Canvas Dx

Canvas Dx is a software-based medical device designed to aid in the diagnosis of Autism Spectrum Disorder in young children by classifying neurodevelopmental and neurobehavioral data. It leverages a machine learning model, continuously updated via a Predetermined Change Control Plan, to provide diagnostic classifications.

Canvas Dx, a device manufactured by Cognoa, Inc., has received clearance for marketing as an aid in the diagnosis of Autism Spectrum Disorder (ASD) for children aged 18 to 72 months who are at risk for developmental delay. Functioning as an adjunct to the clinical diagnostic process, this software and data system-based device provides diagnostic outputs classified as 'Positive for autism,' 'Negative for autism,' or 'Indeterminate,' thereby streamlining the initial assessment phase. A critical element enabling its marketing clearance is the inclusion of a Predetermined Change Control Plan (PCCP). This plan allows for prospective, planned updates and modifications to the device's underlying machine learning model and performance thresholds, utilizing real-world data within its cleared indications. Such modifications can encompass optimizing decision thresholds, retraining the model with new datasets, and incorporating additional neurodevelopmental or neurobehavioral features as inputs, all while maintaining rigorous performance evaluation, software verification, and validation, ensuring ongoing regulatory compliance for device quality, labeling, and reporting.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the specific machine learning model within Canvas Dx process various inputs to arrive at its diagnostic outputs of positive, negative, or indeterminate for autism?; What specific neurodevelopmental and neurobehavioral features are utilized as inputs by the Canvas Dx device, and how are these features objectively collected or assessed?; What are the established performance evaluation metrics and thresholds that trigger or validate updates to the machine learning model under the Predetermined Change Control Plan?; Beyond its role as a diagnostic aid, what are the recommended clinical next steps or pathways for patients who receive 'indeterminate' or 'negative' results from Canvas Dx?; How does the integration of real-world data through the PCCP ensure the long-term diagnostic accuracy and address potential biases or evolving characteristics of ASD in the target population?