S7 Intelligence — Risk Adjustment, engineered by clinicians and machines

The S7 Platform

One knowledge graph. Every risk-adjustment workflow.

S7 Intelligence unifies coding, validation, RAF, and audit around a curated clinical knowledge graph — so every recommendation is explainable, every code is defensible, and every dollar of revenue is traceable back to a source.

Prospective & Retrospective Coding

Suggest, validate, and rank ICD-10-CM codes with HCC impact, RAF delta, and rule-level rationale — pre-visit through post-encounter.

Real-time Validation

Excludes1, age/sex edits, HCC hierarchy trumping, non-billable header checks — blocked before they hit the claim.

RAF Scoring, CMS-HCC V28

Full model: demographic base, hierarchy-trumped HCCs, disease-count factors, and community disease interactions — with per-component breakdown.

Full-catalog Code Search

Sub-100ms search across the entire ICD-10-CM catalog with billable-first ranking, HCC mapping, and RAF weight surfaced inline.

3D Knowledge Graph Explorer

WebGL "graph galaxy" for auditors and clinical informaticists — inspect any code's neighborhood, HCC path, and applied rules in three dimensions.

Grounded NL Assistant

Ask "why HCC 38?" — the assistant answers with graph lookups you can inspect, not hallucinated citations. Tool-calling over 8 read-only graph tools.

Product

S7 Coder for coding teams & QA

A browser-based workspace for HCC coders, CDI specialists, and audit reviewers. Every suggestion is explainable, every code is graph-verified, and every patient's RAF is decomposed to the coefficient.

What you get on day one

Encounter workspace — patient demographics, problem list, prior conditions, and current-visit codes in a single pane.
Suggest & defend — every AI suggestion carries the ICD-10 → HCC → RAF chain and the rule(s) that justify it.
Block-before-you-bill — non-billable headers, Excludes1 conflicts, age/sex edits, and payer-specific rules are enforced pre-submit.
Patient RAF, live — total score plus per-HCC, demographic, interaction, and disease-count decomposition.
Coder Assistant — natural-language Q&A grounded in the graph, with visible tool-call trace.
3D Graph Explorer — full standalone dashboard for auditors and informaticists.
Role-based access — SSO/SAML, granular scopes, and full audit log.
Deployment — SaaS (multi-tenant), single-tenant VPC, or on-prem for regulated environments.

app.s7intelligence.com / coder

E11.9 Type 2 diabetes mellitus without complications HCC 38RAF 0.166

I50.9 Heart failure, unspecified HCC 226RAF 0.360

N18.3 Chronic kidney disease, stage 3 Requires N18.30–.32

F32.9 Depression, unspecified Blocked: Excludes1 with F33.0

Patient RAF

1.034

Demographic0.396

HCC 380.166

HCC 2260.360

Diabetes × HF0.112

98%

HCC capture accuracy at audit

40%

Reduction in retro-review cycle time

<100ms

Code search & validation p95

100%

Recommendations with graph rationale

See it in action

Ask AI — grounded in the graph, not guessed by the model.

A live coder question, answered against S7's clinical knowledge graph. Every claim is backed by the exact tool call and node lookup performed — the same trace an auditor sees.

💬 S7 Coder Assistant live · graph-grounded

SK

Sarah Kim Coder

Should I code E11.9 or E11.65 for a 71-year-old with diabetes and A1c of 9.4? Any HCC impact?

S7

S7 Assistant grounded

🔎 search_codes q: "type 2 diabetes hyperglycemia" ✓

📋 get_code_details code: "E11.65" ✓

🧬 check_hcc_hierarchy codes: ["E11.9", "E11.65"] ✓

📐 rules_for_code code: "E11.65", payer: "MA" ✓

With an A1c of 9.4 the correct code is E11.65 — Type 2 diabetes mellitus with hyperglycemia.

Both E11.9 and E11.65 map to HCC 38 (Diabetes without Complications), so the RAF is the same at 0.166 — but E11.65 is the more specific, defensible code given the documented A1c.

Also flagged: A1c ≥ 9.0 supports adding Z79.4 if the patient is on long-term insulin — no HCC impact, but a common CDI query.

