​

EHR Project

Mining EHR in Kenya for Research Purposes 

Project Brief

Data in the 21st century is like oil in the 18th century, and within an LMIC setting can be an immensely valuable at identifying new innovative solutions. Within AKDN and AKU specifically, little to no work has been done to extract data from our hospital systems (EHR) and mine that data to understand trends and identify better care pathways for the communities we serve. This project aims to change that paradigm by extracting data from key hospital systems in Nairobi and Mombasa, organising the data in a manner optimised for analytics and machine learning and then making the repository available to AKU faculty for research purposes. The project will also build a similar repository for AKHS in Mombasa for them to use for their operations.

 
​

Goals:

We plan to focus on the following goals:

  1. Build an analytic data repository at AKU with data from Care2000 systems in Nairobi and Mombasa to be used for research purposes.

  2. Work with team at AKHS Mombasa to build similar repository for them to use internally with data from Care2000 in Mombasa.

  3. Build a community of users on the AKU repository including Informatics office, Research faculty within AKU, and other AKU teams as needed.

  4. Mine the data and produce 2 – 3 data packages that provide key trends, findings and insights that highlight the value of the investment.

Tenets:

The following tenets are guiding principles which will be used to evaluate the approaches and prioritise activities:

  • Small cross functional team comprised of clinicians and data architects/ engineers.

  • Tangible output and outcomes delivered at the end of every 6-week sprint. 

  • Use of open source/ open data algorithms vs. proprietary/ custom code.

  • Development code/ scripts should be crafted in a manner to be replicable across the Care2000 systems i.e., the same script applied in Nairobi should run in Mombasa.

Approach:

The project will be executed using agile methodologies using a 6-week sprint cycle.

The project will kick off with an Explore/ Planning sprint that will determine key architectural elements including:

  • Detailed understanding of the Care2000 hospital system including data relationships, data quality, and types of data
  • High level understanding of the data model to be used. 
  • Alignment with clinicians on how data will be extracted to align with having a complete working model at the end of each sprint
  • Detailed understanding of what can be done in compliance with Kenya Data Protection Laws.

Key milestones so far:

Significant progress has been made in expanding our EHR data repository. We have successfully integrated all data from the new EHR system, Meditech Expanse, into our existing repository. Data is refreshed every 96 hours, ensuring near real-time updates. The consolidation of legacy systems with Meditech into a single, standardized data model is already enabling impactful research, including:

  • A study on gestational diabetes (GDM) prevalence and risk factors in Kenyan women.
  • A newly launched sepsis research initiative with the CMIO office and CITRIC in Karachi.
  • An experimental hypertension study leveraging Generative AI in collaboration with Google.​