Conversational Analytics
Ask Your Data Like You Are Talking to a Colleague
Vizier understands clinical language. ICD-10 codes, CPT codes, quality measure terminology, payer names, provider NPIs — type the question the way you think it and get a chart or table within seconds. No SQL. No pivot tables. No BI analyst required.
The Dashboard Problem
Dashboards Answer the Questions Someone Thought to Ask Three Months Ago
Every dashboard was designed for a question that made sense when it was built. The moment your focus shifts — a new quality program, a payer contract change, a CMS rule update — the dashboard you have does not answer your new question. You log a request with IT. Six weeks later a new panel appears. By then the question has changed again.
Conversational analytics inverts this entirely. You ask the question you have right now, using the language you use in clinical meetings. The system answers it. If the answer raises another question, you ask that one too. There is no dashboard to maintain, no IT ticket to file, and no delay between a question forming in your mind and an answer appearing on your screen.
Dashboards are not inherently wrong. Vizier lets you save any query result as a persistent view — a "living dashboard" that reruns automatically on each data upload. The difference is that these views emerge from real questions rather than being pre-built by analysts who anticipated what you might eventually want to see.
The typical dashboard lifecycle:
Quality director asks a question
Day 0IT logs the dashboard request
Day 1Analyst scopes the data sources
Day 7Draft dashboard sent for review
Day 28Revisions and approval cycle
Day 42Dashboard published
Day 56Original question no longer relevant
Day 60With Vizier: Day 0 → Answer in 8 seconds.
Real Query Examples
Twelve Queries. Twelve Answers. No SQL Written.
Each example below shows the query as typed, the result type Vizier returns, and the clinical context that makes the answer meaningful. All examples reflect real questions submitted by clinical quality teams.
"What's our 30-day readmission rate for diabetic patients?"
21.3% — 94 of 441 diabetic patient discharges readmitted within 30 days. CMS national average: 17.8%. Patient list available for care management follow-up.
Directly relevant to the HRRP (Hospital Readmissions Reduction Program) under ACA §3025. Excess readmissions can trigger CMS payment reductions of up to 3% on Medicare base payments.
"Show A1C trends for patients over 65 in the last 12 months"
Mean A1C across 318 qualifying patients fell from 8.4 (Jan) to 7.6 (Dec). 28% remain above 9.0. Month-by-month trend line shown with volume overlay.
Relevant to HEDIS measure HbA1c Control for Patients with Diabetes (CDC). A1C above 9.0 is the NCQA threshold for poor control.
"Which providers have the highest no-show rates this quarter?"
Dr. Okafor: 34% (89/261). Dr. Bergmann: 29% (71/245). Dr. Chaudhry: 26% (58/223). Practice mean: 19%. National primary care benchmark: 18–23%.
Each percentage point of no-show rate above 20% represents approximately $8,400 in lost revenue per full-time provider annually at average E&M reimbursement.
"What's our MIPS score as of today?"
Estimated composite MIPS score: 67.4 out of 100. Quality: 42.1/45. Promoting Interoperability: 12/25. Improvement Activities: 13.4/15. Cost: not yet calculated. Score as of last data upload Feb 14.
A MIPS score below 75 results in a payment penalty; above 85 qualifies for a positive adjustment. CY 2024 exceptional performance threshold is 89.
"Show denied claims by payer this quarter"
UnitedHealthcare: 14.2% denial rate ($218,400 denied). Aetna: 11.8% ($164,200). BCBS: 9.1% ($142,600). Medicare: 4.3% ($87,100). Medicaid: 16.1% ($93,400).
Medicaid's 16.1% denial rate is significantly above the 5–7% industry benchmark. Top denial codes shown: CO-4 (incomplete modifier), CO-97 (not separately payable).
"Which diagnosis codes have the longest average length of stay?"
DRG 291 (Heart Failure): 6.8 days vs. 4.7-day geometric mean (+2.1). DRG 871 (Septicemia): 8.2 days vs. 6.1-day geometric mean (+2.1). DRG 194 (COPD): 5.4 days vs. 3.8-day geometric mean (+1.6).
Length of stay above geometric mean affects CMI-adjusted reimbursement under MS-DRG. Excess days are not fully compensated, representing direct margin compression.
"How many patients are due for an Annual Wellness Visit this month?"
387 Medicare patients due for AWV (CPT G0438 or G0439) in February. 141 have had no AWV in 24+ months. 246 are due for their subsequent AWV. Estimated billable value: $67,200.
AWVs (G0438 initial, G0439 subsequent) reimburse at approximately $174 and $111 respectively under 2024 Medicare Fee Schedule. No cost-sharing to patient under ACA preventive care provisions.
"What percentage of our hypertensive patients have BP readings above 160 systolic?"
18.4% of hypertensive panel (214 of 1,163 patients) had most recent reading ≥160 systolic. 87 of these had no BP reading recorded in the past 6 months.
