Adaptive Temporal Mixture of Experts for Predicting Stiffness Metrics from the Ocular Response Analyzer and Identifying Keratoconus: Stiffness Estimates from Ocular Response Analyzer.

PURPOSE

To develop a machine learning-based modeling approach for extracting elastic stiffness estimates from Ocular Response Analyzer (ORA) waveforms.

DESIGN

Prospective observational cohort study with model training and testing using development dataset and validation with independent dataset.

SUBJECTS

Participants were prospectively enrolled into six cohorts for the Development dataset, including control subjects (n=199), individuals diagnosed with keratoconus (n=54), diabetes mellitus with retinopathy (n=78) and without retinopathy (n=75), primary open-angle glaucoma (n=53), and ocular hypertension (n=44). Independent validation dataset comprised of two cohorts including healthy participants (n=145) and individuals diagnosed with keratoconus (n=44).

METHODS

Intraocular pressure and biomechanical data were collected using the ORA and Corvis ST devices for the Development dataset. An Adaptive Temporal Mixture of Experts (AT-MoE) model, incorporating Long Short-Term Memory (LSTM) networks with dynamic expert selection, was trained using ORA waveform parameters and raw signals to predict stiffness parameters (SP) from Corvis ST, including SP-A1 representing corneal stiffness, SP-HC representing scleral stiffness, and SSI representing deformation stiffness. Predicted ORA stiffness estimates were independently validated with a separate dataset of keratoconus and healthy eyes. The ORA waveform parameters, as well as applanation and pressure signals, were used as input to the machine learning models to classify keratoconus versus healthy eyes, which was evaluated with Area Under Receiver Operator Characteristic (AUROC) Curves.

MAIN OUTCOME MEASURES

The primary outcome measures were the predicted ORA stiffness estimates: Corneal Stiffness estimate (CSe), Scleral Stiffness estimate (SSe), and Deformation Stiffness estimate (DSe).

RESULTS

The AT-MoE model significantly outperformed the baseline LSTM in predicting ocular stiffness estimates using ORA input data with reduction of prediction error. The AT-MoE model also resulted in improvement of keratoconus detection performance in the independent validation dataset with AUROC = 0.9512, which is similar to performance of tomographic detection approaches.

CONCLUSIONS

The AT-MoE model provides a novel and accurate framework for deriving elastic stiffness estimates from ORA waveforms and offers improved diagnostic capability for keratoconus compared to traditional machine learning models. This method has the potential to expand the utility of the ORA device for biomechanical assessment in clinical settings.