State Medicaid Brief — Canvas Dx, FWA, and the Diagnostic Gate


Canvas Dx · State Medicaid Policy Brief

Fix the autism diagnostic gate.
Build the infrastructure.

Canvas Dx is State Medicaid’s FDA-cleared diagnostic stop-gap against the FWA inflation that four OIG audits have now documented — and the data infrastructure that builds the precision-care journey at national scale. Full coverage, RHTP-funded on-ramp, $1,000 per assessment with outcomes-linked pricing. Act early; avoid the $15–$25B in national Medicaid obligation that compounds over the next decade under the current diagnostic latency.
Cognoa · Canvas Dx
April 2026
All claims cited to peer-reviewed publications or government sources; internal modeled estimates noted as such

Medicaid mean age at autism diagnosis
~60 months
11+ mo later than the 49-mo national mean; well past the 18–36 mo neuroplasticity window

Qualifying specialists per child 18–72 mo
1 : ~23,500
~850 board-certified DBPs nationally

With Canvas Dx · primary-care enabled
1 : ~300
~80× capacity expansion
~67,000 pediatricians + family physicians

10-yr Medicaid obligation delta per early Dx
$250,600
$385K late − $134K early · modeled
▸ National scale of the problem

$15–$25B in avoidable 10-year Medicaid obligation this decade
At current diagnostic latency, the US Medicaid-enrolled autism cohort is accumulating an estimated $15–$25 billion in avoidable 10-year obligation this decade from the gap between what is medically achievable (18-month diagnosis, FDA-label floor) and what the system operationally delivers (49-month national mean, later in Medicaid). National Medicaid ABA spend reached $14.49B in 2024, +279% since 2018 at 25% CAGR (Distilinfo March 2026; Behavioral Health Business Dec 2025) — growth outpacing oncology, diabetes, and every other healthcare category. Canvas Dx is the FDA-cleared stop-gap that also builds the national precision-care data infrastructure to prevent this debt compounding into the next decade.
Modeled: ~20M US children 18–72 mo × ~3.2% prevalence (CDC ADDM 2022) × ~40% Medicaid share × $250,600 per-child 10-yr delta (modeled) × 30% achievable-at-scale factor.

Navigator · policy brief + interactive dashboard
Click any tab to continue · Tab 01 is the written brief; Tabs 02–04 are live dashboards · the state selector lives on Tabs 03 and 04.

▸ The human arithmetic

One Medicaid child. Two paths.

The economics and the enforcement case both resolve to a single child. Here is that child under today’s pathway, and under Canvas Dx.

Today’s pathway

Specialist-only diagnosis · Medicaid tracks ~60 mo · ~42 months of friction
Age 18–24 months

Parent raises a developmental concern at the well-child visit. M-CHAT administered; returns positive. Pediatrician says “let’s check at your next appointment” — watchful-waiting is the default because PCPs can’t diagnose today.
Age 24–30 months

Next well-child visit. Concern persists; pediatrician places a referral to a developmental specialist (DBP, child neurologist, or licensed psychologist with autism training).
Age 30+ months · joining the waitlist

Specialist waitlists routinely exceed 12–18 months in Medicaid-enrolled populations. Several states have fewer than five qualifying specialists. Interim visits and misrouted therapy begin at $331/mo all-cause (Vu JMCP 2023).
Waitlist duration · ~12–24 months

Family waits. Rural and lower-resource households face 100+ miles and 2–4 hour round-trip to the specialty center when the appointment finally arrives. Missed work, transportation cost, childcare coordination for siblings — all during the highest-value neuroplasticity window.
Specialist visit 1 · intake

Developmental history, parent interview, initial observation. Scheduled separately from instrument administration because capacity is constrained.
Specialist visits 2–3 · instrument administration

ADOS and/or CARS (research tools repurposed as standard of care), clinical interview, scoring, report preparation. 6–8 cumulative billable specialist hours per child. Each visit requires another family commute.
Age ~60 months (Medicaid-adjusted)

Results delivered. DSM-5 autism diagnosis communicated by a specialist whose practice is often tied to the downstream ABA referral network. Next steps of care discussed. National mean is 49 mo; Medicaid-enrolled children track later.
+ 201 days

Lag from diagnosis to ABA enrollment (Leslie-Fournier 2025).
Age ~66 months

ABA begins at 10 hrs/wk under FFS default. Neuroplasticity window long missed. Authorization anchored to subjective specialist diagnosis — the OIG-flagged integrity gap.
10-yr Medicaid obligation · modeled
$385,200

Canvas Dx pathway

Primary care-enabled · algorithm-anchored · ~1 week start-to-results
Age 18–24 months

Same well-child visit. Same concern. M-CHAT returns positive. Pediatrician orders Canvas Dx at the point of care — no specialist referral, no family commute. Parent completes a short app questionnaire and uploads two home videos.
+ 48–72 hours

Canvas Dx returns a DSM-5-mapped result and auto-generated documentation report — objective, algorithm-locked, FDA-cleared, independent of referral relationships. ~1 hour of billable clinician time total. Clinician reviews, confirms, and discusses next steps with the family.
Right-matched care begins · within weeks

For confirmed autism: ABA as primary intervention (~85% of Canvas-Dx-Positive / clinician-confirmed cases, standard of care), delivered through VBC contracts (Catalight or equivalent) at the intensity the evidence supports — ~5 hrs/wk rather than the FFS 10–40 hrs/wk default. Speech, OT, Early Intervention, and Special Ed paired as clinically indicated. Canvas Dx operational target population mean: ~36 months, trending toward the 18-month FDA-label floor as deployment matures.
Age 5–6

37% of early-diagnosed children no longer meet DSM-5 criteria for autism by school age (Harstad JAMA Pediatrics 2023).
10-yr Medicaid obligation · Canvas Dx + VBC · modeled
$134,600
Timeline compressed (Medicaid-adjusted)
−24 mo
Avoidable 10-yr obligation · modeled
−$250,600
ABA intensity (right-sized via VBC)
−50%
Non-persistence by school age
37%

▸ The ask — immediate action

Full coverage. RHTP-funded on-ramp. Durable coverage post-RHTP.