Grounded in 4 graph lookups · view trace →

Rest of the workspace

Small snapshots of the surfaces coders and auditors use every day.

Encounter workspace

MJ

Maria Johnson

71 · F · MA — Humana

Codes Notes RAF

E11.65 Type 2 DM w/ hyperglycemia HCC 38

I50.9 Heart failure, unspec. HCC 226

N18.3 CKD, stage 3 needs N18.30–.32

F32.9 Depression, unspec. Excludes1 vs F33.0

Code search (full catalog)

🔎 diabetes hyperglycemia

E11.65 Type 2 diabetes mellitus with hyperglycemia HCC 38 0.166

E11.9 Type 2 DM without complications HCC 38 0.166

E11.21 Type 2 DM w/ diabetic nephropathy HCC 37 0.302

RAF breakdown, live

Patient RAF (CMS-HCC V28 · CNA)

1.034

▲ +0.472 vs. base

Demographic (71 M)0.396

HCC 38 — Diabetes0.166

HCC 226 — Heart Failure0.360

Interaction · DIAB × HF0.112

3D Graph Galaxy explorer

Neighborhood of E11.65 — click any node to grow the galaxy

API

S7 Coding Intelligence API

Embed S7 into your EHR, CDI workflow, claims scrubber, or coder desktop. RESTful, JSON in / JSON out, OAuth 2.0 client-credentials, versioned, and BAA-covered. FHIR R4 endpoints available for provider integrations.

POST/v1/validate

Validate a candidate code set against Excludes1, HCC hierarchy, age/sex edits, non-billable headers, and payer rules. Returns blocking vs warnings with rule citations.

POST/v1/suggest

Given free-text clinical notes and patient context, return ranked ICD-10-CM suggestions with HCC mapping, RAF delta, and evidence spans from the source note.

GET/v1/codes/search

Sub-100ms full-catalog search across ICD-10-CM. Billable-first ranking, HCC-joined, RAF-annotated. Prefix + fulltext.

GET/v1/codes/{code}

Full code detail: parents, children, chapter, HCC mapping, RAF (per segment), Excludes1/Excludes2, and applicable rules.

POST/v1/raf

Compute patient RAF (CMS-HCC V28). Returns total score plus demographic base, per-HCC contribution, disease-count factor, and community interactions.

GET/v1/graph/{code}

Knowledge-graph neighborhood: nodes (ICD10, HCC, Chapter, Rule) and typed edges (PARENT_OF, MAPS_TO_HCC, EXCLUDES1, HCC_SUPERSEDES).

POST/v1/ask

Natural-language grounded Q&A. Returns an answer plus the exact graph lookups performed — no hallucinated citations.

POST/v1/audit/rescore

Batch RAF rescore for a member panel across a date range. Suitable for retrospective HCC recapture and RADV prep.

curl · validateHTTP

POST https://api.s7intelligence.com/v1/validate
Authorization: Bearer <token>
Content-Type: application/json

{
  "patient": { "age": 71, "sex": "M", "payer": "MA" },
  "codes":   ["E11.9", "I50.9", "N18.3"]
}

200 · responseJSON

{
  "blocking": [
    { "code": "N18.3",
      "rule": "non_billable_header",
      "message": "Requires N18.30–N18.32" }
  ],
  "warnings": [],
  "raf": {
    "total": 1.034,
    "components": {
      "demographic": 0.396,
      "hcc_38":      0.166,
      "hcc_226":     0.360,
      "int_diab_hf": 0.112
    }
  },
  "trace": [ "graph:code:E11.9", "graph:hcc_supersedes",
             "graph:interaction:DIABETES_HF" ]
}

Auth OAuth 2.0 client-credentials · mTLS optional

SLA 99.9% · p95 < 250ms · region-pinned

Regions us-east-1 · us-west-2 · single-tenant VPC

Contracts BAA · DPA · sub-processor list on request

Why knowledge-centric

Every answer is traceable to a node in the graph.

