Red/Blue Referencing Report for Medicaid AI Adoption

Here’s a source-validation pass on the research report you provided (AI Adoption challenges in Medicaid Settings). I pulled every salient number or key factual claim from the report, tracked the report’s own citations, and then followed those to confirm (or not) the figures. For each item, I note whether the report itself cites a source, whether that source actually supports the fact, and whether the best supporting source is primary (original data/official doc/peer-reviewed paper) or secondary (press, blogs, vendor case studies, news, etc.). Where the report relied only on a secondary source, I looked for a better primary source.

Key numbers & facts — sourcing audit

#	Claim (as written in the report)	Where it appears in the report	Did the report cite a source?	What the cited (or best) source says	Primary or secondary?	Verdict
1	“Medicaid covers 70–80 million Americans, roughly 1 in 5 U.S. residents, including about 40% of all children.”	Exec. summary (p.1)	Yes: 1	AHA Fact Sheet (Feb 2025) states Medicaid covers “one in five Americans — more than 70 million people — including 40% of all children.” (American Hospital Association)	Secondary (advocacy fact sheet summarizing federal data)	Supported (wording & figures align).
2	“In 2023, U.S. hospitals reported a $27.5B Medicaid shortfall (payments vs. costs).”	p.1 and p.2	Yes: 3/11	AHA Fact Sheet (Feb 2025) explicitly reports $27.5B Medicaid shortfall in 2023. (American Hospital Association)	Secondary (AHA analysis of cost-report data)	Supported (matches AHA).
3	“There are approximately 300 safety-net hospitals and more than 4,000 Rural Health Clinics (RHCs).”	p.2	Yes: 10	America’s Essential Hospitals says it represents “more than 300” essential (safety-net) hospitals. (America's Essential Hospitals) CMS (July 2025) says there are about 5,200 RHCs (not ~4,000). (Centers for Medicare & Medicaid Services)	AEH: Secondary; CMS MLN: Primary	Partially supported — “300” is fine; RHC count is outdated (best current figure ≈5,200).
4	“FQHCs: ~1,375 organizations serving ~29M patients annually, about half on Medicaid.”	p.2	Yes: 10	HRSA/UDS shows 1,359 awardees (2024 table), and NACHC’s 2024 roll-up shows 1,512 CHCs serving ~34M patients. (HRSA Data)	HRSA UDS/NACHC: Primary (UDS); NACHC: Secondary summarizing UDS	Partially supported — counts have grown; 29M and 1,375 look like older UDS vintages.
5	“As of 2024, the U.S. had ~161,000 primary-care offices and 314,000 specialty offices (~475,000 total).”	p.2	Yes: 3	Closest public reference is a market report (Research & Markets) estimating 161k primary and 314k specialty offices (2024). (Research and Markets)	Secondary (commercial market report)	Supported (secondary only) — No open primary dataset found with these exact splits; NAICS pages don’t match these counts. (SICCODE.com)
6	“A 2023 Commonwealth Fund analysis found one-third of primary-care practices account for 90% of Medicaid office visits.”	p.2	Yes: 3	I couldn’t find a Commonwealth Fund analysis with this 1/3 → 90% concentration statistic.	—	Not verified — needs an exact citation; please provide the specific Commonwealth Fund piece to validate.
7	“Abridge raised $300M Series E at $5.3B valuation; deployed in 150+ health systems.”	p.3	Yes: [19][20]	Reuters (June 24, 2025) reports $300M Series E at $5.3B valuation. (SHADAC)	Secondary (news)	Supported (funding/valuation). Deployment count (150+) not verified in Reuters; needs company/press source.
8	“Suki demonstrated a 72% reduction in documentation time, saving >3 hours/week.”	p.3	Yes: [21]	AAFP News (Dec 15, 2021) press item claims 72% time reduction and ~3.3 hours/week saved. (HRSA Data)	Secondary (press article referencing internal/user study)	Supported (secondary) — no peer-reviewed primary provided.
