msdrg 0.1.7

High-performance MS-DRG (Medicare Severity Diagnosis Related Groups) grouper

mz-drg

High-performance CMS claim processing tools written in Zig with Python bindings.

---

mz-drg provides open-source reimplementations of two CMS tools:

• MS-DRG Grouper — assigns Diagnosis Related Groups based on diagnoses, procedures, and demographics
• Medicare Code Editor (MCE) — validates ICD diagnosis and procedure codes against CMS edit rules

Both are written in Zig, callable from Python, and validated against the CMS reference Java implementations with a 100% match rate on 50,000+ claims.

Why mz-drg?

The official CMS tools are Java applications. While accurate, they come with practical limitations:

	Java (CMS)	mz-drg
Startup	JVM warmup, seconds	Instant
Throughput (Ryzen 5 5600U)	~500 claims/sec	~11,000+ claims/sec
Memory	JVM heap overhead	Minimal, memory-mapped data
Dependencies	JRE 17+, classpath management	Single shared library
Python integration	JPype bridge (fragile)	Native ctypes (simple)
Embedding	Requires JVM process	C ABI, any language

Both engines are ported line-by-line from the decompiled Java source and validated claim-by-claim against the original.

Quick start

Install

pip install msdrg

MS-DRG Grouper

import msdrg

with msdrg.MsdrgGrouper() as grouper:
    result = grouper.group({
        "version": 431,
        "age": 65,
        "sex": 0,
        "discharge_status": 1,
        "pdx": {"code": "I5020"},
        "sdx": [{"code": "E1165"}],
        "procedures": []
    })

print(result["final_drg"])            # 293
print(result["final_mdc"])            # 5
print(result["final_drg_description"])  # "Heart Failure and Shock without CC/MCC"

Medicare Code Editor

import msdrg

with msdrg.MceEditor() as mce:
    result = mce.edit({
        "discharge_date": 20250101,
        "age": 65,
        "sex": 0,
        "discharge_status": 1,
        "pdx": {"code": "I5020"},
        "sdx": [{"code": "E1165"}],
        "procedures": []
    })

print(result["edit_type"])  # "NONE"
print(result["edits"])      # [] — no edits triggered

Unified claim — same dict for both

import msdrg

claim = {
    "version": 431,
    "discharge_date": 20250101,
    "age": 65, "sex": 0, "discharge_status": 1,
    "pdx": {"code": "I5020"},
    "sdx": [{"code": "E1165"}],
    "procedures": []
}

with msdrg.MsdrgGrouper() as g, msdrg.MceEditor() as mce:
    drg = g.group(claim)
    mce_result = mce.edit(claim)

MS-DRG Grouper

Input format

{
    "version": 431,              # MS-DRG version (e.g. 400, 410, 421, 431)
    "age": 65,                   # Patient age in years
    "sex": 0,                    # 0=Male, 1=Female, 2=Unknown
    "discharge_status": 1,       # 1=Home/Self Care, 20=Died
    "hospital_status": "NOT_EXEMPT",  # "NOT_EXEMPT" (default), "EXEMPT", or "UNKNOWN"
    "tie_breaker": "CLINICAL_SIGNIFICANCE",  # "CLINICAL_SIGNIFICANCE" (default) or "ALPHABETICAL"
    "pdx": {                     # Principal diagnosis (required)
        "code": "I5020",
        "poa": "Y"               # Present on Admission: Y/N/U/W (optional)
    },
    "admit_dx": {                # Admission diagnosis (optional)
        "code": "R0602"
    },
    "sdx": [                     # Secondary diagnoses (optional)
        {"code": "E1165", "poa": "Y"},
        {"code": "I10", "poa": "Y"}
    ],
    "procedures": [              # Procedure codes (optional)
        {"code": "02703DZ"}
    ]
}

Hospital status

The hospital_status field controls how Hospital-Acquired Condition (HAC) processing is applied, per CMS rules:

Value	Behavior
"NOT_EXEMPT"	Standard HAC processing. Default.
"EXEMPT"	Hospital is exempt from POA reporting. No HAC/POA ungroupable conditions.
"UNKNOWN"	Stricter POA validation with specific ungroupable return codes.

Tie breaker

The tie_breaker field controls how the grouper resolves attribute matches when multiple secondary diagnoses could match the same DRG formula attribute. This determines which diagnosis "wins" during the marking phase.

Value	Behavior
"CLINICAL_SIGNIFICANCE"	MCC diagnoses get first pick over CC, then by ICD code string. Default, matches CMS Java grouper.
"ALPHABETICAL"	Sort by ICD code string only, ignoring severity.

