OHDSI

Observational Health Data Sciences

and Informatics

OHDSI Africa

Standardizing African Health Data for

Federated, Privacy-First Research

DS-I Africa Virtual Networking Exchange · 8 April 2026

Presenters: Daniel Nsanzabandi & Namanya Abert

�Leads, OHDSI Africa ETL Workgroup · Kigali, Kampala

What Is OHDSI?

A global open-science collaborative founded in 2013

Mission: To improve health by empowering a community to collaboratively generate the evidence that promotes better health decisions and better care.

300+

Partner institutions globally

80+

Countries represented

810M+

Unique patient records in CDM

FREE

All tools — open-source forever

Core Principle: Patient data NEVER leaves the institution. Each site runs the same open-source analysis code locally on their own database and shares only aggregate, anonymized results. Full data sovereignty always.

Why OMOP CDM? The Case for Africa

The problem OMOP solves — and why it matters here

The Problem Without OMOP

• Every health facility stores data in a different structure, language, and format
• Malaria coded as 'Palu', 'B54', 'MAL', 'Malaria NOS' across different facilities
• A researcher cannot compare results across Rwanda, Kenya, and Ethiopia without manual harmonisation — which takes months
• 80% of research time is spent cleaning and reformatting data, not doing science
• African data rarely appears in global research because it is too fragmented to use

What OMOP CDM Solves

• One agreed table structure: person, visit, condition, drug, measurement — across all sites
• One agreed vocabulary: SNOMED CT, RxNorm, LOINC — so malaria is always concept_id 4340390, everywhere
• A researcher writes the analysis code ONCE and every OMOP site can run it the same day
• Results from Rwanda, Kenya, and Ethiopia are directly comparable — without sharing any patient records
• African data can participate in global studies on equal terms with high-income country datasets

How OHDSI Works: Federated, Privacy-First Research

Source Data

(EHR / HMIS / DHIS2)

ETL Pipeline

(Transform & Standardize)

OMOP CDM

(Local Database)

Analysis Code

(Runs Locally)

Only RESULTS shared

(aggregate, anonymized)

Rwanda — RBC RIDS Division

proposed OMOP site

OMOP CDM

Kenya — KEMRI collaborators

ETL design in progress

OMOP CDM

Ethiopia — EPHI / SmartCare

data profiling underway

OMOP CDM

Each site runs identical open-source code on its OWN local data. Patient records never cross any border.

OMOP CDM v5.4 and the OHDSI Toolstack

OMOP CDM v5.4 Key Domains

• Person — demographics, year of birth, gender
• Visit Occurrence — each clinical encounter/admission
• Condition Occurrence — diagnoses (mapped to SNOMED CT)
• Drug Exposure — prescriptions (mapped to RxNorm)
• Measurement — labs and vitals (mapped to LOINC)
• Observation — other clinical facts and findings
• All records linked by a universal person_id across domains

OHDSI Open-Source Tools (all free)

WhiteRabbit & Rabit-in-a-hat

Profile source data before and Design the ETL

USAGI

Map local codes to OMOP concepts

Achilles

Characterize and profile your CDM

DQD

3,000+ automated data quality checks

ATLAS

Define cohorts, design studies

HADES

Population-level evidence in R

Any institution can adopt the full toolstack without licensing fees — all code is on GitHub, all documentation is open.

OHDSI Africa: Who We Are

A growing continental network — active sites, real momentum

Active & Developing OHDSI Africa Sites

Proposed Site

Rwanda

RBC Research, Innovation & Data Science (RIDS) Division

OMOP CDM pipeline under development · ETL Workgroup Lead

Active

Kenya

KEMRI / Academic health system collaborators

ETL design in progress using Rabbit-in-a-Hat

Active

Ethiopia

EPHI / SmartCare EMR collaborators

Data profiling with WhiteRabbit completed

Active

South Africa

SAMRC / Academic medical centre partners

Vocabulary mapping underway for clinical data

Growing

Uganda

Makerere University School of Public Health (MUSPH)

Exploratory discussions underway

Growing

Nigeria

Lagos University Teaching Hospital (LUTH) + partners

Community of practice being established

First OHDSI Africa Training — Rwanda, March 2026

A milestone for African health data standardization capacity

Milestone: In March 2026, OHDSI Africa delivered its first-ever Data Standardization Training on the continent hosted at the Rwanda Biomedical Centre (RBC), Kigali.

What the Training Covered

• Introduction to OMOP CDM v5.4 — structure, purpose, and global adoption
• International health vocabularies — SNOMED CT, RxNorm, LOINC, ICD-10
• OHDSI tools: WhiteRabbit, Rabbit-in-a-Hat, USAGI, Achilles, DQD, ATLAS
• Hands-on ETL concepts: source data profiling, field mapping design
• Data quality fundamentals — plausibility, conformance, completeness
• Cohort analysis basics in ATLAS — defining research-ready patient groups

Key Outcomes

2

weeks of intensive training

10+

RBC technical staff trained

100+

OMOP concept mappings completed

2nd

OHDSI Africa training this year 2026

These reusable materials — 10 presentations, 5 workbooks, setup guides — are available for any OHDSI Africa partner institution.

