Draft SOO ACADM.pdf

DRAFT — Statement of Objectives: Autonomous Coding & Appeals and Denials Management

STATEMENT OF OBJECTIVES

for

AUTONOMOUS CODING &

AUTOMATED APPEALS AND DENIALS MANAGEMENT

1. BACKGROUND

Military Treatment Facilities (MTFs) across the Defense Health Agency (DHA) enterprise experience high encounter coding volume, increased coding errors, a growing volume of denied claims, and limited ability to predict which claims will be denied before submission. To manage these challenges, DHA requires an artificial intelligence/machine learning (AI/ML)-enabled solution capable of adapting and improving dynamically based on data patterns in order to autonomously code clinical encounters;

prioritize and resolve denied claims; and prevent denials before claim submission.

2. KEY DEFINITIONS

2.1. Final coded. An encounter is “final coded” when the solution assigns a complete code set and applicable modifiers sufficient for claim submission and releases the encounter without human intervention, with an audit trail traceable to the source documentation. This is the same unit of measure used for invoicing under Section 7.4; the terms “autonomously coded” and “successfully encoded” carry the same meaning throughout this SOO.

2.2. Coverage. Autonomous coding coverage shall be expressed against total in-scope encounter volume. Any encounter volume the solution excludes from autonomous coding shall be reported together with the basis for exclusion and the resulting exclusion rate.

2.3. Production. The live, operational technical environment utilized for actual day-to-day data operations integrated with MHS GENESIS. This definition refers strictly to the hosting and execution environment for processing live military health data and is distinct from, and should not be confused with, "follow-on production" acquisition pathways, contracts, or transactions as defined under 10 U.S.C. § 4022.

3. APPROACH & OBJECTIVES

3.1. FUNCTIONAL OBJECTIVES

• Autonomous Coding (AC): Provide an AI/ML capability to automatically assign medical billing codes without human intervention. Solutions should process roughly 143K encounters daily (4.35M monthly) with no later than 24-hour turnaround, adhering to current coding guidelines with full transparent auditability (“explainability”).

• Appeals and Denials Management (ADM): Use AI/ML to automate the generation of appeals, reduce future claim denials, assess payer performance, and provide auditing for financial impacts.

3.2. SHARED OBJECTIVES (Joint Vendor Solution Requirements)

If the Government does not select a single vendor to provide both solutions, the selected Vendors must work collaboratively, as needed, to integrate their solutions in order to meet the following requirements:

1. Comprehensive Reporting Dashboards: Vendors will provide unified reporting dashboards that consolidate data and analytics from both vendors’ systems, presenting a cohesive view of the complete solution performance and outcomes.

2. Interdependent Algorithm Learning: The vendors will implement interdependent machine learning algorithms that operate across both the Autonomous Coding (AC) and Appeals and Denials Management (ADM) systems. The algorithms must be designed to share learning insights and optimize performance across both vendors' platforms in a coordinated manner.

3. Secure Data Exchange and Integration:

o Phase 1 (Initial): Vendors will establish secure, compliant file transfer protocols between their respective systems to enable data exchange while maintaining HIPAA compliance and DoD security requirements.

o Phase 2 (Ultimate): Vendors will collaboratively develop and implement integration capabilities with MHS GENESIS that:

Enable seamless data flow between both vendor solutions and MHS GENESIS

Preserve MHS GENESIS core functionality without modification

Utilize approved MHS GENESIS APIs, interfaces, or integration points as designated by the Government

Maintain all security, privacy, and interoperability standards required by the Defense Health Agency and Department of War.

The vendors will designate technical leads responsible for coordination and shall provide joint integration plans, testing protocols, and documentation demonstrating successful interoperability between their systems upon implementation of Capability 1 and 2 as outlined below.

4. Execution Phases & Deliverables The project will be executed via a two-capability, milestone-driven structure. AC and ADM may or may not start in parallel with pending award (s) but will follow the same milestone process. Advancement to the live production environment (Milestone 4) is strictly gated by the successful progressive completion of cybersecurity, onboarding, and performance metrics in the pre-production environment (Milestones 1, 2, and 3).

4.1. CAPABILITY 1: AUTONOMOUS CODING

• Milestone 1: Onboarding & Cybersecurity o Objective: Establish secure baseline for DOD data handling.

o Criteria: All vendor personnel assigned to this effort shall be credentialed (CAC). Risk Management Framework (RMF) artifacts approved for an Interim Authority to Test

(IATT).