9	“Notable (Montage Health) cut referral processing time from 23 days to <2 days, saving 1,200+ staff hours.”	p.3	Yes: [22][23]	Notable customer story reports 23 → <2 days and 1,200+ hours saved. (TechCrunch)	Secondary (vendor case study)	Supported (secondary) — no independent evaluation found.
10	“Waymark Signal (study of 14M Medicaid members) improved care-gap closure by ~35 percentage points.”	p.3	Yes: [25][26]	Press (Accesswire) says npj Digital Medicine study found 85% accuracy and large closure gains; specific “35 percentage points” appears in coverage, not in an accessible peer-reviewed article I could open. (American Medical Association)	Secondary (press; underlying npj paper would be primary but not readily accessible from link provided)	Partially supported — need the actual npj Digital Medicine article or preprint to confirm effect size.
11	“FDA-cleared IDx-DR improved screening in a Sutter Health pilot; integrated into primary care.”	p.3	Yes: [29][30]	AMA News feature (June 23, 2025) describes Sutter’s primary-care workflow using AI retinal imaging; reports improved screening. (NTIA)	Secondary (trade news); the primary would be Sutter or peer-reviewed data	Supported (secondary) — seek a Sutter publication or program evaluation for primary.
12	“12–13% of Americans have no internet at home; 15% are smartphone-only.”	p.4–5	Yes: [37][38]	NTIA (2023 survey) FAQ: 12% of people live in households with no internet; 26.2% lack wired broadband. (BroadbandUSA) Pew (Jan 31, 2024) reports 15% “smartphone-only” adults. (Pew Research Center)	NTIA: Primary; Pew: Primary survey	Supported.
13	“Teladoc Health & Cleveland Clinic SMS-first AI triage bot for rural FQHCs; pilot reduced no-shows and enabled pre-visit symptom review.”	p.4–5 (mini-case)	Yes: [39][40][41][42]	I could not find a Teladoc–Cleveland Clinic SMS triage pilot case study. Cleveland Clinic has a “one-click nurse triage schedule” article; Teladoc has a general health-equity case brief, but not this SMS bot pilot. (Cleveland Clinic)	—	Not verified — appears unsourced/untraceable as written. Needs a concrete case study or press release.
14	“ThriveLink uses a voice-activated AI agent so patients without internet/smartphones can get reminders & complete applications.”	p.4–5	Yes: [40][43]	MIT Solve profile describes ThriveLink’s telephonic AI agents enabling application completion without internet/literacy, consistent with voice-based access. (MIT Solve)	Secondary (competition profile / company materials)	Supported (secondary) — primary would be a deployment evaluation by a health system or public agency.
15	“Nationally, EHR adoption rates are high (over 95% of FQHCs and nearly all public hospitals).”	p.7	Yes: [47][48][49]	ONC reports 96% of non-federal acute care hospitals have certified EHRs (2021, and stable since 2015-17). (HealthIT.gov) Evidence for “over 95% of FQHCs” is older/patchy; HRSA UDS shows near-universal EHR presence but doesn’t publish a single percentage line; older ONC briefs showed high FQHC adoption but not 95%+ in the latest year. (HRSA Data)	ONC: Primary; FQHC %: unclear	Partially supported — hospitals yes; FQHC “>95%” needs a current HRSA/ONC statistic or explicit UDS metric.
16	“~15,000 FQHC sites; many in broadband-limited areas.”	p.4 & p.5	Yes: [35]	Latest NACHC/UDS suggests >17,000 delivery sites (2024). (NACHC)	Secondary (NACHC citing UDS)	Partially supported — “15,000” is low vs. 2024 figures.
17	“Roughly 53,000 Medicaid-serving private practices.”	p.2 & p.5 (derived in text)	No (derived)	This is an internal calculation (from item #5 plus the ‘one-third’ concentration claim). The 1/3→90% premise is unverified (see #6).	—	Not supported — treat as an unsourced estimate until #5 and #6 are pinned down.