Three converging decisions under one coverage commitment: add Canvas Dx to State Medicaid as full coverage today, allocate already-appropriated RHTP dollars as the funding on-ramp with outcome metrics unlocking continued tranches, and transition to standard durable coverage after the RHTP window when the evidence base is in place. The state is not piloting a diagnostic tool — it is establishing the precision-care infrastructure that defends against the next OIG audit and shapes the national standard of care.
▸ The three-part ask
1. Full coverage of Canvas Dx under State Medicaid, effective immediately.
$1,000 per Canvas Dx assessment, paid directly to Cognoa as the NPI vendor — not billed by the ordering clinician as a procedure code. Canvas Dx becomes the accepted diagnostic anchor for ABA prior-authorization. Vendor-direct payment removes billing friction for PCPs (no coding burden, no attribution dispute with downstream ABA providers) and keeps the diagnostic service auditable as a discrete, algorithm-anchored artifact. No pilot phase — full coverage from day one.
2. RHTP-funded deployment and evidence generation as the coverage on-ramp.
The state allocates Rural Health Transformation Program dollars to Canvas Dx deployment during the RHTP window (through October 2030). RHTP priorities — rural access, behavioral health, health-technology innovation — match Canvas Dx directly; federal dollars are already appropriated, no new legislation required. Jointly-designed outcome metrics unlock continued RHTP funding tranches:

  • Age-at-diagnosis compression — initial target 49 → 36 months, trending toward the 18-month FDA-label clinical floor
  • FWA reduction — Negative-cohort ABA authorization rate (baseline Cognoa deployment data: ~5% within label age)
  • DEV-CH loop closure rate at the PCP (CMS Child Core Set reporting)
  • Service-routing diversity — precision-care pathway evidence; correlation of challenge-score-by-domain with downstream referrals
  • 24-month non-persistence rate — outcome-assessment cohort tracking against the Harstad 2023 37% benchmark
3. Durable coverage pathway post-RHTP.
Once the RHTP funding window closes (October 2030 or earlier if the state opts for acceleration), Canvas Dx transitions to standard Medicaid coverage under the negotiated rate structure. The 5+ years of evidence-generation data inside the RHTP window establishes the state-specific durable coverage case — not a hope, a documented track record. The RHTP window is the bridge, not the destination.
▸ Why now — the compounding cost of delay
Three OIG audits remain in the seven-state ABA audit series; the four completed averaged $49.5M improper payments per state. RHTP dollars are appropriated through October 2030 — states that allocate early capture the full evidence-generation window; states that wait face both audit exposure and missed RHTP capture. At the national scale, the US Medicaid-enrolled autism cohort accumulates an estimated $15–$25B in avoidable 10-year obligation this decade (modeled). Every state’s decision window is also a share of the compounding national debt. Acting early — now, with RHTP as the catalyst — is the difference between building the precision-care infrastructure and inheriting the billions in avoidable cost.
▸ Evidence generation & discourse
This brief is put forward as a starting point for discourse, not a closed argument. All cited figures are drawn from peer-reviewed publications or government sources; internal modeled estimates are flagged as such. Cognoa is committed to co-designing the evidence-generation workstream alongside the coverage decision — instrumenting the covered cohort to track the five outcome metrics above, reportable to CMS, the state legislature, and the broader field on a 12-month cadence. The data capture enabled by this workstream is what makes stepped-care pathway design possible: correlating challenge-score-by-domain to downstream service routing, stratifying by clinician type (PCP vs. specialist), and ultimately shifting from blanket multi-modal referral to precision-matched care. Canvas Dx coverage is the diagnostic stop-gap today; the data infrastructure it enables is the precision-care system tomorrow. We welcome challenge to the evidence base and partnership on the outcomes definition.

▸ The supporting data

Below: the full argument that supports the ask above.

Access failure · workforce bottleneck · FWA exposure · Canvas Dx intervention · care pathway · per-child economics · per-assessment value derivation.

▸ The problem in three steps

The diagnostic gate is the bottleneck.

A structured read of how today’s specialist-only pathway produces an access failure, a workforce-impossibility constraint, and an FWA exposure — all traceable to the same diagnostic-integrity root cause.

01
The access failure

Medicaid-enrolled children are diagnosed at a mean of ~60 months: 11+ months later than the 49-month national average.

The national mean is 49 months (CDC ADDM 2022, published April 2025); Medicaid-enrolled children track later due to specialist access barriers documented across the literature (Mandell et al., Shattuck et al.). Reliable autism diagnosis is medically possible at 18 months (Megerian et al. NPJ Digital Medicine 2022). The operational gap for State Medicaid is ~42 months — the 18–36-month window when early intervention produces its largest outcomes effects is not just missed, it’s missed by years. Everything downstream — cost, treatment mix, rural access gap, FWA exposure — flows from this access failure.

Today · Medicaid mean
~60 mo
Medicaid-enrolled children track 11+ months later than the 49-mo national mean due to specialist access barriers (Mandell et al.; Shattuck et al.). National CDC ADDM 2022 baseline shown for comparison.
Canvas Dx operational goal · initial
~36 mo
Realistic compression target from the Medicaid-adjusted mean, through PCP-deployed diagnosis. Moves children into intervention inside the 18–48 month neuroplasticity envelope rather than past it.
FDA-label floor · long-term direction
18 mo
Canvas Dx is FDA-cleared for diagnostic aid at 18 months (Megerian NPJ Digital Medicine 2022). As PCP deployment scales and workflows mature, the operational age trends toward this floor.
02
The workforce bottleneck

Only specialists can diagnose today. ~850 DBPs nationally, for ~20M children in age range.