• Milestone 2: Testing in Pre-Production Environment o Successfully Autonomously Code Outpatient/Clinic encounters utilizing an established set of MHS GENESIS workflows, user roles, and approved test data that will be provided by the Government. Success at this milestone requires autonomous coding of ≥ [60] % of the approved test volume at ≥ 95% code-level accuracy, reported by encounter category, within the 24-hour turnaround objective.

o Recommended priority order (Outpatient/Clinic encounters) for testing in Pre-Prod:

Family Medicine, Internal Medicine, and Flight Medicine Behavioral Health Laboratory Emergency Medicine Radiology to include its subspecialties (outpatient only) Immunizations Additional Primary Care Outpatient/Clinic encounters Specialty Care Day Surgery

• Milestone 3: Authority to Operate Approval o Objective: Secure authorization for live data operations in a Production environment integrated with MHS GENESIS. Hosting environment demonstrated DoD Impact Level 4+ (IL4+) authorization, attained through FedRAMP High or FedRAMP Moderate with DoD Privacy Overlays.

o Criteria: Submit required artifacts and achieve an Authority to Operate — Conditional (ATO-C) at a minimum.

• Milestone 4: Full Integration with MHS GENESIS in Production Environment o Deploy/Implement Autonomous Coding capability to select military treatment facilities for validation. Autonomously code ≥ 80% of total in-scope production volume with ≥ 95% code-level accuracy. Process encounters in under 5 minutes on average and in no case later than the 24-hour turnaround objective, with a 100% reliable fallback mechanism for human routing for a period no less than 90 days.

4.2. CAPABILITY 2: APPEALS AND DENIALS MANAGEMENT

• Milestone 1: Onboarding & Cybersecurity o Objective: Establish secure baseline for DoW data handling.

o Criteria: All vendor personnel assigned to this effort shall be credentialed (CAC). RMF artifacts approved for an Interim Authority to Test (IATT).

• Milestone 2: Testing in Pre-Production Environment o Appeals and Denials Management: Ingest and parse active, open, and aged A/R records.

Generate diagnostic reports segmenting claims achieving a ≥ [80]% correlation match against a human-audited sample of no less than 500 claims, utilizing an established set of MHS GENESIS workflows, user roles and approved test data that will be government provided. Automatically map identified denials to standardized root causes.

• Milestone 3: Authority to Operate Approval o Objective: Secure authorization for live data operations in a Production environment integrated with MHS GENESIS.

o Criteria: Submit required artifacts and achieve an Authority to Operate — Conditional (ATO-C) at a minimum. Hosting environment demonstrates DoD Impact Level 4+ (IL4+) authorization, attained through FedRAMP High or FedRAMP Moderate with DoD Privacy Overlays

• Milestone 4: Full Integration with MHS GENESIS in Production Environment o Ingest and parse 100% of active, open, and aged A/R records. Generate diagnostic reports segmenting claims achieving a ≥ 90% correlation match against a human-audited sample of 500 claims. Automatically map 100% of identified denials to standardized root causes.

Implement automated prevention resulting in a ≥ 20% reduction in claim denials attributable to coding errors. This reduction shall be credited only where net third-party collections per billable encounter hold constant or increase over the same measurement period. Autonomously generates required appeal artifacts for ≥ 50% of the prioritized, recoverable claims for a period of no less than 90 days. Performance under this criterion shall be credited only for claims subsequently collected or overturned, not for appeals generated.

5. SUCCESS CRITERIA:

In accordance with 10 U.S.C. § 4022, successful completion of this prototype project (or portions thereof) is defined by written Government acceptance of required deliverables and demonstrated quantitative success of the defined milestones. To ensure a clear path to product validation, the vendor’s SOW should align with the following quantitative success criteria:

5.1. CAPABILITY 1: AUTONOMOUS CODING

• Autonomous coding accuracy of ≥ 95% at the code level and, in no case, less than the measured baseline manual coding accuracy, for each deployed specialty and encounter type

• Demonstrated reduction in claim denial rates attributable to coding errors

• Reliable data exchange with MHS GENESIS via file transfer with no degradation to existing system performance

• Completion of all five Coding Tests — corresponding to the Outpatient/Clinic, Radiology, Laboratory, Immunizations, and Emergency Department encounter categories identified in Milestone 2 — within the agreed schedule objectives