Notes on gaps, corrections, and better sources

Medicaid coverage basics (items 1–2) are well-supported by the AHA’s 2025 fact sheet, which itself compiles federal program data. For strict primary documentation on “1 in 5” and “40% of children,” the MACPAC enrollment dashboards and Census/ACS insurance tables can also corroborate; the report cites MACPAC for enrollee composition. (American Hospital Association)
RHC count: CMS’s July 2025 briefing lists ~5,200 RHCs, so the report’s “>4,000” should be updated. (Centers for Medicare & Medicaid Services)
FQHC counts & patients: HRSA UDS (2024) shows 1,359 awardees; NACHC’s “By the Numbers” (2024) shows 1,512 CHCs, ~34M patients, >17,000 sites — more current than “1,375 / 29M / ~15,000.” Recommend updating all three. (HRSA Data)
Physician office counts (item 5): only secondary market research publicly states 161k primary-care and 314k specialty offices. NAICS portals don’t provide these figures, and Census County Business Patterns enumerates establishments but not with a simple primary vs. specialty split. Unless you can surface a Census/CBP or AMA Physician Practice Benchmark table with those counts, keep this labeled secondary. (Research and Markets)
Commonwealth Fund “1/3 practices → 90% of Medicaid visits” (item 6): I could not locate this analysis; please add a direct link or reconsider the statistic. — Action needed.
Abridge / Suki / Notable: funding and case-study performance claims are supported only by secondary or vendor sources. They should remain clearly labeled as such in the report. (SHADAC)
Waymark (item 10): Press cites a npj Digital Medicine study covering 14M members; to anchor the “35 percentage-point” effect, link the actual npj article (DOI) or preprint. I didn’t find the open paper via the press links provided. (American Medical Association)
IDx-DR @ Sutter (item 11): AMA News is a useful overview, but a program report or peer-reviewed evaluation from Sutter would be the appropriate primary confirmation. (NTIA)
Connectivity stats (item 12): These are current and primary (NTIA 2023; Pew 2023 fieldwork published 2024). Good. (BroadbandUSA)
Teladoc–Cleveland Clinic SMS triage (item 13): I found no Teladoc+CC SMS case study matching the description. Cleveland Clinic has a nurse-triage scheduling workflow article (not SMS/AI), and Teladoc has a general equity case brief. Unless a specific pilot report exists, this section should be removed or re-sourced. (Cleveland Clinic)
ThriveLink voice agent (item 14): Multiple secondary sources (MIT Solve profiles & company pages) confirm a telephonic AI enrollment/engagement approach that works without internet/smartphones. For stronger evidence, add a deployment evaluation by a public agency/health system partner if available. (MIT Solve)
EHR adoption (item 15): The hospital figure is strongly supported by ONC (96% adoption). The FQHC “>95%” needs a current UDS table or an ONC/FQHC-specific brief; otherwise rephrase to “near-universal EHR adoption among FQHCs” with a UDS hit-capabilities table citation. (HealthIT.gov)

Primary replacements (or best alternates) for secondary-only items

Item 5 (physician office counts) – No authoritative public primary with that exact split. If you keep it, label as market estimate and consider substituting with AMA/CDC high-level counts if accessible behind paywalls. Otherwise, cite Census County Business Patterns with methodology caveats (but it doesn’t cleanly split primary vs. specialty). (SICCODE.com)
Item 8 (Suki 72%) – Primary would be a peer-reviewed or methodology white paper; AAFP News is secondary. If unavailable, keep AAFP press and add any independent evaluation if found later. (HRSA Data)
Item 9 (Notable/Montage) – Vendor case study is secondary. See if Montage Health or a third-party analytics partner published a methods/results report; otherwise, keep as a case study clearly labeled. (TechCrunch)
Item 10 (Waymark) – Replace press with the npj Digital Medicine article (DOI) once located; until then, mark as secondary. (American Medical Association)
Item 11 (IDx-DR @ Sutter) – Add a Sutter Health primary program page or an academic abstract from its pilot; AMA News remains secondary. (NTIA)
Item 13 (Teladoc–Cleveland SMS) – Currently unverified. If this pilot exists, it should have a Teladoc or Cleveland Clinic case study or conference abstract. Otherwise, remove or switch to a documented example (e.g., state Medicaid chatbots in LA and AZ during renewals). (AAFP)

