AppGrid AI: Healthcare reference data
AppGrid AI answers healthcare questions in plain English by pairing your own Salesforce records with more than 30 published healthcare reference sources. Each source teaches the AI something your records don't spell out: what a code means, which group it belongs to, how a measure is defined, or what a fair price is. Below, every source, what it adds, and one question it makes possible.
How the data reaches your org
All reference data lives inside your own Salesforce org — no outside service holds it. It arrives in one of three ways:
Included and loaded automatically. Public sources ship with AppGrid and load when you install or upgrade. Nothing to download, nothing to pick.
Fetched for your own records. A few registries are read from the publisher's public service for just the codes and providers your org uses, and refreshed weekly.
Your own licence. Licensed content such as CPT or HEDIS is loaded from your organization's own subscription. AppGrid never ships it.
The Catalog tab in AppGrid AI. Public sources show as packaged and loaded automatically; licensed sources such as CPT wait for your own licence.
Diagnosis, procedure, lab and unit codes
Codes are how health data is recorded. These sources turn a code into words, and group thousands of codes into chapters a person would name.
ICD-10-CM (CMS / CDC) — every diagnosis code, its description, and its chapter and block.
"Which diagnosis chapters drove the most inpatient admissions this year?"
ICD-10-PCS (CMS) — inpatient procedure codes, grouped by body system and procedure type.
"How many joint replacement procedures did we pay for in Q3?"
HCPCS Level II (CMS) — codes for supplies, equipment, drugs and services beyond CPT.
"What did we spend on durable medical equipment last quarter?"
LOINC (Regenstrief, optional) — lab tests, vital signs and assessments.
"Which lab tests were ordered most often last month?"
SNOMED CT (NLM, optional) — clinical concepts and their synonyms, for records coded from an EHR.
"How many patients have a recorded history of heart failure?"
UCUM (Regenstrief) — units of measure.
"Which glucose results were recorded in a unit other than mg/dL?"
Disease groupings and risk
Analysts rarely ask about a single code. These sources define the conditions and categories they actually mean.
AHRQ CCSR and Elixhauser — clinical categories for every diagnosis and procedure, plus comorbidity flags.
"What are the top 10 disease categories by inpatient cost?"
CMS Chronic Conditions Warehouse — standard definitions of 30 chronic conditions.
"How many members have chronic kidney disease?"
CMS-HCC risk models — the map from diagnoses to risk categories, with their hierarchies and weights.
"Which members carry the most risk-adjusting conditions?"
Quality measures
Quality programs define exactly which patients, codes and time windows count. These sources carry those definitions, so the AI counts the way CMS does.
VSAC value sets (NLM) — the code lists behind each quality measure: which diagnoses, drugs, tests and procedures qualify.
"How many diabetic members had an eye exam this year?"
CMS eCQM specifications — each electronic measure's population, timing, exclusions and exceptions.
"What is our controlling high blood pressure rate for this year?"
AHRQ Prevention Quality Indicators — admissions that good outpatient care can prevent.
"How many asthma admissions could have been avoided last year?"
CMS QPP / MIPS — clinician quality measures and their national benchmarks.
"Which MIPS measures are our clinicians furthest below benchmark on?"
Medicare Star Ratings — Part C and D measures, star cut points and published results.
"Which Star measures are closest to the next star threshold?"
Claims and payment rules
Payment questions need Medicare's published rules beside your own claims.
MS-DRG (CMS) — inpatient groups and their relative weights.
"What was our average case-mix index by hospital last quarter?"
CMS fee schedules and RVUs — Medicare rates and relative values by locality.
"How do our allowed amounts for office visits compare with Medicare rates?"
CMS NCCI edits — procedure pairs that should not be billed together, and unit limits.
"Which claims billed two procedures that should not appear together?"
Medicare Coverage Database — national coverage decisions and the codes they cover.
"Which lab claims lack a diagnosis that supports coverage?"
CMS Place of Service — where each service was given.
"What share of visits were telehealth last quarter?"
Drugs and vaccines
A pharmacy claim carries an NDC — the exact product dispensed. These sources say what that product is and which class it belongs to.
FDA NDC Directory (fetched for your NDCs) — each product's brand or generic status, ingredients, strength, labeler and DEA schedule.
"What is our generic dispensing rate this year?"
RxClass (NLM, fetched) — drug classes such as statins, GLP-1 agonists and ACE inhibitors.
"What did we spend on GLP-1 drugs last quarter?"
RxNorm (NLM) — standard drug names, ingredients and strengths.
"Which ingredients account for the most pharmacy spend?"
DailyMed (NLM, optional) — current drug labeling for each product.
"Which of our top-dispensed drugs had a label update this year?"
CDC vaccine codes (CVX, MVX) — vaccines and their manufacturers.
"How many members 65 and older got a flu shot this season?"
Each organization fetches FDA data with its own free openFDA key.
Providers, facilities and communities
These sources describe who delivered care and the places your members live.
NPPES NPI Registry (CMS, fetched) — each of your providers' name, location and primary specialty.
"Which in-network specialists had the most referrals last quarter?"
NUCC provider taxonomy — the standard list of provider specialties.
"How is our spend split across provider specialties?"
CMS Care Compare — published hospital ratings and readmission results.
"How do our readmission rates compare with each hospital's published rate?"
CDC PLACES — community health estimates by ZIP code.
"Which ZIP codes have the highest local diabetes rates among our members' areas?"
CDC/ATSDR Social Vulnerability Index — social vulnerability by ZIP code and county.
"Do members in high-vulnerability areas have more ER visits?"
Census American Community Survey — income, age, insurance and other demographics by ZIP code.
"What share of our members live in areas where median income is below $40,000?"
Sources you bring under your own licence
Some standards are licensed by their publishers. AppGrid reads them when your organization loads its own copy, and never ships them.
CPT (AMA) — procedure and service codes for professional claims.
"Which office visit levels did our clinicians bill most last month?"
X12 claim adjustment and remark codes — what each denial or adjustment code means.
"What are our top five denial reasons by dollars this quarter?"
NUBC UB-04 codes — revenue codes, bill types and discharge status.
"What share of inpatient stays ended in a discharge to skilled nursing?"
NCQA HEDIS — health-plan quality measures and their value sets.
"What is our breast cancer screening rate against HEDIS?"
PQA measures — medication adherence and safety measures.
"What share of members on statins were adherent this year?"
Medi-Span — drug products and the interactions they flag.
"How many members fill two drugs flagged as interacting?"