State Medicaid coverage typically restricts autism diagnosis to developmental-behavioral pediatricians, child neurologists, and licensed psychologists with autism-specific training. There are ~850 board-certified DBPs nationally (SDBP / ABP 2024 workforce) — a ratio of 1 qualifying specialist per ~23,500 children 18–72 months. Several states have fewer than five qualifying specialists. This is not a waitlist problem. It is a workforce-impossibility problem.

The math makes the impossibility concrete. A specialist autism evaluation today runs 6–8 hours of billable specialist time per child — intake, structured instrument administration (ADOS, CARS), scoring, clinical interview, report. Canvas Dx returns a DSM-5-mapped result at the pediatrician in ~1 hour of billable clinician time. Even at a notional 30% of specialist capacity dedicated to autism evaluation (the rest goes to other developmental conditions), the existing workforce cannot clear the ~143K new in-age-range autism children entering the system each year at 7 hrs each — the arithmetic requires roughly 2× the specialist-hours that exist. The access gap is not a scheduling problem; the specialist-only pathway is structurally incapable of meeting demand at any realistic level of effort. A PCP-deployed, FDA-cleared diagnostic aid is the only way to close the gap.

03
The FWA consequence

A subjective diagnostic gate produces authorization integrity failures. Four OIG audits confirm it.

Four completed HHS OIG audits found improper payments in 100% of sampled enrollee-months — services never rendered, hours beyond physical possibility, caseloads stacked on providers, and authorizations without a confirmed diagnosis. Canvas Dx closes the diagnostic-anchor gap directly; other fraud vectors remain under state enforcement. National Medicaid ABA spend reached $14.49B in 2024 (+279% since 2018; 25% CAGR — Distilinfo March 2026), a signal that compounds every month the gate stays subjective.

HHS OIG ABA audit series · improper payments
Indiana (Dec 2024)  $56.0M
Wisconsin (Jul 2025)  $18.5M
Maine (Jan 2026)  $45.6M
Colorado (Mar 2026)  $77.8M
3 audits remaining in the 7-state series
↳
The Switch

Fix the gate. Canvas Dx is the stop-gap today and the precision-care infrastructure tomorrow.

▸ 03 · The intervention

What Canvas Dx is.

Canvas Dx is the FDA De Novo-authorized AI-assisted diagnostic aid for autism in children 18–72 months. Three dimensions matter for any coverage decision: regulatory status, clinical performance, and operational deployment.

▸ Regulatory

FDA De Novo-authorized

Class II De Novo DEN200069, June 2021. First and only AI-assisted diagnostic aid cleared for autism spectrum disorder. Indication: children 18–72 months at risk for developmental delay. The algorithm is locked; it produces a DSM-5-mapped output that is used as an adjunct to clinician assessment per FDA label — the clinician remains the diagnosing authority.
▸ Clinical performance

95% PPV · 97% NPV

Canvas Dx continues to improve as it ingests more data — current performance supports high diagnostic confidence (95% positive predictive value) on positive calls and high confidence when ruling out (97% negative predictive value) on negatives. Current determinate rate 78%; on Indeterminate reports, the diagnosing clinician still determines next steps of care, leveraging the auto-populated DSM-5-mapped documentation report Canvas Dx generates. Validation method: each Canvas Dx result is compared against a blinded specialist evaluation, independently confirmed by a second specialist blinded to both Canvas Dx and the first specialist’s read. Authorized for ongoing improvement under a predetermined change control plan — 510(k) K243558. Baseline real-world evidence on file: Salomon et al., Nature Scientific Reports 2025 (n=254 prescriptions, 15 states).
▸ Deployment

Primary care — standard pediatric workflow · telehealth-validated

Runs inside the existing well-child visit workflow. Parent completes an app questionnaire and uploads two short home videos; the ordering pediatrician completes a portal entry. A DSM-5-mapped result and auto-generated report return to the clinician in 48–72 hours. No new certification, no specialist-equivalent training. Clinically validated for telehealth use, eliminating the family commute to a specialty center. Total billable clinician time: ~1 hour — vs. 6–8 hours of billable specialist time in the current pathway, plus weeks of family scheduling and travel.
▸ Objective, algorithm-locked diagnostic baseline
Today’s standard of care relies on repurposed research instruments (ADOS, CARS) delivered by specialists whose practices are often tied to the same downstream therapy networks the children they diagnose will be referred to. Canvas Dx changes that dependency: an FDA-cleared, third-party, algorithm-locked diagnostic aid produces a consistent, quantitative baseline regardless of geography, practice, or referral relationship. Clear positives and clear negatives are routed to appropriate care months before a child would typically reach a specialist — freeing specialist capacity for the most complex cases, where clinician judgment modifies the Canvas Dx baseline with case-specific nuance. The national baseline becomes consistent; specialist expertise is deployed where it adds the most clinical value. Canvas Dx closes the diagnostic-anchor portion of ABA authorization integrity; it does not independently prevent over-billing, phantom services, or caseload stacking — those remain under the separate enforcement mechanisms Medicaid programs already operate.

▸ 04 · How it runs in clinical workflow

The Canvas Dx care pathway.

Five steps from parental concern to right-matched care — all inside the pediatric medical home, no specialist referral required.

01
Well-child visit
At the 18- or 24-month AAP-recommended well-child visit, the pediatrician administers the M-CHAT screen during the standard developmental check.
Setting · pediatric primary care
02
Canvas Dx ordered
On a positive M-CHAT or parental developmental concern, the pediatrician orders Canvas Dx at the same visit. No specialist referral. No queue.
Ordering clinician · PCP
03
Parent + clinician inputs
Parent completes app questionnaire and uploads 2 short home videos (~15 min). Clinician completes portal questionnaire.
Cycle time · at-home, async
04
Algorithmic result
FDA-cleared algorithm returns a DSM-5-mapped result in 48–72 hours — positive, negative, or indeterminate — with full audit trail to source inputs.
Turnaround · 48–72 hrs
05
Right-matched routing
Clinician reviews, confirms or overrides. Positive or indeterminate cases route to the service that actually fits — not defaulted to ABA.
Disposition · care plan
▸ Post-diagnosis therapy referral distribution · Cognoa in-label deployment (18–71 mo) · n=442 autism-diagnosed / 1,374 total referrals
Applied Behavior
27%
Speech-Language
27%
Occupational Therapy
20%
Special Education
12%
Early Intervention
10%
Specialist referral
4%

