Search

SummaryObjectiveElectronic health record (EHR) systems contain a wealth of clinical data stored as both codified data and free-text narrative notes, covering hundreds of thousands of clinical concepts available for research and clinical care. The complex, massive, heterogeneous, and noisy nature of EHR data imposes significant challenges for feature representation, information extraction, and uncertainty quantification. To address these challenges, we proposed an efficientAggregated naRrativeCodifiedHealth (ARCH) records analysis to generate a large-scale knowledge graph (KG) for a comprehensive set of EHR codified and narrative features.MethodsThe ARCH algorithm first derives embedding vectors from a co-occurrence matrix of all EHR concepts and then generates cosine similarities along with associatedp-values to measure the strength of relatedness between clinical features with statistical certainty quantification. In the final step, ARCH performs a sparse embedding regression to remove indirect linkage between entity pairs. We validated the clinical utility of the ARCH knowledge graph, generated from 12.5 million patients in the Veterans Affairs (VA) healthcare system, through downstream tasks including detecting known relationships between entity pairs, predicting drug side effects, disease phenotyping, as well as sub-typing Alzheimer’s disease patients.ResultsARCH produces high-quality clinical embeddings and KG for over 60, 000 EHR concepts, as visualized in the R-shiny powered web-API (https://celehs.hms.harvard.edu/ARCH/). The ARCH embeddings attained an average area under the ROC curve (AUC) of 0.926 and 0.861 for detecting pairs of similar EHR concepts when the concepts are mapped to codified data and to NLP data; and 0.810 (codified) and 0.843 (NLP) for detecting related pairs. Based on thep-values computed by ARCH, the sensitivity of detecting similar and related entity pairs are 0.906 and 0.888 under false discovery rate (FDR) control of 5%. For detecting drug side effects, the cosine similarity based on the ARCH semantic representations achieved an AUC of 0.723 while the AUC improved to 0.826 after few-shot training via minimizing the loss function on the training data set. Incorporating NLP data substantially improved the ability to detect side effects in the EHR. For example, based on unsupervised ARCH embeddings, the power of detecting drug-side effects pairs when using codified data only was 0.15, much lower than the power of 0.51 when using both codified and NLP concepts. Compared to existing large-scale representation learning methods including PubmedBERT, BioBERT and SAPBERT, ARCH attains the most robust performance and substantially higher accuracy in detecting these relationships. Incorporating ARCH selected features in weakly supervised phenotyping algorithms can improve the robustness of algorithm performance, especially for diseases that benefit from NLP features as supporting evidence. For example, the phenotyping algorithm for depression attained an AUC of 0.927 when using ARCH selected features but only 0.857 when using codified features selected via the KESER network[1]. In addition, embeddings and knowledge graphs generated from the ARCH network were able to cluster AD patients into two subgroups, where the fast progression subgroup had a much higher mortality rate.ConclusionsThe proposed ARCH algorithm generates large-scale high-quality semantic representations and knowledge graph for both codified and NLP EHR features, useful for a wide range of predictive modeling tasks.

MotivationPredicting molecule–disease indications and side effects is important for drug development and pharmacovigilance. Comprehensively mining molecule–molecule, molecule–disease and disease–disease semantic dependencies can potentially improve prediction performance.MethodsWe introduce a Multi-Modal REpresentation Mapping Approach to Predicting molecular-disease relations (M2REMAP) by incorporating clinical semantics learned from electronic health records (EHR) of 12.6 million patients. Specifically, M2REMAP first learns a multimodal molecule representation that synthesizes chemical property and clinical semantic information by mapping molecule chemicals via a deep neural network onto the clinical semantic embedding space shared by drugs, diseases and other common clinical concepts. To infer molecule–disease relations, M2REMAP combines multimodal molecule representation and disease semantic embedding to jointly infer indications and side effects.ResultsWe extensively evaluate M2REMAP on molecule indications, side effects and interactions. Results show that incorporating EHR embeddings improves performance significantly, for example, attaining an improvement over the baseline models by 23.6% in PRC-AUC on indications and 23.9% on side effects. Further, M2REMAP overcomes the limitation of existing methods and effectively predicts drugs for novel diseases and emerging pathogens.Availability and implementationThe code is available at https://github.com/celehs/M2REMAP, and prediction results are provided at https://shiny.parse-health.org/drugs-diseases-dev/.Supplementary informationSupplementary data are available at Bioinformatics online.

