Development and Pilot Validation of ABHA-O-SHINE: An AI-Ready Oral Health Risk and Insurance Prediction Framework within the Ayushman Bharat Digital Ecosystem

This study presents the development and pilot validation of the ABHA-O-SHINE framework, a scalable tool that successfully integrates clinical and behavioral oral health data with insurance prioritization within the Ayushman Bharat Digital Mission ecosystem, establishing a foundation for future AI-driven automation.

The Gist

Imagine India's healthcare system as a massive, bustling train station. The Ayushman Bharat Digital Mission (ABDM) is the new, high-tech digital ticketing system designed to give every passenger a unique ID (called an ABHA) so they can travel smoothly, access their medical history, and get help when they need it.

However, there's a problem: while the train station is getting a high-tech upgrade, the dental department is still running on an old, manual clipboard. No one is really checking how "at risk" a passenger is for a dental disaster before they even get on the train, and there's no clear plan for who gets priority insurance coverage for dental care.

Enter ABHA-O-SHINE.

Think of ABHA-O-SHINE not just as a study, but as a smart, futuristic "Dental Weather Forecast" app that the authors built and tested. Here is how it works, broken down into simple concepts:

1. The Problem: The "Silent Storm"

Oral health issues (like cavities and gum disease) are like a slow-moving storm. They affect billions of people, but because they don't always hurt immediately, people ignore them until the roof caves in (requiring expensive, emergency surgery). In India, this storm is huge, but the digital system doesn't have a radar to predict it yet.

2. The Solution: The "Dental Risk Scorecard"

The researchers created a 14-point scorecard (the ABHA-O-SHINE framework) to act as a radar. Instead of just looking at teeth, they looked at the whole picture:

• The Habits: Do you smoke? (Like driving with a flat tire). Do you eat too much sugar? (Like pouring gasoline on a fire).
• The Damage: How many teeth are decayed or missing? (The actual crash damage).
• The Gums: How deep are the pockets in your gums? (The foundation of the house).

They gave everyone a score from 0 to 14.

• 0–4 (Low Risk): Your car is running fine.
• 5–9 (Moderate Risk): You have a check-engine light on.
• 10–14 (High Risk): You are about to break down on the highway.

3. The Pilot Test: The "Dry Run"

The authors tested this scorecard on 126 people at a dental college in Nagpur. It was like a "beta test" for a new app.

• What they found: Most people (over 75%) were in the "Moderate Risk" zone, and nearly 20% were in the "High Risk" zone.
• The Smoking Gun: They found a direct link between smoking and high risk. If you didn't smoke, you were almost never in the "High Risk" category. If you smoked multiple types of tobacco, you were almost guaranteed to be in the danger zone.
• The "Fake" Hygiene: Many people said they brushed twice a day, but when the doctors looked, their mouths were actually quite dirty. It's like saying you exercise daily but never actually leaving the couch. The scorecard caught this lie.

4. The Big Idea: Connecting to Insurance

Here is the most creative part. The authors realized that if we know who is "High Risk," we can tell the Insurance Company (the bank paying for the repairs) who needs help first.

• Low Risk = Low Priority (Wait your turn).
• High Risk = High Priority (Jump the line, get treated now).

This turns dental care from a "first-come, first-served" chaos into a smart triage system, ensuring money goes to the people who need it most before their teeth fall out completely.

5. The Future: The "AI Brain"

Right now, the doctors calculated these scores by hand (like using a calculator). But the ABHA-O-SHINE framework is designed to be AI-ready.

Imagine upgrading from a calculator to a super-intelligent robot.

• In the future, a dentist could take a photo of your teeth.
• The AI (Artificial Intelligence) would instantly analyze the photo, check your digital health ID (ABHA), look at your history, and say: "This patient is 90% likely to lose teeth in 2 years. Assign them 'High Priority' insurance coverage immediately."

Why Does This Matter?

• It's Preventive: It stops the storm before it destroys the house.
• It's Fair: It uses data to decide who gets help, not who has the loudest voice.
• It's Digital: It finally puts dental health on the same high-tech map as the rest of India's healthcare.

In a nutshell: The authors built a digital radar to spot dental disasters early, tested it on real people, and proved it works. Now, they are handing the blueprints to the government and AI developers to build a system where your dental health is automatically monitored, and your insurance covers you before you have a dental emergency.

Technical Summary

1. Problem Statement

Oral health in India remains a significant public health burden, affecting approximately 3.5 billion people globally, yet it is inadequately integrated into general healthcare systems. Specifically, within the Ayushman Bharat Digital Mission (ABDM), there is a critical gap in:

• Structured Risk Assessment: Lack of standardized, scalable models to assess oral health risk within the digital ecosystem.
• Data Integration: Failure to link clinical oral health data with digital health records (Ayushman Bharat Health Account - ABHA) for actionable decision-making.
• Preventive vs. Curative Focus: Existing systems prioritize treatment over preventive, risk-based approaches, leading to higher long-term costs.
• AI Readiness: Current frameworks are not designed to support future Artificial Intelligence (AI) and Machine Learning (ML) automation for insurance prioritization and resource allocation.

