Healthcare interoperability project analyzing clinical data using industry-standard medical coding systems across 1,180 synthetic patients.
This project explores how healthcare data is standardized and analyzed using real-world coding systems used in US hospitals, payers, and EHR platforms. Built using Synthea synthetic FHIR R4 data — no real patient data, fully HIPAA-safe.
The analysis covers the full clinical data lifecycle — from raw FHIR resources through terminology mapping, lab standardization, and medication coding, and population health insights.
| Standard | Purpose | Records Analyzed |
|---|---|---|
| HL7 FHIR R4 | Interoperability & data exchange | All resources |
| SNOMED-CT | Clinical condition terminology | 8,766 conditions |
| ICD-10-CM | Diagnosis coding & billing | 4,327 mapped codes |
| LOINC | Lab results & vitals standardization | All observations |
| RxNorm | Medication naming & e-prescribing | All prescriptions |
| HIPAA | Privacy — Safe Harbor approach | Synthetic data only |
Section 1 — Data Loading Loads 1,180 Synthea FHIR R4 patient bundles and inventories all resource types across the dataset.
Section 2 — SNOMED-CT to ICD-10 Crosswalk Maps SNOMED-CT clinical terminology to ICD-10-CM diagnosis codes — the same process used in production EHR systems like Epic and Cerner when exchanging data between clinical and billing systems.
Section 3 — ICD-10 Disease Burden Analysis Analyzes population-level disease patterns using ICD-10 chapter classifications. Identifies top diagnoses, chronic disease prevalence, and onset trends over time.
Section 4 — LOINC Lab & Vitals Standardization Extracts and categorizes LOINC-coded observations including vitals, metabolic panels, lipid panels, CBC, renal function, and liver function. Visualizes distributions against clinical reference ranges.
Section 5 — RxNorm Medication Analytics Analyzes prescription patterns by drug class and therapeutic category. Covers cardiovascular, diabetes, mental health, respiratory, and antibiotic medications with trend analysis over time.
Section 6 — Standards Coverage Summary End-to-end summary of all standards demonstrated, data volumes processed, and patient clinical complexity metrics.
- Most common diagnosis: Viral Infection (J09.X9) — 1,237 patients
- Highest disease burden: Respiratory conditions
- Most prescribed category: Cardiovascular medications
- SNOMED → ICD-10 mapping coverage: 49.4% (4,327 of 8,766 conditions)
- Average conditions per patient: varies across chronic disease cohorts
Python 3.10 pandas numpy matplotlib seaborn json pathlib
Synthea™ — Synthetic Patient Population Simulator by The MITRE Corporation
👉 synthea.mitre.org
# Clone the repo
git clone https://github.com/nipa-analytics/icd10-medical-coding-analytics.git
cd icd10-medical-coding-analytics
# Create environment
conda create -n fhir_env python=3.10
conda activate fhir_env
# Install dependencies
pip install pandas numpy matplotlib seaborn jupyter
# Download Synthea data
# Go to synthea.mitre.org/downloads
# Download FHIR R4 sample → place JSON files in data/fhir/fhir/
# Launch notebook
jupyter notebook
# Open: icd10_medical_coding_analytics.ipynb| File | Description |
|---|---|
icd10_disease_burden.png |
Top ICD-10 diagnoses + disease burden by chapter |
loinc_vitals_dashboard.png |
12-panel vitals & labs with clinical reference ranges |
loinc_category_summary.png |
LOINC category distribution |
rxnorm_medication_dashboard.png |
Prescription patterns by drug class |
standards_summary.png |
Records processed per standard |
patient_complexity.png |
Conditions, meds, observations per patient |
- FHIR Healthcare Analytics
- Patient Segmentation — MIMIC-IV
- 30-Day Readmission Risk — MIMIC-IV
- ICU Deterioration Prediction
Nipa Shah — Data Scientist | Healthcare Analytics | AI/ML
📍 Jersey City, NJ
🔗 LinkedIn
🐙 GitHub
📧 nipashah2311@gmail.com