Reading the referral data requires separating two framings. Per autism-diagnosed patient (in-label age range): ~82% are referred to ABA, ~81% to Speech, ~61% to OT — autism is treated with multi-modal care, ~3.1 referrals per child on average. This is standard of care. Per referral-volume share (shown in the bars above): ABA accounts for 27% of all referrals, Speech 27%, OT 20%. Canvas Dx’s effect on the ABA authorization pipeline is visible in the Negative-output cohort: when Canvas Dx returns Negative and the clinician confirms not-autism, ABA is referred in ~5% of cases (6 of 121 scripts; 2–8% by age band) — those children are kept off the ABA authorization trail OIG audits have flagged. Canvas Dx doesn’t reduce ABA referral among confirmed autism; it anchors every ABA authorization to an objective, algorithm-locked, third-party baseline, and keeps clinically-confirmed not-autism out of the pipeline.

Post-diagnosis: the Catalight care model as cost offset to traditional ABA.

▸ Care delivery partner

Canvas Dx identifies the right children early; the question a Medicaid program then faces is whether those children are routed into the traditional fee-for-service, high-intensity ABA default — or into a value-based care model that delivers at the intensity the clinical evidence supports. Catalight is that alternative. Cognoa and The Catalight Group formalized a national partnership to scale a streamlined journey from early diagnostic evaluation to evidenced-based treatment. In a field where diagnosis alone can take 12–24 months, PCPs use Canvas Dx to diagnose in days and refer directly to Catalight, which places patients into care within 10 business days of the assessment. Every child is tracked on a continuous, transparent outcomes record that payers, providers, and families can see — from the moment of concern through active treatment and beyond.

What Catalight is

A non-profit behavioral health network delivering personalized autism care through 15,000+ practitioners to 25,000+ patients annually. The Catalight Research Institute publishes peer-reviewed research that directly challenges the CASP 25–40 hrs/wk dosing default — showing that fewer than 9 hrs/wk produces significant gains (Sneed et al. 2023, n=106; Behavior Analysis: Research and Practice) and that hours do not correlate with adaptive-behavior outcomes (Samelson, Sneed & Pfingston 2025, n=725; J Autism Dev Disord). Findings codified in the Catalight Practice Guidelines.

30% cost reduction under VBC
Care placement in 10 business days

Why the partnership creates a care cost offset

Canvas Dx + Catalight together operationalize the model the April 2026 partnership announcement describes: objective diagnosis, individualized treatment, outcomes-based payment — what Catalight CEO Susan Armiger calls “equitable, sustainable autism care that focuses on quality of care over quantity of hours.” Early diagnosis plus value-based delivery produces lifetime cost savings that neither intervention produces alone. The 10-year Medicaid delta of $250,600 per early-diagnosed child assumes the post-diagnosis pathway does not default back to the FFS high-intensity ABA that drives the CAGR growth signal. A Catalight-style delivery partner is what makes the Canvas Dx savings durable.

Evidence base for <9 hrs/wk
Continuous outcomes tracking

▸ 06 · Per-child economics

Access integrity — the modeled avoidable Medicaid obligation is $250,600 per child.

State Medicaid’s mandate is best-possible lifetime outcomes for its enrolled children and their families — not P&L protection (the commercial-payer concern). The access-integrity case: when children are correctly identified early and routed to right-intensity care via VBC delivery, avoidable Medicaid obligation compresses by $250,600 per child over 10 years (modeled, components sourced). Under EPSDT, states are federally obligated to cover medically necessary diagnostic and treatment services for Medicaid-enrolled children; the mean annual Medicaid cost per autism child is $22,653 (Shi et al. PMC3534815). Acting on the diagnostic gate today is how that obligation is controlled.

Late Diagnosis · 10-Year Medicaid Obligation

ABA intensity 10 hrs/wk · Yr 1
Catalight Practice Guidelines 2024
$62,400
Continued high-intensity ABA, Yr 2–4
~$62,400/yr × 3 yrs
$187,200
Ongoing support · Yr 5–10
Special ed + support services
$120,000
Carrying cost during diagnostic delay
Vu M et al. JMCP 2023 · $331/mo all-cause
$15,600
10-yr total obligation (modeled)
$385,200

Early Diagnosis + VBC Delivery · 10-Year Medicaid Obligation

ABA intensity 5 hrs/wk · Yr 1
Sneed 2023 (n=106); Samelson 2026 (n=725); Catalight Practice Guidelines 2024
$31,200
Shorter treatment duration, Yr 2
Catalight Practice Guidelines 2024
$31,200
Service utilization offset · Yr 2–5
Cidav JAACAP 2017 · −$19,000/yr
−$76,000
Residual ABA + support services · Yr 3+
After applying 37% non-persistence discount
Harstad JAMA Pediatrics 2023 (37% non-persistence); remaining obligation for non-remitting 63% cohort
$148,200
10-yr total obligation (modeled)
$134,600
Avoidable Medicaid obligation per child · 10-yr delta MODELED
Composite of published line items — Catalight 2024 (ABA dose-response), Cidav 2017 (service-utilization offset), Harstad 2023 (37% non-persistence), Vu 2023 (carrying cost). Assumes VBC delivery in place for the early-dx cohort. Not rolled into the Primary Value Total below (Section 06); shown as the long-horizon consequence of acting on the diagnostic gate.
$250,600

Sources: Shi et al. Autism Research & Treatment PMC3534815 ($22,653/yr Medicaid autism cost); Catalight Practice Guidelines 2024 + Sneed et al. 2023 + Samelson et al. 2026 (ABA dose-response — significant gains at <9 hrs/wk, diminishing returns after ~15 hrs/wk); Cidav Z et al. JAACAP 2017 PMC7007927 ($19K/yr service utilization offset post-EI); Harstad E et al. JAMA Pediatrics 2023 (37% non-persistence with early dx); Lovaas 1987 (47% mainstreaming with intensive early intervention); Vu M et al. J Manag Care Spec Pharm 2023;29(4):378 doi:10.18553/jmcp.2023.29.4.378 ($331/mo all-cause excess spend); Buescher JAMA Pediatrics 2014 ($1.4M–$2.4M lifetime cost avoidance); EPSDT 42 U.S.C. § 1396d(r); CMS 2024 Autism Infographic.