2. Methodology

The study employed a cross-sectional pilot design to develop and validate the ABHA-O-SHINE (Ayushman Bharat Health Account – Oral Smart Health Insurance Neural Engine) framework.

• Study Setting & Population: Conducted at the Outpatient Department (OPD) of Swargiya Dadasaheb Kalmegh Smruti Dental College, Nagpur, India.
• Sample Size: 126 participants (66 males, 60 females), aged 18+, selected via random sampling.
• Data Collection:
  • Questionnaire: Demographic details, behavioral factors (tobacco use, sugar intake), and ABHA awareness.
  • Clinical Examination: Performed using the WHO Oral Health Assessment Form (2013), recording DMFT (Decayed, Missing, Filled Teeth) and CPI (Community Periodontal Index).
• Framework Development (ABHA-O-SHINE Scoring):
  A composite risk score (0–14) was developed by integrating:
  • Behavioral Parameters: Tobacco usage (0–2), Sugar intake (0–2).
  • Clinical Parameters: Intervention need (0–4), DMFT (0–3), Periodontal status (0–2), Oral lesions (0–1).
  • Risk Stratification:
    • Low Risk: 0–4
    • Moderate Risk: 5–9
    • High Risk: 10–14
  • Insurance Priority Mapping: Low/Moderate/High risk groups were mapped to corresponding insurance priority levels.
• Statistical Analysis:
  • Descriptive statistics (Mean, SD, Frequencies).
  • Inferential tests: Chi-square (association), Spearman's rank correlation (correlation), and Binary Logistic Regression (predictors).
  • Note: Logistic regression faced convergence issues due to "quasi-complete separation" (e.g., zero high-risk cases among non-tobacco users), highlighting the strength of the association but limiting model stability in this pilot sample.

3. Key Results

• Risk Distribution: The majority of the population fell into Moderate Risk (75.4%) and High Risk (18.3%), with only 6.3% in the Low Risk category. The mean risk score was 7.52 ± 2.12.
• Behavioral Correlates:
  • Tobacco Use: Showed a highly significant association with risk categories ($\chi^2 = 34.47, p < 0.001$). A clear dose-response relationship was observed: 0% of non-users were high-risk, whereas 41.7% (20/48) of multiple-form tobacco users were high-risk.
  • Oral Hygiene Discrepancy: While 73.8% of participants reported brushing twice daily, clinical examination revealed 38.1% had poor oral hygiene, indicating a gap between self-reporting and clinical reality.
• Clinical Correlates:
  • Periodontal Status: Demonstrated the strongest correlation with the total risk score (Spearman's $\rho$ = 0.661, p < 0.001).
  • Caries Experience (DMFT): Showed a moderate positive correlation with risk score ($\rho$ = 0.347, p < 0.01).
• Treatment Needs: 77% of participants required active intervention (urgent care, scaling, or referral), with only 4.7% requiring no treatment.
• Digital Health Gap: While 50% were aware of ABHA, only 26.9% possessed an account, and a mere 7.3% actively used it at healthcare facilities.
• AI Readiness: The study confirmed that the structured dataset is suitable for future AI/ML integration, despite the current pilot using manual scoring.

4. Key Contributions

1. Novel Framework: Introduction of ABHA-O-SHINE, the first framework to integrate oral health risk assessment with insurance prioritization specifically within the Indian ABHA digital ecosystem.
2. Multidimensional Scoring: Development of a composite index (0–14) that combines behavioral, clinical, and digital health parameters, moving beyond traditional indices that focus solely on clinical metrics.
3. AI-Ready Architecture: The framework is explicitly designed to be "AI-Ready." It establishes a structured data foundation (structured risk scores, categorical variables) necessary for training future Machine Learning models for automated risk prediction and image-based analysis.
4. Policy Relevance: Provides a mechanism for dynamic insurance prioritization, allowing insurers and policymakers to allocate resources based on quantified risk rather than reactive treatment.
5. Identification of Implementation Gaps: Highlights the critical disconnect between ABHA awareness and actual utilization, providing data to guide future digital health policy interventions.

5. Significance and Future Implications

• Scalability: The framework is designed to be scalable from a pilot study to a national level, enabling real-time risk assessment across the ABDM ecosystem.
• Shift to Preventive Care: By linking risk scores to insurance priority, the model incentivizes preventive care and early intervention, potentially reducing the long-term burden of oral diseases.
• Foundation for AI: The study serves as a proof-of-concept for transitioning from manual scoring to automated, AI-driven predictive modeling. Future iterations could utilize intraoral images and longitudinal ABHA data to refine risk predictions using deep learning.
• Resource Optimization: The ability to stratify populations into Low/Medium/High priority groups allows for optimized resource allocation in public health planning and insurance underwriting.

Conclusion:
The ABHA-O-SHINE framework successfully demonstrates the feasibility of integrating oral health risk assessment into India's digital health infrastructure. While the pilot study faced statistical limitations regarding regression modeling due to sample size, the strong associations found between behavioral factors (tobacco) and clinical outcomes validate the framework's logic. It represents a significant step toward a data-driven, AI-enabled, and preventive-oriented oral healthcare system within the Ayushman Bharat Digital Mission.