Introduction — March 31, 2026
Precedent-Based Denial Intelligence

Case law
for claims.

Behavior-aware decision infrastructure for behavioral health revenue cycle. We capture what actually wins — and make it permanent.

82%¹
of denied claims are ultimately overturned on appeal
2–3ײ
higher denial rate in behavioral health vs. general medical
Presented to
Michael Burns, M.D., Ph.D.
Clinical Assoc. Professor, Anesthesiology
Associate CMIO for AI — Michigan Medicine
Presented by
Patrick Lord, MBA, MS
Founder, Stratum Collective
Context
First call — following up on our email exchange re: ARPA-H research partnership
¹ AMA, 2024 — Prior Authorization Physician Survey ² KFF, 2024 — Mental Health Parity Analysis
Where We Left Off
Conflict assessed — no overlap

Same cycle. Different end.

Service
Clinical documentation
Coding
CPT assignment
Submission
Claim filed
Adjudication
Payer review
Denial
Reason issued
Appeal
Evidence assembly
Decimal Code — pre-submission · coding accuracy · NLP on clinical notes · CPT out · perioperative
Stratum — post-denial · overturn patterns · evidence assembly · precedent out · behavioral health
Where they run parallel — without competing
Decimal Code
Stratum
NLP on clinical notes → assigns CPT codes at point of care
Core method
ML on denial records → predicts overturn patterns at appeal
Multi-site case data — model improves as more encounters train it
Network value
Multi-facility corpus — value compounds as more precedents accumulate
Providers lack accurate, consistent coding expertise — especially across payers
Knowledge gap closed
Providers lack payer-specific appeal intelligence — only aggregation reveals it
Perioperative / general medical — codes out of documentation
Domain
Behavioral health / SUD — evidence out of denial patterns
The Problem
Behavioral Health Revenue Cycle

A broken system with
recoverable revenue.

BH denial rate vs. general medical1
85% higher
Appealed denials ultimately overturned2
82%
40+
Percentage point overturn variance3
From evidence assembly alone — before clinical arguments are made. This is the signal Stratum is designed to close.
Why current tools fail

RCM AI optimizes for volume — it handles the predictable cases faster. Behavioral health denials aren't predictable. They're high-variability clinical scenarios with significant payer discretion at the medical necessity level.

The result: facilities that fight denials intelligently — with the right evidence, assembled in the right sequence, referencing the right precedents — overturn 40+ points more than those who don't.

That gap is entirely recoverable. The asymmetry isn't clinical — it's informational.

1 KFF, 2024 — Mental Health Parity Analysis 2 AMA, 2024 — Prior Authorization Physician Survey 3 Stratum early data — preliminary, validation in progress
What We Build
The Infrastructure

Precedent objects —
structured denial decision records.

Every denial, appeal, and outcome becomes a structured precedent object — versioned, queryable, outcome-labeled. Not a note. Not a PDF. A data asset the facility owns.

01 Denial Classification — payer, code, LOC, diagnosis cluster
02 Evidence Sequence — what was submitted, in what order, what worked
03 Narrative Architecture — the argument structure that moved the reviewer
04 Traceability Chain — guideline clause → evidence → outcome (transparent proof chain)
05 Escalation Economics — cost to appeal, expected recovery, break-even
06 Outcome Label — overturned, upheld, partial, withdrawn
07 Governance Metadata — version, author, last validated, drift flags
The compounding effect

Each precedent improves the next decision. The corpus compounds — the same dynamic that makes case law more valuable over time than any single ruling.

No one has built a structured, longitudinal, cross-payer corpus of behavioral health denial decisions. It doesn't exist in the published literature.

We don't draft appeals — we build cases. The corpus is what makes that repeatable.

CORPUS compounds Facility A Facility B Facility C Facility D Facility E Each facility contributes precedents · corpus returns intelligence to all
The Platform
In Practice

The corpus as working infrastructure.

