New technology-based tools can help pediatricians to make an accurate diagnosis of autism spectrum disorder.
Although pediatricians screen for autism spectrum disorder (ASD) in children aged 18 to 24 months during routine health maintenance exams, this neurodevelopmental condition often eludes formal diagnosis until a child is 4 years or older.1,2 One telehealth application is currently expediting the diagnosis of ASD by specialists. Another artificial intelligence (AI)–based diagnostic system, which recently received marketing approval by the US Food and Drug Administration (FDA), may facilitate the diagnosis by pediatricians. This article examines these technology-based tools and explores how they can aid pediatricians.
Children with ASD have communication and social impairments, frequently demonstrate repetitive or restrictive behaviors, and often have a wide range of comorbidities. The condition is common, affecting 1 in 59 children in the United States.3 As most pediatricians are not able to diagnose children with developmental and speech delays as having ASD, they refer children to specialists, who include developmental pediatricians, child psychologists, pediatric neurologists, and pediatric psychiatrists. In many circumstances a child suspected of having ASD is referred to a multidisciplinary diagnostic team composed of one of the above specialists as well as educators, speech pathologists, occupational therapists, and physical therapists. There are often wait times of many months until an ASD evaluation is performed. Even after a child is evaluated, they may not meet threshold criteria for an ASD diagnosis, and the assessment is considered inconclusive. In such situations, parents and pediatricians must wait until the child is older and upon repeat assessment receive a diagnosis of ASD or have the diagnosis excluded. The diagnosis of ASD is very important for families (vs “developmental delay”) as it assures that insurance companies and school systems will provide much-needed behavioral interventions and services, which are mandated by most states.
Many factors may delay the diagnosis of ASD. For instance, individuals in low-income communities and minorities have limited access to services necessary to diagnose ASD.4 Only 60% of pediatricians screen children for developmental delays despite the recommendation from the American Academy of Pediatrics (AAP) to perform screenings at 18- and 24-month well visits. In one study, only two-thirds of those who failed screening were referred for a diagnostic ASD evaluation.5,6 In addition, it was recently shown that the Modified Checklist for Autism in Toddlers Revised With Follow-Up (M-CHAT-R/F), used by most pediatricians to screen for ASD, has sensitivities as low as 39% in detecting children with ASD.7 Lastly, the COVID-19 pandemic has resulted in significant delays in evalutating children with suspected ASD.
Screening for ASD
Pediatricians have a plethora of ASD screening tools available, such as the Ages and Stages Questionnaires, the Communication and Symbolic Behavior Scales, the Parents Evaluation of Developmental Status, the Screening Tool for Autism in Toddlers and Young Children, and M-CHAT-R/F. Most pediatricians use M-CHAT-R/F to screen children for autism at 18 months and 24 months of age. The M-CHAT-R/F queries parents regarding their child’s perception of others, use of gestures, interactive eye contact, vocal communication, and ability to interact with parents and children. When children refer following the ASD screen, or if a parent or pediatrician is suspicious of the diagnosis, the AAP recommends that the child be referred for a diagnostic evaluation.
In 2013, the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) consolidated several previous categories of ASD into just 1 category. To meet diagnostic criteria for ASD, a child must have 3 symptoms relating to social communication and interaction, as well as 2 out of 4 symptoms relating to repetitive or restricted behaviors (Box).8
This recharacterization and simplification of the criteria needed for the diagnosis of ASD has facilitated the diagnosis of ASD by specialists. Many use a chart similar to the Box as an intake screening form before proceeding with a diagnostic evaluation.
Developmental pediatricians or child psychologists use the Autism Diagnostic Observation Schedule (ADOS) test and/or the Autism Diagnostic Interview-Revised (ADI-R) for evaluation. The ADOS is an observational play and activity assessment of the child that takes up to an hour to complete, whereas the ADI-R is a 93-point questionnaire that may take several hours.9 Many developmental pediatricians and child psychologists use their own screening methods rather than performing a full ADOS or ADI-R, sometimes using other screening tools such as the Social Responsive Scale and the Childhood Autism Rating Scale. Most importantly, although these aids help with diagnosis, ASD is a clinical diagnosis best rendered by clinicians with the training and expertise to do so.
