Endocrine Abstracts

JOINT258

Background: The utilisation of data is revolutionising the operations and impact of non-profit organisations (NGOs) worldwide. The effectiveness of NGOs is increasingly contingent upon their capacity to collect, analyze, and apply data at every level of their operations. Action4Diabetes (A4D) is a UK diabetes charity that is making sustainable and scalable progress in the battle to provide quality type 1 diabetes healthcare in emerging countries across South-East Asia (SEA). CorrelAid is a nonprofit organization of data scientists dedicated to improving the world through the application of data science. CorrelAid e. V. is the registered nonprofit association behind the CorrelAid community. CorrelAid and A4D have collaborated to develop a centralized database in the public cloud, which provides access to all historical data in a single repository.

Methods: A4D is exchanging data with the local hospitals in the programme on a monthly basis via Microsoft Excel files. A preprocessing pipeline was implemented to account for differences between files from different hospitals and differences over time for the same hospital. The pipeline extracts the patient and medical product data in a standardized and unified manner, rendering it suitable for storage in a database. The data is anonymized and subsequently uploaded to a secure storage in the public cloud, where it is processed and stored in a centralized database. An error reporting system is in place to monitor the pipeline’s progress and alert in the event of critical errors or data quality issues.

Results: The public cloud database provides access to all historical data for the years 2017 to 2023. The database encompasses six countries, 31 clinics, and approximately 1, 000 patients along with their monthly medical data. Furthermore, the database contains information about the distribution and usage of approximately 200 medical products.

Conclusions: The formulation of a data strategy and the implementation of a centralized database solution for healthcare data is a challenging but achievable objective when approached in a cost-effective manner, particularly when utilizing services from modern public cloud providers. The subsequent stages of the process entail the identification of initial use cases that build upon the data, the implementation of dashboards and reports that facilitate access to the data, the automation of the data pipeline to ingest newly arriving data without manual intervention, and the establishment of a data governance strategy to oversee data content, structure, usage, and safety.