Output format

{
    "initial_drg": 293,
    "final_drg": 293,
    "initial_mdc": 5,
    "final_mdc": 5,
    "initial_drg_description": "Heart Failure and Shock without CC/MCC",
    "final_drg_description": "Heart Failure and Shock without CC/MCC",
    "initial_mdc_description": "Diseases and Disorders of the Circulatory System",
    "final_mdc_description": "Diseases and Disorders of the Circulatory System",
    "return_code": "OK",
    "pdx_output": {
        "code": "I5020",
        "mdc": 5,
        "severity": "CC",
        "drg_impact": "BOTH",
        "poa_error": "POA_NOT_CHECKED",
        "flags": ["VALID", "MARKED_FOR_INITIAL", "MARKED_FOR_FINAL"]
    },
    "sdx_output": [...],
    "proc_output": [...]
}

Supported DRG versions

Version	CMS Fiscal Year
400	FY 2023 (Oct 2022 – Apr 2023)
401	FY 2023 (Apr 2023 – Sep 2023)
410	FY 2024 (Oct 2023 – Apr 2024)
411	FY 2024 (Apr 2024 – Sep 2024)
420	FY 2025 (Oct 2024 – Apr 2025)
421	FY 2025 (Apr 2025 – Sep 2025)
430	FY 2026 (Oct 2025 – Apr 2026)
431	FY 2026 (Apr 2026 – Sep 2026)

Medicare Code Editor (MCE)

The MCE validates ICD diagnosis and procedure codes against CMS edit rules. It checks for sex conflicts, age conflicts, unacceptable principal diagnoses, E-codes as PDX, non-covered procedures, bilateral procedures, and more.

Input format

{
    "discharge_date": 20250101,  # YYYYMMDD integer (required for MCE)
    "icd_version": 10,           # 9 or 10 (default: 10)
    "age": 65,
    "sex": 0,                    # 0=Male, 1=Female, 2=Unknown
    "discharge_status": 1,
    "pdx": {"code": "I5020"},
    "admit_dx": {"code": "R0602"},
    "sdx": [{"code": "E1165"}],
    "procedures": [{"code": "02703DZ"}]
}

Output format

{
    "version": 20260930,
    "edit_type": "PREPAYMENT",    # NONE, PREPAYMENT, POSTPAYMENT, or BOTH
    "edits": [                    # List of triggered edits (empty if NONE)
        {
            "name": "E_CODE_AS_PDX",
            "count": 1,
            "code_type": "DIAGNOSIS",
            "edit_type": "PREPAYMENT"
        }
    ]
}

Example — E-code as principal diagnosis

import msdrg

with msdrg.MceEditor() as mce:
    result = mce.edit({
        "discharge_date": 20250101,
        "age": 65, "sex": 0, "discharge_status": 1,
        "pdx": {"code": "V0001XA"},  # E-code
        "sdx": [], "procedures": []
    })

print(result["edit_type"])  # "PREPAYMENT"
print(result["edits"][0]["name"])  # "E_CODE_AS_PDX"

Supported edit types

The MCE detects ~35 edit types including:

• INVALID_CODE — code not in CMS master for date range
• SEX_CONFLICT — code restricted by patient sex
• AGE_CONFLICT — code restricted by patient age
• E_CODE_AS_PDX — E-code used as principal diagnosis
• MANIFESTATION_AS_PDX — manifestation code used as PDX
• UNACCEPTABLE_PDX — code unacceptable as principal diagnosis
• NON_COVERED — procedure not covered by Medicare
• BILATERAL — bilateral procedure without bilateral PDX
• OPEN_BIOPSY — open biopsy without prior biopsy

MCE validation

The MCE implementation is validated against the CMS Java MCE 2.0 v43.1 with a 100% match rate on 50l,000 test claims.

Architecture

┌──────────────────────────────────────────────────────┐
│  Python (msdrg)                                      │
│  ctypes ──► C API (c_api.zig, mce_c_api.zig)        │
│                │                                     │
│    ┌───────────┴───────────┐                         │
│    ▼                       ▼                         │
│  MS-DRG Grouper         MCE Editor                  │
│  (GrouperChain)         (MceComponent)              │
│    │                       │                         │
│    ▼                       ▼                         │
│  Chain of Links:        Validation Pipeline:        │
│  Preprocess → Group     Code Check → Edit Rules     │
│  → HAC → Final DRG      → Output Counts            │
│    │                       │                         │
│    ▼                       ▼                         │
│  Memory-mapped .bin files (22 total)                │
└──────────────────────────────────────────────────────┘

Both engines share the same shared library and data files. The grouping pipeline is a chain of composable processors; the MCE is a linear validation pipeline. Both mirror the original Java architecture for validation purposes.