• Measurable reduction in Discharged-Not-Final-Coded (DNFC) volume against the pre-prototype baseline

• Measurable reduction in Accounts Receivable (A/R) days against the pre-prototype baseline

5.2. CAPABILITY 2: APPEALS AND DENIALS MANAGEMENT

• Increased revenue recovery from denied claims compared to pre-prototype baseline, measured in dollars against the billable third-party (Other Health Insurance) population

• Reliable data exchange with MHS GENESIS via file transfer with no degradation to existing system performance

• Operational Denials Prevention capability demonstrated at pilot site(s)

6. FINANCIAL CONTROLS & DATA RIGHTS

6.1. FINANCIAL AND AUDIT CONTROLS

• The solution shall not autonomously write off, adjust, settle, or otherwise financially dispose of any claim. All financial dispositions require human review and approval.

• Every automated action shall generate a complete, FFMIA-aligned audit trail sufficient to support financial statement audit and Inspector General review, identifying the code or action selected, the supporting source documentation, and the approving user where human approval is required.

• The solution may generate coding determinations and appeal artifacts within the parameters of this SOO. It shall not make final determinations of claim collectability or claim value.

6.2. DATA RIGHTS AND ENGINE PROVENANCE

The vendor shall identify the underlying coding or denials engine proposed, whether the vendor owns or licenses it, and the vendor’s rights to maintain, patch, and retrain it across the period of performance and any follow-on production period. The vendor shall propose data rights for the trained model, tuned model parameters, and any Government-derived datasets.

7. DELIVERABLES:

Information contained in written (hard copy or electronic) deliverables (reports, etc.) shall be subject to distribution within US Government agencies and their contractors (Distribution Statement C). Acceptable deliverable format acceptable: MS Office/ PDF; financial data in MS Excel. Deliverables must NOT be marked proprietary and must have Distribution Statement C. All deliverables will be submitted to the DHMSM Product Manager (PM) and Agreement Officer Representative (AOR). Deliverables must include at a minimum:

7.1. MONTHLY PROGRESS REPORTS

Submitted monthly no later than the 10th of the month. Each monthly report, at a minimum, must contain:

• Summary of monthly progress for each coding test and denials management test, including relevant AI model performance metrics

• Identification of any task or milestone that is slipping or missed, with discussion of path forward and impact on other milestones

• Summary of risks, discussion of potential impacts, and efforts to mitigate

• Summary of overall schedule and changes from previous month

• Financial summary of overall project costs and budget to date

7.2. QUARTERLY IN-PROCESS REVIEWS

Scheduled as needed (generally quarterly) at DHA’s facilities (or virtual, at DHA’s discretion).

Duration: 8 hours maximum. Face-to-face review of previous quarter’s activities. Informative in nature to keep the Government apprised of project progress and to discuss issues that may require joint resolution. Meeting minutes will be submitted within 14 days of meeting occurrence.

7.3. AI MODEL VALIDATION REPORTS

Submitted upon completion of each Coding Test and Denials Management Test. Each report shall include model performance metrics, validation methodology, comparison to baseline, identified limitations, and recommendations for the next phase.

7.4. MONTHLY INVOICING

The vendor should propose a monthly invoice process that ensures timely submission of accurate invoices for successfully encoded encounters. Invoices must be supported by detailed report on encounter volumes by tier, applicable rates, and etc. The proposed SOW should include:

• Format and level of detail for invoice backups and support reports.

• Reconciliation and dispute resolution procedures.

7.5. FINAL PROJECT REPORT

Submitted no later than 90 days after successful completion of the prototype. Final report shall summarize the project plan objectives and the success achieved in meeting those objectives; summary of risks incurred, impacts, and mitigation; AI model performance across all coding tests and denials tests; financial summary; schedule summary comparing original to final; and recommendations for enterprise-wide deployment.

8. SCHEDULE AND MILESTONE PAYMENTS

8.1. SCHEDULE OBJECTIVES:

The overall schedule for this effort shall be derived from the authorization critical path rather than from a fixed calendar completion date. Authority to Operate – Conditional (ATO-C) is the pacing milestone: production operation may not begin before ATO-C is granted, and the audit monitoring period may not begin until production operation begins. The Government’s schedule objectives are therefore expressed as durations from the date of award (Aug 2026), sequenced so that the Outpatient/Clinic category reaches production at the earliest defensible point on the critical path.

