Objective:

To evaluate the effectiveness of a camera-based screening approach using facial images for identifying hypertension, specifically focusing on the zygomatic and cheek regions.

Key Findings:

• Deep learning analysis achieved 83% accuracy in identifying hypertension.
• Zygomatic and cheek regions alone achieved 82% accuracy, comparable to full-face models.
• Traditional statistical models and contact-based methods had lower accuracy (73% to 83%).
• The segmentation model achieved a mean Intersection over Union of 98%.
• The proposed framework achieved an F1-score of 0.75 and AUC of 84%.

Interpretation:

The study indicates that facial image analysis can serve as a scalable, non-invasive initial screening tool for hypertension, addressing barriers like low screening adherence, asymptomatic disease onset, and measurement biases.

Limitations:

• Sample size is relatively small, potentially limiting generalizability.
• Future studies should include larger, multicenter, and diverse cohorts to ensure applicability across different ethnicities, genders, and age groups.

Conclusion:

The facial AI approach is intended as a complementary tool for hypertension screening, not a replacement for traditional methods.