Canvas Dx Product Development

Leveraging Machine Learning to Support Early Autism Diagnosis

Canvas Dx is the first FDA-authorized diagnosis aid for autism.1 Learn about the key milestones in our product development journey, from foundational research to clinical validation and post FDA marketing authorization algorithmic optimization. You can find summaries of all our major peer reviewed scientific publications and journal links here as well.

Leveraging Machine Learning to Support Early Autism Diagnosis

Early Stage Development

Foundational research began at Harvard Medical School’s Center for Biomedical Informatics, where Cognoa founder Dr. Dennis Wall PhD was on faculty. Early research focused on addressing the problem of ongoing delays in time to diagnosis for children with autism spectrum disorder (ASD),2 and exploring whether machine learning approaches could be leveraged to design accurate, efficient, objective and user-friendly ASD digital diagnostic tools. In this early work machine learning approaches were applied to derive maximally predictive ASD behavioral features using archived electronic patient record data from thousands of children with diverse conditions, presentations and comorbidities. Six peer reviewed scientific publications3-8 analyzed score data from 11,298 individuals with varying ASD presentations, and 1,356 individuals without ASD. Performance metrics for an initial set of eight unique machine learning ASD classifiers were derived, tested and published.3-5,8


child playing

Independent Validation of Classifiers

The accuracy of the two best performing ASD classifiers (ADTree7 and ADTree8) were then independently validated against medical record ASD score data not previously used in training, testing or classifier construction. Results were published in peer reviewed scientific journals: 5,7 


Validation Classifiers Image

Prospective Validation Studies Begin

Prospective validation studies of the evolving device inputs and algorithm are conducted and published in peer reviewed scientific journals.6, 10-12 


Prospective Validation Image

Evolution into Canvas Dx

Building on the foundational work described above, Cognoa’s research scientists refined the device inputs and algorithm to arrive at the product that was validated in the pivotal study13 and authorized by the FDA on June 2, 2021 as Cognoa ASD Diagnosis Aid, and then commercialized under the brand name Canvas Dx.1

Canvas Dx incorporates inputs from a caregiver, healthcare provider and trained video analyst. This multimodular approach is consistent with best practice recommendations that ASD evaluation include both caregiver and clinician input, as well as a structured observation of the child.14 See below for a summary of the Canvas Dx Prospective Clinical Validation Pivotal Study.13


Canvas Dx incorporates 3 separate, user-friendly inputs that feed into Canvas Dx’s algorithm:

3 types of inputs
How it works bucket

The Canvas Dx algorithm evaluates all 3 inputs, generating a device result that the physician uses when diagnosing or ruling out an ASD diagnosis in conjunction with their clinical judgement13

bbottom of the funnel image

* Trained professional with at least a Master’s degree and more than 5 years diagnosing and/or treating children with ASD.
**One of Canvas Dx’s device inputs is a 13 or 15 item age-dependent healthcare provider (HCP) questionnaire collected via a healthcare provider portal.

FDA Marketing Authorization Granted

On June 2, 2021, Canvas Dx becomes the first FDA authorized diagnostic device for autism.1


Algorithmic Update

In June of 2022, the decision thresholds underlying the Canvas Dx algorithm were updated (algorithm V2) under the predetermined change control plan that was part of the de novo request granted by the FDA in 2021 (DEN200069). Key changes to device performance are reported in Table 5. Algorithm V2 is currently included in Canvas Dx.


Algorithmic Update Image

Our Pipeline

Cognoa’s AI approach enables an extensible, equitable platform that supports the company’s product pipeline. Cognoa’s R&D programs target multiple behavioral health conditions including autism, speech and language, ADHD, and childhood anxiety, working towards developing early, efficient, and accurate diagnoses and precision treatments.

product development


  1. U.S. Food & Drug Administration. FDA Authorizes Marketing of Diagnostic Aid for Autism Spectrum Disorder
  2. Gordon-Lipkin, E., Foster, J. & Peacock, G. Whittling down the wait time: exploring models to minimize the delay from initial concern to diagnosis and treatment of autism spectrum disorder. Pediatr. Clin. 63, 851–859 (2016).
  3. Kosmicki, J. A., Sochat, V., Duda, M. & Wall, D. P. Searching for a minimal set of behaviors for autism detection through feature selection-based machine learning. Transl. Psychiatry 5, e514–e514 (2015).
  4. Levy, S., Duda, M., Haber, N. & Wall, D. P. Sparsifying machine learning models identify stable subsets of predictive features for behavioral detection of autism. Mol. Autism 8, 1–17 (2017).
  5. Wall, D. P., Dally, R., Luyster, R., Jung, J.-Y. & DeLuca, T. F. Use of artificial intelligence to shorten the behavioral diagnosis of autism. (2012).PLoS One. 2012;7(8):e43855.
  6. Duda, M., Daniels, J. & Wall, D. P. Clinical evaluation of a novel and mobile autism risk assessment. J. Autism Dev. Disord. 46, 1953–1961 (2016).
  7. Duda, M., Kosmicki, J. A. & Wall, D. P. Testing the accuracy of an observation-based classifier for rapid detection of autism risk. Transl. Psychiatry 4, e424–e424 (2014).
  8. Wall, D. P., Kosmicki, J., Deluca, T. F., Harstad, E. & Fusaro, V. A. Use of machine learning to shorten observation-based screening and diagnosis of autism. Transl. Psychiatry 2, e100–e100 (2012).
  9. Tariq, Q. et al. Mobile detection of autism through machine learning on home video: A development and prospective validation study. PLoS Med. 15, e1002705 (2018).
  10. Kanne, S. M., Carpenter, L. A. & Warren, Z. Screening in toddlers and preschoolers at risk for autism spectrum disorder: Evaluating a novel mobile‐health screening tool. Autism Res. 11, 1038–1049 (2018).
  11. Abbas, H., Garberson, F., Glover, E. & Wall, D. P. Machine learning approach for early detection of autism by combining questionnaire and home video screening. J. Am. Med. Inform. Assoc. 25, 1000–1007 (2018).
  12. Abbas, H., Garberson, F., Liu-Mayo, S., Glover, E. & Wall, D. P. Multi-modular AI approach to streamline autism diagnosis in young children. Sci. Rep. 10, 1–8 (2020).
  13. Megerian, J. T. et al. Evaluation of an Artificial Intelligence-Based Medical Device for Diagnosis of Autism Spectrum Disorder. Nat. Partn. J.- Digit. Med. (2022) doi:10.1038/s41746-022-00598-6.
  14. Brian, J. A., Zwaigenbaum, L. & Ip, A. Standards of diagnostic assessment for autism spectrum disorder. Paediatr. Child Health 24, 444–451 (2019).
  15. Shannon, J. et al. Optimizing a de novo artificial intelligence-based medical device under a predetermined change control plan: Improved ability to detect or rule out ASD in general pediatric settings. J. Am. Acad. Child Adolesc. Psychiatry 61, S242-243 (2022).

Learn about our efforts to advance pediatric behavioral healthcare.

to top