Building from source

Prerequisites

• Zig 0.16+ — download or via package manager
• Python 3.11+
• uv (recommended) or pip

Setup

git clone https://github.com/Bedrock-Billing/mz-drg.git
cd mz-drg

# Create venv and install
python3 -m venv .venv
source .venv/bin/activate
pip install -e .

This compiles the Zig shared library and bundles all data files into the Python package.

Run tests

# Zig unit tests (60+ tests)
cd zig_src && zig build test

# Python tests (MS-DRG + MCE)
python -m pytest tests/

Data pipeline

The binary data files (data/bin/*.bin) are prebuilt and included in the repository. To regenerate them from the raw CMS CSVs:

bash scripts/setup_data.sh

This runs extract → import → compile → zig build in sequence. See scripts/ for individual steps.

Comparison testing

The tests/ directory contains tools for validating mz-drg against the reference Java implementations.

# Generate random test claims
python tests/generate_test_claims.py --count 1000 --out tests/claims.json

# Compare MS-DRG grouper
python tests/compare_groupers.py --file tests/claims.json

# Compare MCE editor
python tests/compare_mce.py --file tests/claims.json

# Benchmark
python tests/compare_groupers.py --file tests/claims.json --benchmark

The Java comparisons require JDK 17+ and the reference JARs in jars/. This is only needed for validation — the Python package itself has no Java dependency.

C API

mz-drg exposes a C ABI for integration with any language. A complete header is auto-generated at zig-out/include/msdrg.h after building.

JSON API (simple — single call)

#include "msdrg.h"

void* ctx = msdrg_context_init("/path/to/data/bin");
const char* result = msdrg_group_json(ctx, "{\"version\":431,...}");
msdrg_string_free(result);
msdrg_context_free(ctx);

Structured API (no JSON — fine-grained control)

#include "msdrg.h"

MsdrgContext ctx = msdrg_context_init("/path/to/data/bin");
MsdrgVersion ver = msdrg_version_create(ctx, 431);
MsdrgInput inp = msdrg_input_create();

msdrg_input_set_pdx(inp, "I5020", 'Y');
msdrg_input_add_sdx(inp, "E1165", 'Y');
msdrg_input_set_demographics(inp, 65, 0, 1);

MsdrgResult res = msdrg_group(ver, inp);
int32_t drg = msdrg_result_get_final_drg(res);
int32_t mdc = msdrg_result_get_final_mdc(res);
const char* desc = msdrg_result_get_final_drg_description(res);

msdrg_result_free(res);
msdrg_input_free(inp);
msdrg_version_free(ver);
msdrg_context_free(ctx);

The structured API gives C/C++/Rust callers direct access to all 47 result fields without JSON parsing overhead. For Python, the JSON API (group()) is faster due to fewer FFI crossings. See zig-out/include/msdrg.h for the full function reference.

MCE Editor

#include "msdrg.h"

MceContext mce = mce_context_init("/path/to/data/bin");
const char* result = mce_edit_json(mce, "{\"discharge_date\":20250101,...}");
msdrg_string_free(result);
mce_context_free(mce);

Functions are thread-safe after initialization. The context is immutable and can be shared across threads.

Project structure

mz-drg/
├── msdrg/                       # Python package
│   ├── __init__.py
│   ├── grouper.py               # MsdrgGrouper class
│   └── mce.py                   # MceEditor class
├── zig_src/                     # Zig source
│   ├── build.zig
│   ├── main.zig
│   └── src/
│       ├── c_api.zig            # MS-DRG C ABI exports
│       ├── json_api.zig         # MS-DRG JSON in/out
│       ├── msdrg.zig            # GrouperChain + version routing
│       ├── chain.zig            # Composable processor chain
│       ├── models.zig           # Data models
│       ├── preprocess.zig       # Exclusion & attribute handling
│       ├── grouping.zig         # DRG formula matching
│       ├── marking.zig          # Code marking logic
│       ├── hac.zig              # Hospital-Acquired Conditions
│       ├── mce.zig              # MCE main editor
│       ├── mce_c_api.zig        # MCE C ABI exports
│       ├── mce_json_api.zig     # MCE JSON in/out
│       ├── mce_data.zig         # MCE data loading
│       ├── mce_enums.zig        # MCE attributes & edits
│       ├── mce_editing.zig      # MCE edit rules
│       └── mce_validation.zig   # MCE validation logic
├── data/bin/                    # Prebuilt binary data (22 files)
├── scripts/                     # Data extraction & compilation
├── tests/                       # Comparison & benchmark tools
├── pyproject.toml
└── setup.py

License

MIT — see LICENSE.

Documentation

Full documentation is available at Bedrock-Billing.github.io/mz-drg.

Acknowledgments

This project is intended for healthcare IT professionals who need fast, embeddable, and auditable claim processing tools.