Clinical Scorecard: AI in Practice

At a Glance

Key Highlights

• AI models, particularly convolutional neural networks, outperform traditional methods in predicting optimal RGP lens fits for keratoconus.
• Machine learning algorithms incorporating multiple ocular parameters enhance first-fit success rates in orthokeratology lens fitting.
• Large databases enable AI to identify trends that improve empirical fitting accuracy and clinical outcomes.

Guideline-Based Recommendations

Diagnosis

• Utilize corneal topography imaging as input data for AI-assisted fitting models.
• Incorporate multiple ocular parameters such as keratometry, spherical refraction, and corneal eccentricity for comprehensive assessment.

Management

• Apply AI-driven algorithms to select initial contact lens parameters to reduce trial lens numbers.
• Leverage empirical fitting supported by AI to increase first-fit success rates.

Monitoring & Follow-up

• Evaluate lens centration and fit outcomes using topography and clinical follow-up to refine AI model predictions.

Risks

• Ensure AI models are validated with large datasets to avoid inaccurate lens selection.
• Maintain clinical oversight to confirm AI recommendations align with individual patient needs.

Patient & Prescribing Data

Nearly 200 eyes with keratoconus and approximately 800 eyes fitted with orthokeratology lenses

AI-assisted fitting methods demonstrated higher accuracy in predicting optimal lens parameters and improved first-fit success compared to traditional fitting guides.

Clinical Best Practices

• Incorporate AI tools that analyze corneal topography images for lens fitting decisions.
• Use large-scale clinical data to train and validate AI algorithms for contact lens fitting.
• Combine AI predictions with clinical judgment to optimize lens selection and patient outcomes.

References

• Abadou et al, 2025 - AI versus conventional methods for RGP lens fitting in keratoconus
• Zhou et al, 2024 - AI-assisted fitting method enhances orthokeratology lens success