Machine Learning IOL Power Prediction from Preoperative Biometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional calculators for determining intraocular lens (IOL) power require interoperative data, increasing surgical costs and time, and are often inaccurate for individual patients due to population-based data sets.
Innovation Solution
A machine learning-based system that utilizes pre-operative and intra-operative data, including independent measurements to predict post-operative lens settlement, using a determination engine to select a most predictive subset of features for accurate IOL power prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional calculators are used to determine IOL power, then lens position estimation can be provided, but interoperative data is required which increases surgical costs and time
Solution Approach 1:
The system performs preliminary calculations using pre-operative biometric data to predict post-operative refraction before surgery begins. By pre-processing the data and generating predictions in advance, the system eliminates the need for time-consuming interoperative measurements and calculations during the actual surgical procedure.
Solution Approach 2:
The invention extracts and utilizes specific pre-operative biometric parameters (axial length, corneal curvature, anterior chamber depth) to predict lens settlement without requiring interoperative data. By selecting only the essential pre-operative features needed for accurate prediction, the system removes the dependency on additional intraoperative measurements.
2Productivity
If conventional calculators use population-based data sets, then calculations can be performed, but accuracy for individual patients is reduced
Solution Approach 1:
The system transitions from population-based average predictions to individualized predictions by analyzing each patient's specific pre-operative biometric characteristics. The machine learning model is trained on individual patient data and applies local, patient-specific features (such as individual corneal curvature and axial length measurements) to generate personalized predictions, thereby improving accuracy for each unique patient.
Solution Approach 2:
The invention changes the approach from using fixed population-based parameters to dynamically adjusting predictions based on individual patient biometric parameters. The system uses machine learning to identify and weight the most relevant individual patient parameters, allowing the prediction model to adapt to each patient's specific anatomical characteristics rather than applying generic population averages.
3Measurement precision
If conventional calculators perform overt calculations to determine lens position, then recommendations can be provided, but costs increase due to required interoperative data acquisition
Solution Approach 1:
The system extracts and utilizes only the essential pre-operative biometric data needed for accurate predictions, eliminating the need for additional interoperative data acquisition. By identifying and using only the critical parameters (axial length, corneal curvature, anterior chamber depth) available before surgery, the system reduces resource consumption while maintaining prediction accuracy.
Solution Approach 2:
The invention creates a virtual model of the patient's eye using pre-operative biometric data and uses this digital copy to perform predictions and simulations. This virtual copying allows the system to calculate lens settlement and provide recommendations without requiring physical interoperative measurements or additional surgical resources.
Data Source
AI summary
Described are implementations of systems and methods for an improved machine learning-based system that incorporates pre-operative and intraoperative measurements captured during surgery, as well as additional patient-specific data, to provide an individualized, highly accurate post-operative manifest refraction prediction. According to some embodiments, a determination engine generates a predictive feature set of one or more predictors associated with diagnostic measurements of one or more eyes and performs a recursive selection operation using one or more combinations within the predictive feature set and one or more models to produce a most predictive subset, the most predictive subset having a highest prediction accuracy among other predictive subsets for post-operative manifest refraction. The determination engine generates a determination model by refining and retraining the one or more models of the recursive selection operation utilizing the most predictive subset.