A method for generating synthetic longitudinal health data

Abstract

Getting access to administrative health data for research purposes is a difficult and time-consuming process due to increasingly demanding privacy regulations. An alternative method for sharing administrative health data would be to share synthetic datasets where the records do not correspond to real individuals, but the patterns and relationships seen in the data are reproduced. This paper assesses the feasibility of generating synthetic administrative health data using a recurrent deep learning model. Our data comes from 120,000 individuals from Alberta Health’s administrative health database. We assess how similar our synthetic data is to the real data using utility assessments that assess the structure and general patterns in the data as well as by recreating a specific analysis in the real data commonly applied to this type of administrative health data. We also assess the privacy risks associated with the use of this synthetic dataset. Generic utility assessments that used Hellinger distance to quantify the difference in distributions between real and synthetic datasets for event types (0.027), attributes (mean 0.0417), Markov transition matrices (order 1 mean absolute difference: 0.0896, sd: 0.159; order 2: mean Hellinger distance 0.2195, sd: 0.2724), the Hellinger distance between the joint distributions was 0.352, and the similarity of random cohorts generated from real and synthetic data had a mean Hellinger distance of 0.3 and mean Euclidean distance of 0.064, indicating small differences between the distributions in the real data and the synthetic data. By applying a realistic analysis to both real and synthetic datasets, Cox regression hazard ratios achieved a mean confidence interval overlap of 68% for adjusted hazard ratios among 5 key outcomes of interest, indicating synthetic data produces similar analytic results to real data. The privacy assessment concluded that the attribution disclosure risk associated with this synthetic dataset was substantially less than the typical 0.09 acceptable risk threshold. Based on these metrics our results show that our synthetic data is suitably similar to the real data and could be shared for research purposes thereby alleviating concerns associated with the sharing of real data in some circumstances.

A comparison of synthetic data generation and federated analysis for enabling international evaluations of cardiovascular health

“…The analytical conclusions were the same for the models developed using the pooled partially synthetic dataset as the ground truth model developed using federated analysis in various analytical steps including descriptive, univariable analysis and multivariable main effects and country interaction models. While previous observational studies have compared synthetic and real data 34 – 36 , there has been no population-based study testing the use of SDG for pooling datasets across jurisdictions and comparing it to a federated approach.…”

On the evaluation of synthetic longitudinal electronic health records

“…Note that other models have been developed which are able to generate EHRs with static attributes and sequential data [9][10][11]29]. We opt for DoppelGANger and CPAR since they are contained in easy-to-use open-source libraries, promoting reproducibility of this research.…”

“…, such that approximating p(z|X) can be framed as a simple binary classification task, since the optimization problems in Equation (11) and Equation ( 12) are equivalent. In this case, note that we make a continuous approximation of the binary latent variable on the (0,1) interval.…”

“…Recent advancements in generative modelling bring forth model architectures which are able to learn and generate synthetic samples for almost any data structure [8]. Likewise, many model architectures have been developed which are able to generate synthetic longitudinal EHRs [9][10][11]. However, literature on evaluating the quality of generated synthetic data is lagging behind.…”

New Approach for Generating Synthetic Medical Data to Predict Type 2 Diabetes

“…Therefore, the methods proposed for generating synthetic medical data differ. For example, generational neural diffusion models, variation autoencoders, and a generative adversarial network (GAN) are mainly used to develop synthetic medical images/data [ 6 , 7 , 8 ], whereas algorithms such as Bayesian networks [ 9 ] and classification and regression trees [ 10 , 11 ] are used to develop numerical (quantitative) and non-numerical (qualitative), and recurrent deep learning models are used to build time-series databases [ 12 ]. Because this study aims to develop a numerical database, the issues of generating images and developing non-numerical databases are beyond the scope of this study.…”