Items explicitly unsourced or untraceable in the report (action required)

“One-third of primary-care practices account for 90% of Medicaid office visits” — no Commonwealth Fund source found. Provide the exact citation or revise.
Teladoc–Cleveland Clinic SMS triage pilot — could not locate any confirmatory case study, press release, or peer-reviewed abstract. Replace or remove.
Derived figure: ~53,000 Medicaid-serving practices — relies on #5 and #6; until both are verified, treat as an internal estimate (and label as such).

Minor tightening recommendations for the report

Refresh provider counts to 2024/2025 (FQHC orgs/sites/patients; RHC count). (HRSA Data)
Differentiate primary vs. secondary in footnotes: AHA/ONC/HRSA/NTIA/Pew are closest to primary; vendor case studies, press, and trade outlets are secondary.
Where vendor outcomes are used (Notable, Suki, Nabla, Waymark), clearly label as case study/press claims, and (when possible) add a peer-reviewed or third-party corroboration. (HRSA Data)
For EHR adoption among FQHCs, either (a) cite a current UDS HIT table explicitly stating the % with EHR, or (b) soften to “near-universal” with a general UDS HIT citation. (HRSA Data)

References I used to verify claims (selection)

AHA Fact Sheet: Medicaid (Feb 2025): enrollment share, children share, $27.5B shortfall. (American Hospital Association)
America’s Essential Hospitals / AEH: >300 essential hospitals. (America's Essential Hospitals)
CMS (MLN RHC brief, July 2025): ~5,200 RHCs. (Centers for Medicare & Medicaid Services)
HRSA UDS Table (HIT/EHR) & Data: awardees, HIT tables. (HRSA Data)
NACHC “By the Numbers” (2024): 1,512 CHCs, >17,000 sites, ~34M patients. (NACHC)
Reuters (June 24, 2025): Abridge $300M / $5.3B. (SHADAC)
AAFP News (Dec 15, 2021): Suki 72% reduction / 3.3 hr/wk. (HRSA Data)
Notable x Montage case study: 23 → <2 days, 1,200+ hours. (TechCrunch)
AMA News (June 23, 2025): Sutter Health IDx-DR pilot overview. (NTIA)
NTIA (2023 Internet Use Survey FAQ): 12% with no internet at home; wired broadband gap. (BroadbandUSA)
Pew (Jan 31, 2024): 15% smartphone-only adults. (Pew Research Center)
ONC (EHR adoption quickstats): 96% of non-federal acute-care hospitals. (HealthIT.gov)

Excerpts from your report for context

Medicaid program scale & shortfall figures (p.1):
Safety-net/RHC/FQHC counts (p.2):
Vendor outcomes (p.3):
Connectivity/digital divide stats (p.4–5):
Teladoc–Cleveland Clinic SMS triage (p.4–5):
EHR adoption statement (p.7):

Bottom line

The report’s Medicaid scale and financial pressure claims are solid.
Several counts are stale (FQHC orgs/sites/patients; RHCs). Update to 2024–2025 figures.
A few eye-catching performance stats (Suki, Notable, Waymark) rely on vendor or press; keep them, but label secondary and add primary evaluations when available.
Two claims need immediate attention: the Commonwealth Fund “1/3 → 90%” concentration statistic (source missing) and the Teladoc–Cleveland Clinic SMS triage pilot (no traceable source). Consider revising/removing unless proper documentation is found.