OHDSI Africa ETL Workgroup: Current Status

Coordinating CDM standardization across the continent — 2025–2026

What the Workgroup Does

• Coordinates OMOP CDM implementation across African research sites
• Develops and shares open ETL templates, scripts, and vocabulary mapping files via GitHub
• Runs weekly OHDSI Africa workgroup calls — all ETL members
• Developed ETL automated ETL pipeline
• Provides technical mentorship to new sites joining the network.
• Advocates for Africa-specific concepts in the global OHDSI vocabulary

Site Progress (2025–2026)

Rwanda

OMOP CDM pipeline built on sample data; training delivered

55%

Kenya

OMOP analysis tools built by APHRC — integration & CDM validation in progress

40%

ETL process: Manual vs Automated

OHDSI Africa ETL Workgroup — OKRs 2026

5 objectives driving capacity, production, and community across the continent

Purpose: Build ETL capacity across Africa — developing OpenMRS→OMOP training programs, creating production implementations, and establishing a self-sustaining network of skilled OMOP practitioners.

OBJ 1

Map OpenMRS → OMOP CDM

• KR 1.1: Complete validated OpenMRS→OMOP ETL pipeline on GitHub — Nov 2026
• KR 1.2: Document 15+ core OpenMRS modules mapped to OMOP domains

OBJ 2

Develop ETL Training Programme

• KR 2.1: Publish shareable curriculum guides, exercises, assessments (Mar–Apr 2026)
• KR 2.3: Produce 10–15 instructional ETL training.

OBJ 3

Increase Visibility & Knowledge Sharing

• KR 3.1: Present at 3+ African health data conferences / OHDSI Africa Symposium
• KR 3.2: Conduct 2–3 hands-on ETL workshops across different African regions.

OBJ 4

Build ETL Capacity — 20+ Practitioners

• KR 4.1: Recruit 20+ trainees from 8+ countries, 3+ per region (Jun–Aug 2026)
• KR 4.2: Weekly office hours for ongoing support (Jun–Dec 2026)
• KR 4.3: Launch peer learning community — bi-weekly calls, Teams channel

OBJ 5

First Production ETL Implementation

• KR 5.1: Convert 1 full OpenMRS site dataset to OMOP CDM — site using data for analytics (Nov 2026)
• KR 5.2: Publish replicable case study, lessons learned, and replication toolkit

UPCOMING EVENT

OHDSI Africa Symposium 2026

Location

Addis Ababa, Ethiopia

Dates

September 28–30, 2026

Audience

OHDSI Africa network members

+ invited researchers & partners

What to Expect

Progress Presentations

Each OHDSI Africa site presents their CDM implementation status and early research findings

Federated Study Results

First results from the multi-country OHDSI Africa network studies shared across participating sites

Capacity Building

Hands-on workshops on advanced ATLAS cohort analysis, DQD interpretation, and ETL best practices

Partnership Forum

Open sessions for new institutions to learn how to join, funder engagement, and collaboration planning

The Terminology Gap: Africa's Hidden ETL Challenge

Why mapping local clinical terms to international standards is harder than it looks

The Problem: African health systems document clinical data in local languages, abbreviations, and non-standard terms. When mapping to SNOMED CT, RxNorm, and LOINC, many concepts simply don't exist — or the closest match is clinically misleading. This is one of the biggest practical barriers to African data joining global research networks.

Diagnoses → SNOMED CT

• "Palu" — Swahili/French for malaria; not in SNOMED
• "Mal.Sev." — local abbreviation for severe malaria
• "Paludisme compliqué" — French term, no direct match
• "Fever NOS" — maps to 12+ different SNOMED concepts
• "URTI" — no single standard concept equivalent
• "CCF" — Congestive Cardiac Failure abbreviation

Drugs → RxNorm

• "Artesunate IV" vs "Artesunate 2.4mg/kg" — same drug, different entries
• "Paracetamol" ≠ "Acetaminophen" — RxNorm uses US names
• "Coartem" — brand name, must map to ingredient pair
• Many local generics have no RxNorm concept ID at all
• "Quinine ampoule" — dose strength not captured
• "SP" (Sulfadoxine-Pyrimethamine) — combination drug

Lab Tests → LOINC

• "Hgb" vs "Haemoglobin" — same test, 3 different LOINC codes
• "Malaria RDT" — qualitative vs quantitative LOINC code differ
• "Blood glucose" — fasting, random, or OGTT? Different codes
• "MUAC" — Mid-Upper Arm Circumference has NO LOINC code
• "Malaria smear" — species, method, and specimen all matter
• "BNP" — specimen type changes the LOINC code entirely

Addressing the Terminology Gap: Our Three-Layer Strategy

No single tool solves this — we use a layered approach that combines automation, African-specific resources, and community contribution.