▸ The FWA stop-gap and the per-assessment value case

Per-assessment value, derived — anchored by authorization integrity.

The structural gap every completed OIG audit has pointed to is not clinician record-keeping — it is the misalignment between the specialist evaluating a child and the downstream therapy networks the same child will be referred to, combined with the subjectivity of repurposed research instruments (ADOS, CARS) used as standard of care. Canvas Dx inserts an FDA-cleared, third-party, algorithm-locked diagnostic aid at the authorization gate: a consistent quantitative baseline that is independent of referral relationships, independent of geography, and produced months before a child would typically see a specialist. Specialist capacity is then reserved for the complex cases where clinical nuance modifies the Canvas Dx baseline; clear positives and clear negatives are routed to the right care immediately. Requiring Canvas Dx to validate active ABA authorizations does not replace over-billing, phantom-service, or caseload-stacking enforcement mechanisms states already operate; it closes the specific diagnostic-anchor gap the OIG series has repeatedly identified as the entry point for the rest. The per-assessment value case below quantifies that stop-gap plus the additional state-Medicaid-specific drivers.

Six state-Medicaid-specific value drivers, summed to a Primary Value Total, converted to a WTP ceiling and a target price. Line items are scoped to the state Medicaid agency buyer; MCO-level drivers (HCC/RAF uplift, member retention, OON SCA avoidance) are reported separately in the embedded Value Matrix at the bottom of the Tab 03 ROI calculator because the state captures them indirectly through capitation rate-setting. Every magnitude is either cited to a published source or explicitly labeled as modeled.

Value driverPer-case valueBasis / source
FWA preventionAuthorization integrity, per case$2,500Indiana OIG $56M improper ÷ ~25K active ABA cases/yr = $2,240 floor; four-audit average extrapolated. HHS OIG Audit Series 2019–2026 (IN $56M, WI $18.5M, ME $45.6M, CO $77.8M).
Neuroplasticity NPVLifetime dependency reduction, discounted$5,0005% reduction in $1.4M lifetime Medicaid cost per autism-without-ID child, 7% discount rate, 30-yr horizon. Buescher et al. JAMA Pediatrics 2014 ($1.4M lifetime cost) + CMS lifetime cost modeling. Modeled.
Specialist capacity reallocationState workforce optimization$1,000State-level developmental workforce time redirected from routine autism evaluation to complex cases. SDBP 2024 workforce data × loaded hourly cost. Modeled.
ED / crisis avoidanceBehavioral crisis prevention$3,7500.3 psychiatric ED encounters prevented per child × $12,500 Medicaid FFS ED cost. HCUP State ED Database 2022; AHRQ behavioral crisis benchmarks.
DEV-CH quality complianceCMS Child Core Set MY2024$450$300–$600 per compliant child range midpoint; Canvas Dx closes the screen-to-diagnosis loop at the PCP, enabling state quality reporting. CMS Child Core Set 2024 (DEV-CH mandatory); Oregon CCO quality metrics benchmark.
EPSDT / OON avoidanceNetwork adequacy compliance$800SCA-obligation reduction when in-network access is established via Canvas Dx at primary care. MHPAEA + EPSDT enforcement benchmarks; Mercer 2024 OON rate analysis. Modeled.
Primary Value Total · per Canvas Dx assessment$13,500Sum of primary sourced and modeled drivers to the state Medicaid agency
▸ WTP ceiling (~20% of value)
$2,700
State Medicaid captures 80% of the value upside; vendor captures 20% for delivery and evidence-generation workstream. Above this threshold, the value equation turns unfavorable for the state.
▸ Target price
$1,000
Conservative at ~7% of Primary Value Total. State Medicaid captures 93% of the upside. Priced for rapid coverage adoption rather than value-share extraction.

▸ Secondary · lifetime cost avoidance ($250,600 per child, modeled)
The 10-year per-child Medicaid obligation delta of $250,600 (Section 06) is a long-horizon consequence of correct diagnosis + right-matched care. It is shown as a modeled estimate — components sourced, composite derived — and is not rolled into the Primary Value Total above to avoid double-counting with the per-assessment drivers. Retained in the brief because the state Medicaid mandate is best-possible lifetime outcomes for Medicaid-enrolled children and their families; the $250,600 is the quantified consequence of acting on the diagnostic gate today.

▸ Primary Sources

Regulatory & enforcement
HHS OIG Audit Series 2019–2026 (IN, WI, ME, CO completed)
FDA De Novo DEN200069 (Canvas Dx clearance, June 2021)
EPSDT 42 U.S.C. § 1396d(r) · MHPAEA enforcement guidance
CMS Child Core Set 2024 (DEV-CH mandatory MY2024)

Clinical evidence
Megerian JT et al. NPJ Digit Med 2022;5(1):57 — pivotal clinical trial
Salomon et al. Nature Scientific Reports 2025;15:29503 — real-world evidence (n=254, 15 states)
Harstad E et al. JAMA Pediatrics 2023 — 37% non-persistence with early dx
Lovaas 1987 — 47% mainstreaming with intensive early intervention
McNally Keehn et al. Pediatrics 2023 — 82% PCP-specialist diagnostic consistency
Catalight Practice Guidelines + Sneed et al. 2023 + Samelson et al. 2025 — ABA dose-response
Cidav Z et al. JAACAP 2017 PMC7007927 — service-utilization offset post-EI