• Milestone 1 — Onboarding and Cybersecurity through Interim Authority to Test (IATT) shall be completed not later than [90] days from award.

• Milestone 2 — Pre-production testing shall be completed not later than [90] days following IATT. Testing shall be sequenced by encounter category, with the Outpatient/Clinic category completed first so that it is positioned to enter production at the earliest point on the critical path.

Radiology, Laboratory, Immunizations, and Emergency Department follow.

• Milestone 3 — Authority to Operate – Conditional (ATO-C) shall be achieved not later than [120] days following completion of pre-production testing. This duration reflects that artifact review and adjudication are Government-dependent activities and are not compressible by vendor effort alone.

• Milestone 4 — Production entry shall occur upon grant of ATO-C, beginning with the Outpatient/Clinic category. Each deployed encounter category shall then operate for a continuous audit monitoring period of not less than 90 days and up to six (6) months before the Government will consider the prototype objectives met for that category. The 90-day human-routing fallback period required at Milestone 4 runs concurrently within this window.

Estimated Timeline: Applying these durations to an August 2026 award places ATO-C at approximately June 2027 and completion of the Outpatient/Clinic six-month audit monitoring period at approximately October-December 2027, yielding an estimated period of performance of [13-16] months. Remaining encounter categories may be phased into production during the outpatient monitoring window and shall each complete their own six-month monitoring period.

Where a fixed completion date is required for funding or fiscal-year reasons, the Government will adjust scope — by limiting the number of encounter categories entering production within the period of performance — rather than compress the authorization or audit monitoring periods, neither of which can be shortened without accepting audit and accreditation risk.

8.2. DELIVERY SCHEDULE TARGET DATES:

Phase/Milestone Target Dates

Capability 1: Autonomous Coding Integrated with MHS GENESIS and validated in Production

30 June 2027

Capability 2: Appeals and Denials Management Integrated with MHS GENESIS and validated in Production

30 June 2027

8.3. MILESTONE PAYMENT SCHEDULE:

The following table outlines the proposed payment structure for the Autonomous Coding and Appeals and Denials Management solution. Milestone payments in this table fund onboarding, pre-production testing, and authorization activities. Production operations following Milestone 4 shall be invoiced on a per-encounter basis by complexity tier in accordance with Section 7.4. Vendors shall propose both elements solution.

Qty Unit Price Total

Capabilities 1 and 2

Milestone 1 – Security & Onboarding 1 TBD TBD

Milestone 2 – Testing in Pre-Production Environment

1 TBD TBD

Milestone 3 – Authority to Operate Approval

1 TBD TBD

Milestone 4 – Full integration with MHS GENESIS and Validation in Production

1 TBD TBD

Total Estimated Value TBD TBD

9. RISK MANAGEMENT & SECURITY

9.1. RISK MANAGEMENT OBJECTIVES:

The vendor Statement of Work (SOW) shall develop and actively maintain a risk register. This includes creating a comprehensive Contingency Plan to mitigate identified risks, such as data loss, network failures, or schedule delays, ensuring mission-critical operations are protected.

9.2. SECURITY:

The highest level of classification for this effort is Unclassified. All data handling shall comply with DoD 5200.01, DHA privacy and cybersecurity policies, HIPAA, and all applicable federal regulations governing Protected Health Information (PHI). Should access to a higher classification be required in the completion of this effort, the DoD will coordinate a DD 254 with the Company.

APPENDICES

Appendix A

List of Cyber Artifacts for IATT

• Vulnerability Scans

• Security Technical Implementation Guides (STIGS)

• Applicable Control List

• Plan of Action and Milestones (POA&M)

• Draft Incident Response Plan

• Draft Contingency Plan

• Draft Configuration Management Plan

• Detailed Hardware/Software Inventory

• Boundary Diagram

• Draft Security Plan

• Draft Audit Policy

Appendix B

List of Cyber Artifacts for ATO-C

• Vulnerability Scans

• Security Technical Implementation Guides (STIGS)

• Applicable Control List

• Plan of Action and Milestones (POA&M)

• Incident Response Plan

• Contingency Plan

• Configuration Management Plan

• Detailed Hardware/Software Inventory

• Boundary Diagram

• Security Plan

• Audit Policy