BRENNEN

Industries / Health / Healthcare / Behavioral health / Autism therapy

Assessment

Assessment is the work of finding out whether a child or adult is autistic: short screening checklists at routine checkups, a fuller diagnostic evaluation by specialists, and the counts of how many people are diagnosed.

7

Milestones achieved, from 1980 to 2025.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1980

    Autism gets its own diagnosis

    DSM-III, the third edition of the American Psychiatric Association's handbook of psychiatric diagnoses, listed "infantile autism" as an official diagnosis; Nicole Rosen, Catherine Lord and Fred Volkmar describe this as its first appearance there. Studies in the 1970s had set it apart from childhood schizophrenia, the label it had often been filed under, by age of onset and family history, and a 1977 twin study by Susan Folstein and Michael Rutter pointed to a biological origin.

    pmc.ncbi.nlm.nih.gov

    Credit

    American Psychiatric Association; Israel Kolvin, Michael Rutter, Susan Folstein and others for the research behind it

  2. 1994

    Standard interview and observation

    The Autism Diagnostic Interview-Revised (ADI-R), published in 1994, gave clinicians a semi-structured interview, a set list of questions with room to probe, for a parent or carer about a person's development, linked to the diagnostic criteria of the time. In 2000 the same group added the Autism Diagnostic Observation Schedule-Generic (ADOS-G), a standard session of play and conversation in four 30-minute modules chosen by how much the person speaks; in a study of 223 children and adults, raters scoring it agreed closely, and its second edition is now among the most widely used measures of autism traits, by Rosen and colleagues' account.

    pubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov

    Credit

    Catherine Lord, Michael Rutter, Ann Le Couteur and colleagues (Lord receives royalties from both tools, as that account discloses)

  3. 2001

    A parent checklist for toddlers

    The Modified Checklist for Autism in Toddlers (M-CHAT) asks a parent 23 yes-or-no questions; in its first study 1,293 children were screened, and 39 of the 58 then evaluated were diagnosed with autism or a related condition. Its revised form with a follow-up interview (M-CHAT-R/F) screened 16,071 toddlers at 18- and 24-month checkups in Atlanta and Connecticut, the ages at which the American Academy of Pediatrics (AAP) recommends autism screening, and the children it found were diagnosed about two years younger than the US median.

    pubmed.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov

    Credit

    Diana Robins, Deborah Fein, Marianne Barton and colleagues, University of Connecticut and Georgia State University

  4. 2011

    A three-month waiting standard

    England's National Institute for Health and Care Excellence (NICE) told the health service in 2011 to set up local multi-professional autism teams for children, and its 2014 quality standard measures the share of people whose diagnostic assessment starts within 3 months of referral. NHS England's monthly figures for June 2026 counted 294,792 people with an open referral for suspected autism, 86.8% of them open 13 weeks or more; the count was 310,794 in May, and most of the fall came from one provider that did not submit data.

    LessonNHS England says its figures undercount, because most children in the UK are assessed in child development services that its data set does not cover, so even the size of the queue is not fully known.

    nice.org.uknice.org.uknhsd-proxy.openprescribing.netnhsd-proxy.openprescribing.net

    Credit

    NICE; NHS England

  5. 2013

    One autism spectrum

    DSM-5 replaced autistic disorder, Asperger's disorder and pervasive developmental disorder not otherwise specified (PDD-NOS) with a single diagnosis, autism spectrum disorder, judged on two groups of traits, social communication and restricted or repetitive behavior, with levels of severity. The World Health Organization's ICD-11 classification also adopted autism spectrum disorder as the umbrella term, with eight subcategories by intellectual and language ability.

    Lessona 2012 reanalysis of DSM-IV field-trial cases by McPartland and colleagues suggested nearly 80% of people with Asperger's or PDD-NOS diagnoses might lose their label, while studies using fuller item data did not find such losses, so DSM-5 let people with well-established earlier diagnoses keep them; how many would lose a diagnosis depended on which data were used.

    pmc.ncbi.nlm.nih.gov

    Credit

    American Psychiatric Association; World Health Organization

  6. 2019

    Screening misses many children

    In 2016 the US Preventive Services Task Force, an independent expert panel, found the evidence insufficient to weigh the benefits and harms of screening toddlers whose parents and doctors had raised no concern, while the AAP kept recommending screening every child at 18 and 24 months. A Children's Hospital of Philadelphia team then followed 25,999 children screened at checkups in 2011 to 2015 with the original M-CHAT and its follow-up interview until age 4 to 8: the checklist flagged 38.8% of the children later diagnosed, 14.6% of those it flagged were diagnosed, and it was less accurate for children of color and from lower-income households.

    Lessonstudies that follow up only the children who screen positive cannot count the ones a checklist misses, and in routine care with every child followed it missed most of them.

    pubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.gov

    Credit

    A. L. Siu and the US Preventive Services Task Force; S. L. Hyman, S. E. Levy and S. M. Myers for the AAP; W. Guthrie, J. S. Miller and colleagues, Children's Hospital of Philadelphia

  7. 2025

    Counting autism by records

    The US Centers for Disease Control and Prevention's Autism and Developmental Disabilities Monitoring (ADDM) Network has counted autism among 8-year-olds every two years since 2000 by reviewing health and education records in selected communities. Its count rose from 1 in 150 children in 2000 to 1 in 36 in 2020 and 1 in 31 (32.2 per 1,000) in 2022, ranging from 9.7 per 1,000 in Laredo, Texas, to 53.1 in its California site; across 99 estimates from 71 studies in 34 countries published since 2012, a global review found a median of 100 per 10,000, or 1%.