HEDIS CBP measure (Controlling High Blood Pressure) uses <140/90 as the target. JNC 8 guidelines recommend BP <150/90 for patients 60+ as the primary threshold for pharmacological intervention.
"Show CPT 99213 and 99214 billing distribution by provider for Q4"
Practice mean: 31% 99213, 54% 99214, 9% 99215. Dr. Williams: 12% 99213, 71% 99214 — statistically high 99214 rate may warrant documentation audit. Dr. Park: 61% 99213 — may be undercoding.
E&M coding distribution outliers (2+ SD from peer mean) can trigger MACs (Medicare Administrative Contractors) to initiate post-payment review. Internal audits at 12-month intervals are standard compliance practice.
"Which patients with COPD had an ED visit in the last 90 days?"
43 patients with ICD-10 J44.x who had an ED encounter in the past 90 days. 18 were subsequently admitted (COPD-related hospitalization rate: 41.9%). 11 had no follow-up visit within 7 days of discharge.
Post-discharge follow-up within 7 days for COPD patients is a HEDIS FUH measure component and a CMS quality reporting metric under the IPFQR program.
"What is our average days to close for patient complaints this year?"
Mean time to complaint resolution: 12.4 days YTD. Longest category: billing disputes (18.7 days). Shortest: scheduling (4.2 days). 7% of complaints exceeded 30 days — CMS QAPI standard requires 30-day closure.
CMS Conditions of Participation (42 CFR §482.13) require hospitals to have a grievance process. QAPI programs under §482.21 track complaint resolution time as a performance improvement indicator.
"Show revenue cycle KPIs: AR days, clean claim rate, and denial rate"
AR Days: 34.2 (target <35). Clean Claim Rate: 91.4% (industry benchmark 95%+). First-pass Denial Rate: 10.8% (target <5%). Net Collection Rate: 94.2%. Days in Accounts Receivable over 90 days: 18.3%.
A clean claim rate below 95% and denial rate above 5% are widely cited industry underperformance thresholds. Each additional AR day represents approximately $28,000 in working capital tied up for a 100-physician group practice.
Clinical Language Intelligence
Why Clinical Language Matters in Healthcare Analytics
Generic natural language query tools — including Microsoft Power BI's Q&A feature and Tableau's Ask Data — are designed for business data. They understand "revenue by region" and "units sold last quarter." They do not understand "HEDIS CDC measure," "CMS-HCC risk score," "DRG geometric mean length of stay," or "30-day readmission under the HRRP."
Vizier's natural language layer is trained on clinical and healthcare operations terminology. It resolves ICD-10 codes and code ranges (J44.x for all COPD variants). It understands that "diabetic patients" means patients with ICD-10 E10.x, E11.x, or E13.x. It knows that "annual wellness visit" maps to CPT G0438 and G0439.
When you ask "which providers are outliers on 99214 billing," Vizier calculates the peer distribution, applies statistical outlier detection, and flags providers who are 1.5 or more standard deviations above the practice mean — without you needing to specify that calculation.
Vizier vs. Generic NLQ Tools
"Show me diabetic patients with poor control"
Resolves to ICD-10 E10.x, E11.x, E13.x AND A1C > 9.0. Returns patient-level results.
Returns an error or asks "what does diabetic mean?"
"What are our MIPS quality measures?"
Matches your data to applicable 2024 MIPS quality measures by specialty and patient population.
Cannot resolve MIPS — no healthcare regulatory context.
"Which DRGs are above geometric mean?"
Pulls CMS FY2024 geometric mean LOS table, compares against your encounter data, ranks by variance.
Cannot resolve DRG or geometric mean without manual reference data upload.
"Show J44.x encounters this quarter"
Expands J44.x to all COPD subcategories (J44.0, J44.1, J44.9), returns encounter list.
Requires exact code. Cannot interpret ICD-10 wildcard ranges.
Contextual Conversation
Each Answer Opens the Next Question
Vizier maintains context within a session. After an initial answer, follow-up questions can reference "those patients," "that provider," or "the same time period" without repeating the full specification.
You
What is our readmission rate for CHF patients discharged in Q4?
Vizier
18.4% — 34 of 185 CHF discharges readmitted within 30 days. National CMS benchmark: 15.2%. Do you want to see the patient list or break this down by provider?
You
Break it down by attending physician.
Vizier
Dr. Holloway: 26.7% (8/30). Dr. Tanaka: 22.2% (6/27). Dr. Mbeki: 14.8% (4/27). Dr. Santos: 12.5% (3/24). Remaining 4 attendings: 11–14%. Dr. Holloway is 8.3 percentage points above practice mean.
You
Show me Dr. Holloway's readmitted patients — what were their discharge dispositions?
Vizier
Of Dr. Holloway's 8 readmitted patients: 5 were discharged to home without services, 2 to home health, 1 to SNF. 6 of the 8 were readmitted within 10 days, suggesting early decompensation rather than post-SNF decline.
See Conversational Analytics Against Your Own Data
Bring a question you have been unable to answer quickly. We will answer it in the first five minutes of the demo session.