Children affected with ASD are a heterogeneous group, with varying degrees of functional limitations. Many have comorbidities that include seizures, attention-deficit/hyperactivity disorder, oppositional defiant disorders, sleep disorders, and speech delay. Researchers have long looked to sensor-based technologies, many employing artificial intelligence algorithms to screen for children with ASD. AI is particularly useful for identifying patterns within data and therefore can be advantageous for investigators to try to identify markers associated with an ASD diagnosis. However, AI algorithms are only as good as the training data sets inputted into a diagnostic system. Over the past few years, several AI-related technologies have been developed to improve medical care.10 Additionally, investigators have had some promising results using sensors to analyze facial expressions, vocalizations, touch sensitivity, eye tracking, movements, and interactions with robots to try to identify children with ASD. Unfortunately, to date, none of these technologies have proved sensitive enough to be used diagnostically.11
Telehealth-based tool facilitates diagnosis
In 2005, Behavior Imaging Solutions (BIS) was founded by Ronald and Sharon Oberleitner after their 3-year-old son received an ASD diagnosis. Developed over the course of several years and first made available nationwide in 2018, the system, called the Naturalistic Observation Diagnostic Assessment (NODA), consists of 2 components. Parents use smartCapture, a smartphone- based application, to fill out a developmental questionnaire and record and upload four 10-minute videos of their child (Figure 1). Various scenarios include the child playing alone, the child playing with others, a family mealtime, and a behavior of parent’s concern.
NODA Connect, a web-based portal, allows an autism specialist to analyze submitted videos for features of ASD (Figure 2). They then complete a DSM-5 checklist for ASD and determine whether the child has ASD. The clinician, not the parent, pays a small fee to BIS for use of the system. Several studies have shown that the BIS system is easy to use and renders diagnoses comparable to those produced by a traditional ADOS evaluation.12,13 In 2017, investigators compared the diagnostic accuracy of the NODA system to in- person evaluations by experienced ASD diagnosticians for 40 families seeking an evaluation for ASD and 11 families with normally developing children. The diagnostic clinicians were blinded as to which group they were evaluating and used ADOS, ADI-R, or other diagnostic tools to render a determination of whether the child being evaluated had ASD. Sensitivity between NODA and the in-person exam was 85%, and the specificity was 94%.14
BIS is in the process of integrating AI into NODA, using “vision” algorithms to tag video frames containing suspect diagnostic features. This AI is expected to be incorporated into the NODA Connect portal over the next 2 years and will speed the diagnostic process.
The advantage is that pediatricians can refer to specialists who use the NODA system rather than a formal ADOS, which can considerably shorten the wait time to diagnosis. Best of all, the evaluation can be done remotely, even in a pandemic, and expedite appropriate interventions for the child with ASD.
An AI-based ASD diagnostic looks promising
Cognoa, a digital health care company, has been working on developing a multimodular AI-based system that will enable pediatricians to diagnose ASD.15,16 In June, the FDA granted the company approval to market the first FDA-authorized diagnosis aid designed to help physicians diagnose autism in a primary care setting. The prescription only system is called Canvas Dx and is expected to be available before the end of this year.17
Cognoa AI software was trained on data sets compiled from ADOS and ADI-R score sheets of children aged 18 months to 7 years supplied by numerous ASD evaluation and treatment centers. Cognoa’s system includes a parental questionnaire, a pediatrician questionnaire, and an analysis of 2 or 3 uploaded videos that are each 1 to 2 minutes long of the child at home during mealtime or playtime (Figure 3). The videos scored by trained analysts for features of ASD who respond to a behavior questionnaire. AI algorithms analyze the questionnaires and video report, then decide on whether the child has ASD or if the evaluation is inconclusive.
A study published in March 2020 validated Cognoa’s system on 375 patients over the course of 2 years, indicating that the system could identify children with ASD with sensitivity and specificity as high as 90% and 83%, respectively.18
More recently, the company completed a double-blind clinical trial at 14 sites around the United States, using an improved algorithm to gather data for submission to the FDA. The trial involved 425 participants, aged 18 months to 6 years, whose parents or doctors expressed concern about their development but who had not previously been evaluated for ASD. Each child was assessed twice: using Cognoa’s diagnostic and by a specialist clinician based on DSM-5 criteria. The study ran from July 2019 through May 2020, so some of the children were evaluated remotely during the pandemic. The tool performed equally well when administered remotely, according to the company. The trial also showed that Cognoa’s diagnostic is highly accurate across male and female patients, as well as different ethnic and racial backgrounds.
Cognoa plans to promote their diagnostic to pediatricians to be used (1) when a child does not pass a routine ASD screening tool, or (2) the diagnosis is suspected by parent or provider by history or observed behavior. Using the diagnostic in this fashion will help identify children who should be referred to ASD specialists to corroborate the diagnosis and help educate families and recommend services. Undoubtedly, Canvas Dx will be scrutinized via independent studies. Pediatricians will look to the AAP’s Council on Children With Disabilities for recommendations regarding its use. Additionally, it may take considerable time until insurance companies pay for the test. former AAP president Colleen A. Kraft, MD, MBA, FAAP, is working with Cognoa to ensure that the company has an exceptional understanding of the needs of children, caregivers, and pediatric practices.