Most Risk Adjustment tools guess with a model and hope the auditor agrees. S7 does the opposite: every suggestion, every RAF component, every denial-risk flag is anchored to a specific edge in our curated clinical knowledge graph — the same graph auditors can inspect.

115

CMS-HCC V28 categories with per-segment RAF coefficients

70k+

ICD-10-CM codes indexed with billable/header flags and chapter context

7

Disease segments (CNA · CND · CFA · CFD · CPA · CPD · INS)

6

Community disease interactions modeled (Diabetes × HF, HF × Kidney, …)

8

Read-only graph tools available to the grounded NL assistant

100%

Rules authored, reviewed, and versioned by clinical coders

Case Studies

What S7 Intelligence delivers in the field.

Selected outcomes from Medicare Advantage plans, ACO REACH entities, and multi-specialty provider groups. Metrics are directional representations of production deployments; details available under NDA.

Payer · Medicare Advantage

Retrospective HCC recapture across a 60k-member panel

A regional MA plan engaged S7 to reprocess two years of encounters ahead of the sweeps deadline. S7 Coder plus the /v1/audit/rescore API ingested claims and provider notes, surfaced missed HCCs with graph-cited rationale, and queued them for coder review.

+0.11 avg RAF uplift per member reviewed
62% of suggestions accepted by senior QA
3.4× throughput vs. prior workflow
0 audit findings on sampled charts

Provider · Multi-specialty group

Prospective coding embedded in the point-of-care workflow

A 1,200-provider group integrated the S7 Coding Intelligence API into their EHR's problem-list workflow. Suggestions surfaced with HCC impact and RAF delta, and Excludes1 / age-sex edits blocked non-billable combinations before the note was signed.

−41% denial rate on risk-adjusted claims
+27% HCC documentation completeness
<180ms p95 in-EHR suggestion latency
90 days to production deployment

Payer · RADV audit prep

RADV-ready evidence packages, generated at scale

For an ACA-market carrier preparing for a RADV audit, S7 rescored the flagged sample, walked each HCC back to its source documentation, and produced an evidence bundle per contract — every citation traceable to a node in the knowledge graph.

100% of flagged HCCs traced to source evidence
−58% attorney-hours vs. prior audit cycle
0 unsupported diagnoses at bundle submission
2 weeks end-to-end for a 400-chart sample

ACO REACH

Grounded coder-assist for a downside-risk ACO

An ACO REACH entity replaced a generative coding tool that produced unsourced recommendations. S7's grounded NL assistant answers every "why this code?" question with the exact graph lookups performed — the same evidence its coders and QA teams use.

0 hallucinated citations across 12k queries
+18% coder confidence in flagged suggestions
−34% QA rejection rate on AI-assisted codes
SOC 2 & HIPAA requirements satisfied on day one

Leadership & Careers

Founder-led. Coder-obsessed. Built to last.

S7 Intelligence is led by a team of operators with two decades of hands-on experience running Risk Adjustment operations for US Payers and Providers. Our leadership is the reason we insist every recommendation is traceable back to a graph node — because we've sat on the audit side of the table too.

Navin Sethia

Managing Director

Sets the strategic direction of S7 Intelligence and owns key payer & provider relationships. Two decades operating US-focused Risk Adjustment programs from Chennai.

Vinay Sethia

Chief Executive Officer

Runs the company end-to-end — product, delivery, growth, and P&L. Anchored in the US market from the Fremont office, Vinay leads the shift from a services-first shop to a knowledge-graph-powered product platform for Payers and Providers.

Pavan Sethia

Chief AI & Product Officer

Owns the S7 product platform, the clinical knowledge graph, and the applied-AI roadmap. Draws the line between "model says so" and "graph proves so" — the reason every S7 recommendation ships with citations an auditor can inspect.

Vikas Sethia

Chief Financial Officer

Owns finance, contracting, and audit readiness. Anchors the compliance program that underpins the company's ISO/IEC 27001, SOC 2, and HIPAA attestations.

Praveen Sethia

Chief Legal Officer

Oversees legal, regulatory, and BAA/DPA contracting with US covered entities. Chairs the privacy program and sub-processor governance.

