematic database searches, and applying strict inclusion/exclusion criteria.

Search queries and results across major academic databases.

The initial search yielded 225 references from PubMed, Scopus, and Web of Science. After removing duplicates and screening based on title, abstract, and full-text eligibility, 60 high-quality studies were selected for the final analysis.

PRISMA flow diagram illustrating the study selection process.

Key Insights: The "Predict and Prevent" Paradigm

A central theme emerging from the review is the fundamental industry shift from a "detect and repair" model to a "predict and prevent" paradigm. By 2030, AI is expected to move insurers away from simply paying for "repairs" (claims) toward predicting health risks and preventing illness through personalized interventions.

Micro-segmentation and Personalization

Traditional insurance models rely on broad demographic categories like age and gender. AI enables "micro-segmentation," allowing insurers to analyze clusters of individual behavior. This leads to hyper-personalized policies and more accurate risk selection, benefiting both the insurer and the policyholder through tailored coverage and pricing.

Fraud, Waste, and Abuse (FWA)

AI significantly outperforms manual audits in identifying complex patterns of Fraud, Waste, and Abuse. The review highlights AI's ability to detect sophisticated schemes such as self-referral, double billing, and "upcoding"—patterns that often elude human reviewers.

Results and Impact

The review categorized AI applications into nine core dimensions, revealing significant quantitative and qualitative benefits:

  • Financial Performance: One study demonstrated that AI-driven mortality models could reduce claims by 9% among the healthiest applicants by optimizing the underwriting process.
  • Operational Efficiency: AI systems for analyzing claims have successfully reduced hospital "weight deductions" (payment penalties) and streamlined the appeals process for denied claims.
  • Clinical Integration: Advanced models, specifically Convolutional Neural Networks (CNNs), showed high performance in predicting chronic kidney disease (CKD) using insurance claim data, identifying risk factors earlier than traditional methods.
  • Public Health Policy: AI tools have been used to evaluate national health insurance schemes, helping policymakers identify geographical variations in maternal health and infant mortality to better target subsidies.

Ethical Governance and Future Directions

Despite the promise of AI, the review emphasizes that its success hinges on solving critical data privacy and ethical challenges. The authors advocate for a "Responsible AI" (RAI) mandate, arguing for human-centered systems that prioritize transparency.

The "Black Box" and Explainability

A major hurdle is the "black box" nature of many AI algorithms. Future work must focus on Explainable AI (XAI) to ensure that when an AI denies a claim or raises a premium, the reasoning is transparent, fair, and legally defensible.

Limitations

The authors acknowledge that the search was limited to three major databases and English-only publications, which may have excluded regional innovations. Furthermore, as a scoping review, the focus was on the applications of AI rather than the specific technical architectures of the underlying algorithms.

Conclusion

AI is poised to redefine the health insurance industry, moving it toward a more efficient, personalized, and preventive model. However, realizing this potential requires robust governance frameworks and international standards to protect personal medical data and ensure that AI applications remain ethical and transparent.

Summary updated

Abstract

Policies for AI applications in health insurance should prioritize the protection of personal health and medical data, address ethical concerns, and ensure robust data privacy and security. Additionally, these policies should promote the use of AI to enhance customer experiences, optimize risk selection, and generate revenue for both insurers and policyholders.
BibTeX
@inproceedings{ramezania2025artificial,
  author = {Maryam Ramezani-a and Ahad Bakhtiari and Mohammadreza Mobinizadeh and Rajabali Daroudi and Hamid R. Rabiee and Alireza Olyaeemanesh and Ali Akbar Fazaeli and Hakimeh Mostafavi and Maryam Ramezani-b and Saharnaz Sazgarnejad and Sanaz Bordbar and Amirhossein Takian},
  title = {{Artificial intelligence applications in health insurances: a scoping review}},
  booktitle = {Cost},
  year = {2025},
  doi = {10.1186/s12962-025-00640-w},
}
Figure