Stratum
Dashboard Library Pipeline Intelligence
Appeal Pipeline
12 active · 2 due this week · $51,000 pending recovery
+ New Appeal
Evidence Gathering 4
Anthem PHP MedNec
Maya · Mar 24 · $8,000
UHC IOP Freq Limit
Maya · Mar 27 · $3,000
Submitted 3
Anthem PHP LOC
Maya · Mar 11 · $6,000
Magellan SUD MedNec
Alex · Mar 9 · $9,000
Awaiting 2
BCBS IOP MedNec
Maya · Feb 27 · $7,000
Outcome 3
UHC PHP MedNec
+$7,000 ✓
Aetna IOP LOC
$0 ✗
Stratum
Dashboard Library Pipeline Intelligence
Denial Intelligence
Behavioral health denial patterns, corpus coverage, and recovery metrics
Payer Trends Corpus Health Executive Summary
BH Denial Volume by Payer
Payer
Denials
Overturn
YoY
Volume
Anthem
1,247
78%
↑12%
UHC
892
71%
↓3%
BCBS
734
65%
↑8%
Cigna
456
82%
↑5%
Magellan
312
60%
↓7%
Stratum
Dashboard Library Pipeline Intelligence
Denial Intelligence
Behavioral health denial patterns, corpus coverage, and recovery metrics
Payer Trends Corpus Health Executive Summary
Precedent Coverage Heatmap
ANTHEM
UHC
BCBS
CIGNA
MAGELLAN
AETNA
Medical Nec.
Strong
Weak
Strong
Weak
Level of Care
Weak
Freq. Limit
Strong
Weak
Prior Auth
Mod.
Doc. Gap
Coverage: 8 of 30 combinations have precedents · 27%
Appeal Pipeline
Payer Trends
Corpus Coverage ← research surface
Research Opportunity
Three Open Questions

The research that needs to happen — and hasn't.

Question 01

Knowledge Transfer & Decay

How does appeal decision quality degrade when key staff turn over? What's the half-life of institutional RCM knowledge, and can it be modeled? Behavioral health has the highest staff turnover of any care setting — this is the most acute version of the question.

Question 02

Payer Drift Detection

Payer denial logic shifts without announcement. How quickly can a corpus-based system detect behavioral change in denial patterns — and how does that compare to a human team's detection latency? This is an ML problem with a real-world outcome.

Question 03

Decision Consistency Under Disruption

Can precedent-based systems maintain overturn rates during workforce disruption (staff loss, mergers, system migrations)? This is the resilience question that ARPA-H's Resilient Systems ISO was written for — and it's untested in the RCM setting.

These questions require real denial data at scale — structured, longitudinal, cross-payer. That dataset is what we're building. It doesn't currently exist anywhere in the published literature.

ARPA-H Alignment
Funding Context

Resilient Systems ISO — a direct fit.

Partnership structures — scoped to fit

Advisory Role

Validation methodology input. Study design review. ~2 hrs/month in-kind. Publication opportunity on corpus analysis findings.

Funded Evaluation Role

Named academic evaluator on the ARPA-H application. Budget line for formal evaluation deliverables and publication rights.

Sub-awardee

UMich as institutional sub-awardee. Larger scope — drift detection methodology, knowledge transfer modeling, corpus analysis.

The opportunity

ARPA-H-SOL-24-103 (Resilient Systems Office) funds health systems that maintain function under disruption. Stratum's thesis — that precedent-based infrastructure preserves decision quality through workforce instability — maps directly to the ISO's mandate.

We're preparing a solution summary and need an academic partner with credibility in AI/ML validation, health system research, and study design.

This isn't a primary revenue path for us — it's a validation accelerant. The goal is to fund the rigorous evaluation that makes the corpus credible at scale.

Paths Forward
Beyond ARPA-H

Multiple ways to work together.

ARPA-H Application Partner

Named academic partner on solution summary. Scoped to validation and study design methodology. Submission target Q2 2026.

Independent Corpus Research

First-mover publication rights on denial pattern analysis. IRB-compatible de-identified structure by design. Dataset doesn't exist anywhere else.

Development Partnership

Early access to the precedent corpus and platform. Inform design of the analytics and research export layer.

What I'm asking for today

Nothing yet. The goal of this conversation is to find out if the research questions are interesting to you and whether there's a natural intersection with your current work.

If it's interesting, we can talk about what the right scope looks like. If ARPA-H is the natural vehicle, great. If a more informal research collaboration makes more sense, also fine.

No commitment needed. I'd rather find the right fit than move fast on the wrong one.

The Parallel
Why Michigan. Why now.

Same pattern, different domain.

Stratum Corpus

  • Multi-facility denial corpus, growing longitudinally
  • Pattern extraction from denial records across payers
  • Overturn variance driven by evidence assembly differences
  • Knowledge transfer across facility types and markets
  • Intelligence that no single facility could produce alone
  • Validated, reusable — the compounding asset

MPOG Consortium

  • 84-system consortium, 15M procedural cases
  • Pattern extraction from clinical documentation across institutions
  • Outcome variance driven by evidence differences
  • Knowledge transfer across provider archetypes and geographies
  • Intelligence that no single site could produce alone
  • Validated, reusable — not one-off

The methodology you've already developed for multi-site outcome pattern aggregation is exactly the kind of expertise this work needs. The research questions are new, but the infrastructure problem is one you've already solved in a harder setting.