Automated Analysis

This service uses machine learning (ML) to analyse nocturnal pulse oximetry data in children. Computed statistical, non-linear, spectral, and oximetric features are used to estimate the presence or absence, and severity of sleep-disordered breathing.

Sleep staging of pulse oximetry data is performed using binary classification. Thirty-second segments (epochs) of the recording are classified as Sleep or Wake, and these segments are filtered before analysis for sleep-disordered breathing. Sleep staging enables the computation of sleep statistics traditionally obtained from overnight polysomnography, such as sleep duration, latency, efficiency, and WASO. While this step may be disabled, it is recommended for data collected in uncontrolled environments (e.g., in the home).

Analysis of recording uses a regressor to produce a point estimate of the apnea-hypopnea index (AHI) with accompanying uncertainty bounds, or a classifier to predict whether the AHI is ≥5. Configurable cluster analysis enables characterisation of endotypes, assignment of endotypes to recordings, and identification of genuine anomalies or outliers.

Peer-reviewed article containing out-of-sample performance data coming soon.

PaedOx API

An API for inference requests, feature generation, and other storage and management operations is available for use by third-party applications. Contact us if you require additional functionality not currently supported, or a separate instance capable of serving large numbers of requests. Reference client code is available here.

Online Retraining

This service offers a streamlined pipeline for training and validating models using a bank of pre-computed features and datasets. Newly trained models are immediately available for use after saving, enabling rapid prototyping and rollout of updated models. All trained models are automatically calibrated, and provide estimates of uncertainty alongside predictions.

Model training uses a set of custom solvers. Gaussian Processes are solved exactly. Linear models are trained using coordinate descent. Logistic regression is trained using Iteratively Reweighted Least Squares with coordinate descent for the inner solve. Support Vector Machine solvers include Sequential Minimal Optimization and the Iterative Single Data Algorithm. Solver performance is comparable to commonly used machine learning libraries. Solver testing and validation data are available here.