Epidemiology & access
CDC ADDM 2022 — Shaw KA et al. MMWR 2025;74(SS-2), pub. April 2025 · 1:31 prevalence, 49-mo national mean
Mandell et al., Shattuck et al. — Medicaid diagnostic delay disparities
Leslie-Fournier et al. Research in Autism 2025 — 201-day Dx-to-ABA lag
SDBP / ABP 2024 workforce data — ~850 board-certified DBPs nationally

Fiscal & market
Shi L et al. Autism Research & Treatment PMC3534815 — $22,653/yr Medicaid autism cost
Vu M et al. J Manag Care Spec Pharm 2023;29(4):378 — $331/mo all-cause carrying cost
Buescher JAMA Pediatrics 2014 — $1.4M–$2.4M lifetime cost avoidance
HCUP State ED Database 2022 — $12,500/encounter Medicaid FFS psychiatric ED
Distilinfo March 2026; Behavioral Health Business Dec 2025 — $14.49B ABA spend, 25% CAGR
CMS 2024 Autism Infographic · House Energy & Commerce 10-state ABA investigation

▸ The stop-gap and the infrastructure

Canvas Dx is two interventions compressed into one coverage decision. Today: the FDA-cleared diagnostic stop-gap that anchors ABA authorization to a reproducible, auditable record — halting the fraud inflation four OIG audits have now documented. Tomorrow: the longitudinal data infrastructure that makes precision-care pathway design possible at the national scale — stratifying by challenge-score domain, by clinician type, by age band, so that the next generation of Medicaid-enrolled children receives the right service at the right intensity from the start. Every month of delay compounds the national avoidable obligation. Every month matters.

▸ 03 · Interactive ROI calculator

Model your state’s assumption set.

Covered children, autism prevalence, Canvas Dx price, and eight value drivers as user-adjustable inputs → live waterfall and projected return multiple. Ported from the analyst version; every value driver carries its own source and formula. Adjust the population and prevalence to your state; toggle drivers off to see conservative or constrained modeling.

Step 1 — Population & price

▸ Adjust to match your state’s covered children and procurement position

State Medicaid pediatric enrollment estimate, age 18–72 months (~6% of total state population)

1 : 4Autism center
1 : 10Dev. peds
1 : 31General pop.
CDC ADDM 2022 general-population prevalence: 1:31 (pub. April 2025). Determinate rates scale with underlying prevalence.
Canvas Dx price · fixed assumption
$1,000
Per Section 06 WTP derivation — ~7% of the $13,500 Primary Value Total to the state.
▸ Population funnel
Covered children (slider)15,000
1 in 4 at-risk of developmental delay3,750
1:31 autism cohort484
M-CHAT false positives / yr (52%)936
37% remission candidates (dx’d 2–3 yr)179

Step 2 — Value drivers

▸ Toggle off any driver your state’s analysis excludes


▸ Step 3 · Your modeled ROI — live

National Medicaid ROI — $0 addressable value

15,000 covered children · Canvas Dx at $1,000 · 0.0× projected return

Total addressable value
$0
Across selected drivers
FWA prevention alone
$0
The structural argument, before any other driver
Breakeven per child
— mo
Carrying-cost savings vs. Canvas Dx price
Return on investment
0.0×
$1 invested → $0.0 in value
▸ ROI waterfall

▸ Value by driver

Calculator math ported from the analyst version (state_medicaid_asset_1.html) verbatim. State-level projections are modeled estimates; formulas per-driver documented in the Value Matrix embedded below.

▸ Value matrix · expanded detail

Per-assessment value drivers with per-card formula and citation.

Expanded-detail view of the six per-assessment drivers summarized in Tab 01 Section 06 (Primary Value Total → WTP ceiling → target price). The Primary band reconciles line-by-line to the $13,500 Primary Value Total — each card carries the per-case value, formula, and source citation shown in Section 06. A second band captures MCO-level drivers that the state Medicaid agency captures indirectly through capitation rate-setting. A third band documents the clinical mechanisms that produce the Primary per-case magnitudes — shown for audit purposes, not summed into the Primary Value Total.

▸ Primary drivers · state Medicaid agency, per Canvas Dx assessment · sum → $13,500 Primary Value Total
FWA prevention · authorization anchor

Primary · 01

$2,500 per case
Indiana OIG found $56.0M in improper ABA payments (Dec 2024) across ~25,000 active ABA cases/year, implying a $2,240 floor per case. Wisconsin, Maine, and Colorado audits in the same series reach similar orders of magnitude (WI $18.5M, ME $45.6M, CO $77.8M). The root cause every OIG report cites is the diagnostic-anchor gap: no defensible, independent Dx record to justify ABA authorization. Canvas Dx installs the authorization anchor before improper billing starts — an FDA-cleared, algorithm-locked Dx tied to the authorization, independent of referral relationships and geography. Enforcement mechanisms against over-billing, phantom-service, and caseload-stacking remain; Canvas Dx closes the diagnostic-integrity gap those enforcement mechanisms repeatedly surface downstream.
Four-audit improper-payment average ÷ active ABA cases/yr = per-case floor → $2,500 target
HHS OIG Audit Series 2019–2026 (IN A-05-22-00025, WI A-05-21-00020, ME A-01-22-00501, CO A-07-22-02821).
Neuroplasticity NPV · lifetime dependency reduction

Primary · 02

$5,000 per case · modeled
The largest long-horizon driver. Buescher et al. JAMA Pediatrics 2014 establishes the $1.4M lifetime Medicaid cost per autism-without-ID child and $2.4M with ID. A 5% reduction in lifetime dependency — supported by Harstad 2023 (37% non-persistence at early dx) and Cidav 2017 (service-utilization offset post-EI) — produces a $70K undiscounted lifetime savings per early-dx child. Discounted to present value at 7% over a 30-year horizon yields ~$5,000 per assessment. Accrues to the Medicaid agency because the Medicaid program carries the lifetime obligation for enrolled children. Modeled: components sourced, composite derived.
$1.4M lifetime × 5% dependency reduction × NPV@7%, 30-yr → ~$5,000 per assessment
Buescher et al. JAMA Pediatrics 2014 ($1.4M–$2.4M lifetime); Harstad JAMA Pediatrics 2023 (37% non-persistence); Cidav JAACAP 2017 (service-utilization offset). Modeled.
Specialist capacity reallocation