Pradeep Sethia

Director of IT & Infrastructure

Runs the platform, cloud, and information-security engineering across all offices. Owner of the technical controls behind the ISO 27001:2022 ISMS.

Careers

Join a knowledge-centric team.

We're hiring across coding, clinical informatics, ML engineering, and platform. If you've spent time inside HCC, ICD-10-CM, or payer-side risk-adjustment operations — or if you build tools that clinicians actually trust — we'd like to meet you.

Senior HCC Coder

Chennai · Full-time

Clinical Informaticist

Chennai / Remote · Full-time

ML Engineer, Coding Intelligence

Coimbatore · Full-time

Platform / SRE Engineer

Chennai · Full-time

Solutions Engineer, US Payers

Fremont, CA · Full-time

CDI Lead

Chennai · Full-time

Compliance & Security

Built for the auditors, not around them.

S7 Intelligence operates a formally documented Information Security Management System and privacy program. We handle PHI under a Business Associate Agreement and maintain independently attested controls.

ISO

ISO/IEC 27001:2022

Certified Information Security Management System covering people, processes, and platforms — including risk treatment, incident response, and supplier controls.

SOC

SOC 2 Type I

Independently attested controls across Security, Availability, and Confidentiality trust services criteria. Type II in progress.

HIPAA

Covered as a Business Associate. Encryption in transit and at rest, minimum-necessary access, PHI segregation, and a documented breach-response program.

SDLC

Secure SDLC

SAST/DAST on every merge, dependency SBOM, quarterly third-party penetration testing, and annual tabletop exercises.

Locations

Global delivery, US-anchored partnerships.

Six offices across India and the United States, aligned to our payer and provider customers' time zones and audit cycles.

Headquarters

Chennai

329, Fountain Plaza, 6th Floor, Pantheon Road,
Chennai, Tamil Nadu 600008, India

Branch Office

Chennai

Prakash Towers, 1st Floor, OMR Service Rd,
Elango Nagar, Perungudi,
Chennai, Tamil Nadu 600041, India

Branch Office

Chennai

76, Mint Tower, Bethel Nagar Street,
Industrial Estate, Perungudi,
Chennai, Tamil Nadu 600096, India

Branch Office

Chennai

21, Ambit Park Road, Ambattur Industrial Estate,
Ambattur, Chennai 600058, Tamil Nadu, India

Branch Office

Coimbatore

Covai Tech Park, Plot No. 12, 4th South Cross St,
Kovai Thirunagar, Indira Nagar,
Civil Aerodrome Post, Nehru Nagar West,
Coimbatore, Tamil Nadu 641014, India

Sales Office · USA

Fremont, California

S2M Inc, 38133 Cambridge Ct,
Fremont, CA 94536, USA

See S7 Intelligence on your data.

Book a 30-minute working session. We'll walk a sample encounter through S7 Coder, run the API against a de-identified sample of your claims, and share a written RAF-uplift estimate within 5 business days.

De-identified sample under BAA
Written RAF-uplift and denial-risk report
API sandbox credentials on request

Turn every chart into accurate revenue and cleaner care.

One knowledge graph. Every risk-adjustment workflow.

Prospective & Retrospective Coding

Real-time Validation

RAF Scoring, CMS-HCC V28

Full-catalog Code Search

3D Knowledge Graph Explorer

Grounded NL Assistant

S7 Coder for coding teams & QA

What you get on day one

Ask AI — grounded in the graph, not guessed by the model.

Rest of the workspace

S7 Coding Intelligence API

Every answer is traceable to a node in the graph.

What S7 Intelligence delivers in the field.

Founder-led. Coder-obsessed. Built to last.

Navin Sethia

Vinay Sethia

Pavan Sethia

Vikas Sethia

Praveen Sethia

Pradeep Sethia

Join a knowledge-centric team.

Built for the auditors, not around them.

ISO/IEC 27001:2022

SOC 2 Type I

HIPAA

Secure SDLC

Global delivery, US-anchored partnerships.

Chennai

Chennai

Chennai

Chennai

Coimbatore

Fremont, California

See S7 Intelligence on your data.