Ensuring electronic medical record simulation through better training, modeling, and evaluation.

OBJECTIVE: Electronic medical records (EMRs) can support medical research and discovery, but privacy risks limit the sharing of such data on a wide scale. Various approaches have been developed to mitigate risk, including record simulation via generative adversarial networks (GANs). While showing promise in certain application domains, GANs lack a principled approach for EMR data that induces subpar simulation. In this article, we improve EMR simulation through a novel pipeline that (1) enhances the learning model, (2) incorporates evaluation criteria for data utility that informs learning, and (3) refines the training process. MATERIALS AND METHODS: We propose a new electronic health record generator using a GAN with a Wasserstein divergence and layer normalization techniques. We designed 2 utility measures to characterize similarity in the structural properties of real and simulated EMRs in the original and latent space, respectively. We applied a filtering strategy to enhance GAN training for low-prevalence clinical concepts. We evaluated the new and existing GANs with utility and privacy measures (membership and disclosure attacks) using billing codes from over 1 million EMRs at Vanderbilt University Medical Center. RESULTS: The proposed model outperformed the state-of-the-art approaches with significant improvement in retaining the nature of real records, including prediction performance and structural properties, without sacrificing privacy. Additionally, the filtering strategy achieved higher utility when the EMR training dataset was small. CONCLUSIONS: These findings illustrate that EMR simulation through GANs can be substantially improved through more appropriate training, modeling, and evaluation criteria.

A Distributed Framework for the Construction of Transport Maps.

The need to reason about uncertainty in large, complex, and multimodal data sets has become increasingly common across modern scientific environments. The ability to transform samples from one distribution P to another distribution Q enables the solution to many problems in machine learning (e.g., Bayesian inference, generative modeling) and has been actively pursued from theoretical, computational, and application perspectives across the fields of information theory, computer science, and biology. Performing such transformations in general still leads to computational difficulties, especially in high dimensions. Here, we consider the problem of computing such "measure transport maps" with efficient and parallelizable methods. Under the mild assumptions that P need not be known but can be sampled from and that the density of Q is known up to a proportionality constant, and that Q is log-concave, we provide in this work a convex optimization problem pertaining to relative entropy minimization. We show how an empirical minimization formulation and polynomial chaos map parameterization can allow for learning a transport map between P and Q with distributed and scalable methods. We also leverage findings from nonequilibrium thermodynamics to represent the transport map as a composition of simpler maps, each of which is learned sequentially with a transport cost regularized version of the aforementioned problem formulation. We provide examples of our framework within the context of Bayesian inference for the Boston housing data set and generative modeling for handwritten digit images from the MNIST data set.

The Use of Cardiac Orienting Responses as an Early and Scalable Biomarker of Alcohol-Related Neurodevelopmental Impairment.

BACKGROUND: Considered the leading cause of developmental disabilities worldwide, fetal alcohol spectrum disorders (FASD) are a global health problem. To take advantage of neural plasticity, early identification of affected infants is critical. The cardiac orienting response (COR) has been shown to be sensitive to the effects of prenatal alcohol exposure and is an inexpensive, easy to administer assessment tool. The purpose of this study was to evaluate the COR effectiveness in assessing individual risk of developmental delay. METHODS: As part of an ongoing longitudinal cohort study in Ukraine, live-born infants of women with some to heavy amounts of alcohol consumption in pregnancy were recruited and compared to infants of women who consumed low or no alcohol. At 6 and 12 months, infants were evaluated with the Bayley Scales of Infant Development-II. CORs were also collected during a habituation/dishabituation learning paradigm. Using a supervised logistic regression classifier, we compared the predictive utility of the COR indices to that of the 6-month Bayley scores for identification of developmental delay based on 12-month Bayley scores. Heart rate collected at each second (Standard COR) was compared to key features (Key COR) extracted from the response. RESULTS: Negative predictive values (NPV) were 85% for Standard COR, 82% for Key COR, and 77% for the Bayley, and positive predictive values (PPV) were 66% for Standard COR, 62% for Key COR, and 43% for the Bayley. CONCLUSIONS: Predictive analysis based on the COR resulted in better NPV and PPV than the 6-month Bayley score. As the resources required to obtain a Bayley score are substantially more than in a COR-based paradigm, the findings are suggestive of its utility as an early scalable screening tool based on the COR. Further work is needed to test its long-term predictive accuracy.