Primary · 03

$1,000 per case · modeled
Today’s specialist evaluation is 6–8 hours of billable specialist time plus family commute to a specialty center. Canvas Dx delivers a DSM-5 Dx report at the pediatrician in ~1 hour of billable clinician time — the state’s developmental-specialist workforce is freed to evaluate the Indeterminate cohort (18.77% Mid + 13.22% Low + 4.82% High, per Salomon 2025) where clinical nuance modifies the Canvas Dx output. Monetizes the redirected specialist time at loaded hourly cost against SDBP 2024 workforce benchmarks. Modeled: workforce data × loaded hourly cost × redirected-hours per Canvas Dx assessment.
Hours redirected × loaded specialist hourly cost → per-assessment value
SDBP 2024 workforce data; Megerian NPJ Digital Medicine 2022 (~1 hr Canvas Dx clinician time, telehealth-validated); specialist-evaluation time benchmarks. Modeled.
ED / crisis avoidance

Primary · 04

$3,750 per case
Undiagnosed and incorrectly-routed children present to the Medicaid-financed emergency department at higher rates for behavioral crisis events. HCUP state ED database 2022 shows ~0.3 Medicaid psychiatric ED encounters per under-6 autism-cohort child over a three-year window; per-encounter Medicaid FFS cost at ~$12,500. Canvas Dx + right-matched early intervention reduces crisis rates by establishing stepped-care before the behavioral-escalation pattern embeds. 0.3 × $12,500 = $3,750 per case.
0.3 psychiatric ED encounters avoided × $12,500 Medicaid FFS ED cost → $3,750 per case
HCUP State ED Database 2022; AHRQ behavioral crisis benchmarks.
DEV-CH quality compliance

Primary · 05

$450 per case
CMS Child Core Set 2024 made DEV-CH mandatory for measurement year 2024 — states are measured on developmental screening completion for children under three. Canvas Dx closes both halves of the loop (screen + documented Dx outcome) at the pediatrician, in the same visit. 41% of screen-positive children never reach formal evaluation under the specialist-only pathway (Monteiro et al. Pediatrics 2019); the loop closes at the PCP with Canvas Dx. Per-compliant-child value sits in a $300–$600 range across state programs (Oregon CCO quality metrics, DHCS CalAIM quality withhold); midpoint $450.
$300–$600 midpoint × DEV-CH-compliant children × deployment rate
CMS Child Core Set 2024 (DEV-CH mandatory MY2024); Monteiro et al. Pediatrics 2019 (41% unevaluated); Oregon CCO quality metrics 2024; DHCS CalAIM quality withhold.
EPSDT / OON avoidance (state-captured)

Primary · 06

$800 per case · modeled
Under EPSDT (42 U.S.C. § 1396d(r)) the state is obligated to deliver timely diagnostic and treatment services to Medicaid-enrolled children. Network-adequacy failures cascade into out-of-network single-case agreement obligations — typically $4K–$7.5K per specialist eval. Canvas Dx at the pediatrician establishes documented in-network access, removing the OON trigger at the state level. The state-captured share (net of MCO-captured benefit, which is accounted for in the Secondary band) is modeled at $800 per case.
OON trigger rate × state-captured share × $4K–$7.5K SCA rate → $800 per case
EPSDT 42 U.S.C. § 1396d(r); Mercer 2024 (OON SCA benchmarks); MHPAEA enforcement guidance. Modeled.
▸ Secondary drivers · MCO-level, captured by state through capitation rate-setting · not additive to $13,500
Risk adjustment revenue · HCC/RAF

Secondary · MCO

+$3.20 PMPM per coded child
Under CMS HCC102 (autism) methodology, accurate autism coding produces +$3.20 PMPM per diagnosed child for Medicaid MCOs operating under risk-adjusted contracts. Recurring revenue, not one-time. A 4,000-child cohort generates ~$153,600/year capitated uplift. The diagnosis must be documented through a defensible record — Canvas Dx is engineered to produce that record. Captured by the state indirectly through capitation rate-setting pressure; not additive to the state’s Primary Value Total.
$3.20 PMPM × 12 × autism cohort (coded) = annual capitated uplift
CMS HCC RAF methodology; HCC102 autism code benchmarks; MCO capitation contract structures.
MCO network-adequacy SCA avoidance

Secondary · MCO

$4K–$7.5K per OON case avoided
Applies directly to Medicaid managed care contracts, not to state FFS. Under MHPAEA (which governs Medicaid MCOs and CHIP) and EPSDT, when no in-network autism evaluation is available within adequate time or distance, the MCO is obligated to approve an out-of-network single-case agreement — typically $4K–$7.5K per evaluation (Mercer 2024). Canvas Dx at the pediatrician removes the OON trigger. The state captures this benefit indirectly, through capitation rate-setting on MCO bids. The state-captured portion is already reflected in Primary Driver 06 ($800); the full MCO-level magnitude is reported here for completeness.
OON utilization rate × MCO autism cohort × $4K–$7.5K SCA rate
Mercer 2024 (1.1M+ covered lives); Title XIX EPSDT 42 U.S.C. § 1396d(r); MHPAEA enforcement guidance (applicable to Medicaid managed care and CHIP); SCA rate arbitration benchmarks.
▸ Mechanism · how Canvas Dx produces the Primary magnitudes · not additive; shown for audit
ABA routing precision

