Synthetic Data Generation of Health and Demographic Surveillance Systems Dataset, Kenya, 2019-2020

Surveillance data play a vital role in estimating the burden of diseases, pathogens, exposures, behaviors, and susceptibility in populations, providing insights that can inform the design of policies and targeted public health interventions. The use of Health and Demographic Surveillance System (HDSS) collected from the Kilifi region of Kenya, has led to the collection of massive amounts of data on the demographics and health events of different populations. This has necessitated the adoption of tools and techniques to enhance data analysis to derive insights that will improve the accuracy and efficiency of decision-making. Machine Learning (ML) and artificial intelligence (AI) based techniques are promising for extracting insights from HDSS data, given their ability to capture complex relationships and interactions in data. However, broad utilization of HDSS datasets using AI/ML is currently challenging as most of these datasets are not AI-ready due to factors that include, but are not limited to, regulatory concerns around privacy and confidentiality, heterogeneity in data laws across countries limiting the accessibility of data, and a lack of sufficient datasets for training AI/ML models. Synthetic data generation offers a potential strategy to enhance accessibility of datasets by creating synthetic datasets that uphold privacy and confidentiality, suitable for training AI/ML models and can also augment existing AI datasets used to train the AI/ML models. These synthetic datasets, generated from two rounds of separate data collection periods, represent a version of the real data while retaining the relationships inherent in the data. For more information please visit The Aga Khan University Website.