Diagnosing ASD remains a challenge, and recent advances in technologies may expedite the identification of affected children. A child identified with ASD at a younger age can receive behavioral therapy and other interventions at a time when they can be most beneficial. Looking ahead, various smart device–based applications are either in development or will soon be available that may assist parents in socializing children with ASD. Progress is being made, and soon parents and providers will leverage technologic advances to better care for patients with ASD.
1. Baio J, Wiggins L, Christensen DL, et al. Prevalence of autism spectrum disorder among children aged 8 years autism and developmental disabilities monitoring network, 11 sites, United States, 2014. MMWR Surveill Summ. 2018;67(6):1-23. doi:10.15585/mmwr.ss6706a1
2. Øien RA, Schjølberg S, Volkmar FR, et al. Clinical features of children with autism who passed 18-month screening. Pediatrics. 2018;141(6):e20173596. doi:10.1542/peds.2017-3596
3. Hyman SL, Levy SE, Myers SM. Identification, evaluation, and management of children with autism spectrum disorder. Pediatrics. 2020;145(1):e20193448. doi:10.1542/peds.2019-3447
4. Siu AL, USPSTF. Screening for autism spectrum disorder in young children: US Preventive Services Task Force recommendation statement. JAMA. 2016;315(7):691-696. doi:10.1001/jama.2016.0018
5. Monteiro SA, Dempsey J, Berry LN, Voight RG, Goin-Kochel RP. Screening and referral practices for autism spectrum disorder in primary pediatric care. Pediatrics. 2019;144(4):e20183326. doi:10.1542/peds.2018-3326
6. Chlebowski C, Robins DL, Barton ML, Fein D. Large-scale use of the modified checklist for autism in low-risk toddlers. Pediatrics. 2013;131(4). doi:10.1542/peds.2012-1525
7. Guthrie W, Wallis K, Bennett A, et al. Accuracy of autism screening in a large pediatric network. Pediatrics. 2019;144(4):e20183963. doi:10.1542/peds.2018-3963
8. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5). American Psychiatric Association; 2013
9. Le Couteur A, Haden G, Hammal D, McCanochie H. Diagnosing autism spectrum disorders in pre-school children using two standardised assessment instruments: The ADI-R and the ADOS. J Autism Dev Disord. 2008;38(2):362-372. doi:10.1007/s10803-007-0403-3
10. Schuman AJ. AI in pediatrics: past, present and future. Contemporary Pediatrics. July 14, 2019. Accessed August 26, 2021. https://www.contemporarypediatrics.com/view/ai-pediatrics-past-present-and-future
11. Cabibihan JJ, Javed H, Aldosari M, Frazier TW, Elbashir H. Sensing technologies for autism spectrum disorder screening and intervention. Sensors. 2017;17(1). doi:10.3390/s17010046
12. Nazneen N, Rozga A, Smith CJ, Oberleitner R, Abowd GD, Arriaga RI. A novel system for supporting autism diagnosis using home videos: iterative development and evaluation of system design. JMIR Mhealth Uhealth. 2015;3(2):e68. doi:10.2196/mhealth.4393
13. Oberleitner R, Laxminarayan S, Suri J, Harrington J, Bradstreet J. The potential of a store and forward tele-behavioral platform for effective treatment and research of autism. Conf Proc IEEE Eng Med Biol Soc. 2004;2004:3294-6. doi:10.1109/IEMBS.2004.1403926
14. Smith CJ, Rozga A, Matthews N, Oberleitner R, Nazneen N, Abowd G. Investigating the accuracy of a novel telehealth diagnostic approach for autism spectrum disorder. Psychol Assess. 2017;29(3):245-252. doi:10.1037/pas0000317
15. Wall DP, Dally R, Luyster R, Jung J-Y, DeLuca TF. Use of artificial intelligence to shorten the behavioral diagnosis of autism. PLoS One. 2012;7(8):e43855. doi:10.1371/journal.pone.0043855
16. Abbas H, Garberson F, Glover E, Wall DP. Machine learning approach for early detection of autism by combining questionnaire and home video screening. J Am Med Inform Assoc. 2018;25(8):1000-1007. doi:10.1093/jamia/ocy039
17. FDA authorizes marketing of diagnostic aid for autism spectrum disorder. News release. US Food and Drug Administration. June 2, 2021. Accessed August 26, 2021. https://www.fda.gov/news-events/press-announcements/fda-authorizes-marketing-diagnostic-aid-autism-spectrum-disorder
18. Abbas H, Garberson F, Liu-Mayo S, Wall DP. Multi-modular AI approach to streamline autism diagnosis in young children. Sci Rep. 2020;10(1). doi:10.1038/s41598-020-61213-w