Layer 1 — USAGI Mapping Tool

• OHDSI's semi-automated mapping tool
• Algorithm scores similarity between your local term and all SNOMED/RxNorm/LOINC concepts
• Data steward reviews and approves each suggested mapping
• Handles spelling variants, abbreviations, French and Kinyarwanda terms
• RBC completed 100+ concept mappings during the March 2026 training
• Export: approved CSV file feeds directly into your ETL pipeline

Layer 2 — CIEL Reference Terminology

• Columbia International eHealth Lab — built specifically for African EMR systems (OpenMRS/eBuzima)
• Maps common African clinical terms directly to SNOMED CT and LOINC
• Covers Kinyarwanda, Swahili, and French clinical terminology
• Already integrated into OpenMRS deployments across Africa
• Acts as a translation bridge: local code → CIEL concept → SNOMED/LOINC
• Maintained by Columbia University with ongoing African community contribution

Layer 3 — Custom Local Concepts

• OMOP CDM reserves concept_id 2B domain for locally-defined concepts
• Used when no SNOMED/RxNorm/LOINC concept exists anywhere
• Examples: MUAC measurements, malaria RDT result subtypes, CHW visit types
• OHDSI Africa workgroup maintains a shared repository of custom concepts
• Custom concepts submitted to OHDSI vocabulary team for potential global inclusion
• Ensures zero data is lost — even unmapped terms are preserved with source value

What OHDSI Africa Offers Research Partners

Infrastructure · Methods · Capacity · Community — all open-source, all free

01

Federated Research Infrastructure

Run multi-country studies on standardized CDM data across Rwanda, Kenya, Ethiopia, and South Africa without any patient record crossing a border. Same code, comparable results, full sovereignty.

02

ETL Templates & Training Curriculum

Open-source ETL scripts, vocabulary mapping files, and the full 2-week OHDSI Data Standardization Training curriculum reusable by any African institution to build their own OMOP CDM.

03

Data Quality Assurance Framework

Achilles characterization and DQD (3,000+ automated checks) as standard practice for every site. Any partner joining our network can be quality-validated to international OHDSI standards.

04

Africa-Specific Terminology Library

Our growing library of CIEL-bridged and custom OMOP concepts covers local African clinical terms, drug names, and community health indicators reducing the mapping burden for new sites.

05

OMOP CDM Technical Expertise

We can support your institution to convert its EHR, HMIS, or DHIS2 data to OMOP CDM from WhiteRabbit source profiling through to production pipeline with DQD validation.

06

Global OHDSI Network Access

Joining OHDSI Africa connects you to the global OHDSI network of 300+ institutions opening access to federated studies, the HADES methods library, and the annual OHDSI Global Symposium.

What We Seek From Research Partners

Building together — not in parallel

We are looking for partners who bring:

Health Data Sources

Institutions with longitudinal EHR, HMIS, or DHIS2 datasets who want to standardize their data and join the OHDSI Africa federated network — especially in countries we do not yet cover.

Data Science Capacity

Universities and research institutes with data science, bioinformatics, or epidemiology expertise who can support ETL engineering, vocabulary development, and study design across sites.

Clinical Research Expertise

Clinicians, epidemiologists, and public health researchers who can define meaningful research questions that take full advantage of OMOP CDM data across multiple African sites.

Funding Partners

Funders committed to scalable, open-source health data infrastructure for Africa. Our model is highly replicable — once the tooling is in place, the cost per new site is low.

Terminology Contributors

Partners who can help expand our Africa-specific concept library — particularly for community health indicators, MUAC, verbal autopsy, nutrition surveillance, and local medicine terms.

Training Collaborators

Organizations that want to co-deliver or host the OHDSI Africa Data Standardization Training in their country or institution to grow CDM capacity across the continent.

Join OHDSI Africa

Building the Future of Health Data Science in Africa — Together

Get Started

Visit ohdsi.org/join-ohdsi

Contact us to join the Africa ETL Workgroup

We guide you through the full setup — free

Collaborate

Have a dataset? Let's talk

Have a research question? We have the infrastructure

Want to train your team? We have the curriculum

Ethiopia 2026

Join us at the OHDSI Africa Symposium

September 28–30, 2026

Meet the full Africa network in person

Daniel Nsanzabandi | Data Scientist, RBC | Namanya Abert |Data Scientist, OHDSI AFRICA | Leads, OHDSI Africa ETL Workgroup

"Patient data stays local — only knowledge crosses borders"

ohdsi.org · ohdsi.org/ohdsi-africa · danielnsanzabandi@gmail.com · Ethiopia Symposium: Sep 28–30, 2026