Mechanism

~$62,400/yr avoided per correctly Negative child
Canvas Dx output distribution (Salomon et al. Nature Scientific Reports 2025, n=254, 15 states): Positive 42.23% · Negative 20.95% · Mid Ind 18.77% · Low Ind 13.22% · High Ind 4.82%. The 20.95% Negative rate is the structural FWA correction: each child correctly cleared by algorithmic Negative avoids ~$62,400/yr in unnecessary ABA authorization (10 hrs/wk × $120/hr × 52 wk). On Indeterminate reports, the diagnosing clinician still determines next steps of care — stepped speech/OT or specialist bridging — leveraging the auto-populated DSM-5-mapped documentation report, and not defaulting to ABA. This is the clinical mechanism underlying Primary Driver 01 (FWA prevention); the $62,400/yr is not additive to the $2,500/case Primary figure, which is the OIG audit-grounded floor.
20.95% Negative × age-range cohort × $62,400/yr avoided ABA authorization
Salomon et al. Nature Scientific Reports 2025 (n=254 prescriptions, 15 states); Catalight Practice Guidelines 2024 (ABA rate benchmarks).
M-CHAT queue clearance

Mechanism

~$466K/yr per 15K-member cohort
M-CHAT-R/F has 47.5% PPV (Robins et al. Pediatrics 2014, n=16,071) — 52% of screen-positives are false positives. Under the current pathway, all screen-positives enter the specialist queue. Canvas Dx’s 97.6% NPV resolves false positives at the pediatrician in 48–72 hours, clearing the queue structurally rather than waiting it out. Per 15,000-member cohort: ~449 M-CHAT positives/year → 233 false positives → $466K+ in avoided specialist evaluations annually. This is the clinical mechanism underlying Primary Driver 03 (specialist capacity reallocation); the $466K/yr is how the redirection happens and is not additive to Driver 03.
(cohort × M-CHAT positive rate × 0.525 FP rate) × $2K specialist eval avoided
Robins et al. Pediatrics 2014 (M-CHAT-R/F, n=16,071, PPV 47.5%); Salomon Nature Sci Rep 2025 (Canvas Dx 97.6% NPV).
▸ Cross-budget federal leverage · not a Medicaid driver

Early autism identification also triggers IDEA Part C (birth-to-3 early intervention) and Part B (special education) federal allocations that flow through the state Department of Public Instruction — approximately $10,041–$14,940 per eligible child (NC DPI Developmental Day proxy). These dollars do not reduce Medicaid spend directly; they accrue to the state education budget. Included here for completeness: the state captures more federal funding in aggregate when children are identified earlier, even though the Medicaid line item is not the beneficiary.

FMAP reminder: realized state share of savings is subject to the Federal Medical Assistance Percentage match rate, which varies by state. Figures shown are total-dollar estimates prior to FMAP apportionment. Primary drivers reconcile to Section 06 line-by-line. Secondary and Mechanism drivers are reported separately and are not summed into the $13,500 Primary Value Total.

▸ 02 · State detail · 51-jurisdiction drill-down

Every state. Same structural pattern, different numbers.

The state table below is the full 51-jurisdiction view. To model a specific state’s ROI or generate a state one-pager, continue to the Tab 03 ROI calculator and Tab 04 one-pager, which carry the state selector. Population estimates from U.S. Census 2020; autism cohort from CDC ADDM 2022 prevalence (1:31); state-level ABA spend and CAGR from Distilinfo March 2026 and state-level Medicaid reporting where published. Savings potential is modeled (autism cohort × 37% Harstad non-persistence × $250,600 per-child 10-yr obligation delta), contingent on VBC delivery model adoption.

State—OIGstatusAvg Dx agemodeledAge reductionvs. 36-mo targetChildren18–72 moAutism1:31 prev.1-yr / autism childmodeledCohort 1-yrmodeledABA spendMedicaidABA CAGR2018–2410-yravoidable

10-year avoidable obligation = autism cohort × 37% non-persistence × $250,600 per-child modeled delta. Assumes VBC delivery model. ABA spend figures are most-recent-reported state-level Medicaid ABA expenditure (Distilinfo March 2026 + state sources); CAGR calculated from 2018 baseline where available. All state-level projections are modeled estimates.

Dec 2024

Indiana · $56.0M improper

Improper payments in 100% of sampled enrollee-months. Findings include services not rendered, hours billed exceeding what is physically possible in a day, caseloads stacked on single providers, and authorization without confirmed diagnosis.

Jul 2025

Wisconsin · $18.5M improper

Same structural findings. Lower absolute dollar figure reflecting state-level program size; the 100% enrollee-month finding is consistent.

Jan 2026

Maine · $45.6M improper

Findings pattern consistent. By this audit, the root-cause language in the OIG report becomes explicit about diagnostic-anchor integrity.

Mar 2026

Colorado · $77.8M improper

Largest single-state finding to date. Four audits now complete; three audits remaining in the HHS OIG seven-state series.

Pending

Three audits remaining · timing disclosed on OIG work plan

Target states not publicly named; OIG work plan lists the seven-state series as ongoing. House Energy & Commerce has flagged Nebraska (+1,800% Medicaid ABA growth) as a candidate for state-level IG action.

▸ National Medicaid ABA spend trajectory

$3.82B (2018) → $14.49B (2024) — +279%.

Distilinfo March 2026 and Behavioral Health Business Dec 2025. A +25% CAGR that compounds at roughly 3× the growth rate of total US healthcare (7.2% CMS). Per Node 05 in the brief (Tab 01), the growth signal is understood as fraud inflation on children already in the system, not the undiagnosed population finally being captured.

2018 spend
$3.82B
Baseline
2024 spend
$14.49B
Current year
Growth multiple
3.8×
+279% in 6 years
Annual CAGR
25%
3.5× total US healthcare

Sources: Distilinfo March 2026; Behavioral Health Business December 2025; CMS National Health Expenditure data; House Energy & Commerce ABA investigation.

▸ Printable one-pager · state-specific

One-page summary, populated by state selector.

Pick a state in the selector below and the one-pager populates with that state’s autism cohort, ABA spend, and modeled avoidable obligation. Print-ready for email attachment, governor’s briefing, or legislative handoff.

Select a state in Tab 03 State Detail to populate this one-pager.