    Lessonboth teams link the spread and the rise to differences in awareness, services and diagnostic practice, so the numbers measure identification as well as autism itself.

    pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov

    Credit

    CDC's ADDM Network (K. A. Shaw and colleagues); J. Zeidan, M. Elsabbagh and colleagues for the global review

Watch for

  1. AI diagnosis aid in clinics

    Canvas Dx, AI software from Cognoa, was authorized by the US Food and Drug Administration (FDA) in 2021 to help diagnose autism in children aged 18 to 72 months. In its pivotal study of 425 children only 31.8% got a yes-or-no result, the rest being told the device could not decide; Cognoa's analysis of its first 254 prescriptions in routine use reports a yes-or-no result for 63.0% of children, with 92.4% of positive results and 97.6% of negative results matching the clinician's diagnosis.

    Achieved whena team with no ties to Cognoa compares the device's output with specialist diagnosis in several hundred children in routine care; a report from George Washington University and Children's National Hospital in November 2026, by authors with no stated tie to Cognoa, covered 28 children and called its own estimates imprecise.

    pubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.govpubmed.ncbi.nlm.nih.gov

    Claimed by

    Cognoa (Carmela Salomon, D. P. Wall and colleagues; Wall is a co-founder), August 2025

  2. Eye tracking for older children

    The EarliPoint system measures where a child looks while watching videos of social scenes and scores the pattern; a 2023 study of 475 children aged 16 to 30 months, led from Emory University and the Marcus Autism Center in Atlanta, found 71.0% sensitivity (the share of autistic children it flagged) and 80.7% specificity (the share of other children it cleared) against expert diagnosis. In March 2026 the FDA cleared it for ages 16 months to 7 years on the strength of the company's study in children aged 31 months to 7 years, which reports 77.1% sensitivity and 78.6% specificity once results in a new "borderline" band are set aside, and 63.5% sensitivity when they are counted as not positive.

    Achieved whena group with no ties to EarliTec, Emory University or the Marcus Autism Center, tests it in routine clinics in children aged 31 months and older.

    accessdata.fda.govpubmed.ncbi.nlm.nih.gov

    Claimed by

    EarliTec Diagnostics, March 2026

Open

  1. 01

    One table of diagnosis waits

    A public table, updated each quarter, of the official waiting figures for autism assessment from every country or region that publishes them, with each figure's definition written out: what starts and stops the clock, which ages, and which services are counted, so readers can see which figures can fairly be compared.

    The brief
    Credit · unclaimed
  2. 02

    An open queue forecast

    An open model of England's queue for autism assessment, built only from NHS England's published monthly counts of new, closed and open referrals, that forecasts the queue 6 and 12 months ahead and shows how it would change under stated changes in assessment capacity.

    The brief
    Credit · unclaimed

Open 01 · the brief

One table of diagnosis waits

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public table, updated each quarter, of the official waiting figures for autism assessment from every country or region that publishes them, with each figure's definition written out: what starts and stops the clock, which ages, and which services are counted, so readers can see which figures can fairly be compared.

Why it is still open

England publishes monthly counts, but NHS England labels them official statistics in development and says they leave out the child development services where most UK children are assessed. PubMed searches in November 2026 found no study that sets official waiting figures from different countries side by side.

nhsd-proxy.openprescribing.net

Built with what exists

National statistics releases such as NHS England's monthly Autism Statistics; health ministry reports and answers to parliamentary questions; a web-search AI agent to find new releases each quarter, with a person checking every number; a spreadsheet or SQLite database; a static site on GitHub Pages.

  1. First step

    Search the health ministries and statistics offices of the four UK nations, Ireland, Australia and Canada for published autism assessment waits, and record which publish one and how each defines the wait.

  2. Done looks like

    At least 10 countries or regions, each with a figure no more than 12 months old, linked to its release and coded on 5 definition fields; two people check 30 random figures and agree on every one; four quarterly updates published.

Open 02 · the brief

An open queue forecast

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An open model of England's queue for autism assessment, built only from NHS England's published monthly counts of new, closed and open referrals, that forecasts the queue 6 and 12 months ahead and shows how it would change under stated changes in assessment capacity.

Why it is still open

In June 2026 NHS England counted 17,509 new referrals and 13,492 closed ones, so the queue was still growing. The demand-and-capacity work published so far covers single local services, such as Cambridgeshire's adult autism service, where referrals rose from 430 in 2019 to 887 in 2023, and no open national forecast was found in searches of Europe PMC in November 2026.

nhsd-proxy.openprescribing.netpmc.ncbi.nlm.nih.gov

Built with what exists

NHS England's Autism Statistics data files back to April 2019; Python with pandas and statsmodels for time-series forecasts; SimPy for a simple simulation of the queue; a public GitHub repository.

  1. First step

    Download the national monthly series, fit a model of new minus closed referrals, and compare its prediction for the latest month with the published count.

  2. Done looks like

    Forecasts published each month for 6 months and scored against each new release, with the mean error in referrals stated; every scenario listed with its assumptions; code and data public.