ObjectiveTo discover pharmacotherapy prescription patterns and their statistical associations with outcomes through a clinical pathway inference framework applied on real-world data.Materials and MethodsWe apply machine learning steps in our framework using a 2006 to 2020 cohort of veterans with major depressive disorder (MDD). Outpatient antidepressant pharmacy and emergency department visits, self-harm, and all-cause mortality data were extracted from the Department of Veterans Affairs Corporate Data Warehouse.ResultsOur MDD cohort consisted of 252,179 individuals. During the study period there were 98,417 cases of emergency department visits, 1,016 cases of self-harm, and 1,507 deaths from all causes. The top ten prescription patterns accounted for 69.3% of the data for individuals starting antidepressants at the fluoxetine equivalent of 20-39mg. Additionally, we found associations between outcomes and dosage change.DiscussionFor 252,179 Veterans who served in Iraq and Afghanistan with subsequent MDD noted in their electronic medical record, we documented and described the major pharmacotherapy prescription patterns implemented by VHA providers. Ten patterns accounted for almost 70% of the data. Associations between antidepressant usage and outcomes in observational data may be confounded. The low numbers of adverse events especially associated with all-cause mortality make our calculations imprecise. Furthermore, our outcomes are also the indications for both disease and treatment. Despite these limitations, we demonstrate the usefulness of our framework in providing operational insight into clinical practice, and our results underscore the need for increased monitoring during critical points of treatment.

The growing availability of electronic health records (EHR) data opens opportunities for integrative analysis of multi-institutional EHR to produce generalizable knowledge. A key barrier to such integrative analyses is the lack of semantic interoperability across different institutions due to coding differences. We propose a Multiview Incomplete Knowledge Graph Integration (MIKGI) algorithm to integrate information from multiple sources with partially overlapping EHR concept codes to enable translations between healthcare systems.The MIKGI algorithm combines knowledge graph information from (i) embeddings trained from the co-occurrence patterns of medical codes within each EHR system and (ii) semantic embeddings of the textual strings of all medical codes obtained from the Self-Aligning Pretrained BERT (SAPBERT) algorithm. Due to the heterogeneity in the coding across healthcare systems, each EHR source provides partial coverage of the available codes. MIKGI synthesizes the incomplete knowledge graphs derived from these multi-source embeddings by minimizing a spherical loss function that combines the pairwise directional similarities of embeddings computed from all available sources. MIKGI outputs harmonized semantic embedding vectors for all EHR codes, which improves the quality of the embeddings and enables direct assessment of both similarity and relatedness between any pair of codes from multiple healthcare systems.With EHR co-occurrence data from Veteran Affairs (VA) healthcare and Mass General Brigham (MGB), MIKGI algorithm produces high quality embeddings for a variety of downstream tasks including detecting known similar or related entity pairs and mapping VA local codes to the relevant EHR codes used at MGB. Based on the cosine similarity of the MIKGI trained embeddings, the AUC was 0.918 for detecting similar entity pairs and 0.809 for detecting related pairs. For cross-institutional medical code mapping, the top 1 and top 5 accuracy were 91.0% and 97.5% when mapping medication codes at VA to RxNorm medication codes at MGB; 59.1% and 75.8% when mapping VA local laboratory codes to LOINC hierarchy. When trained with 500 labels, the lab code mapping attained top 1 and 5 accuracy at 77.7% and 87.9%. MIKGI also attained best performance in selecting VA local lab codes for desired laboratory tests and COVID-19 related features for COVID EHR studies. Compared to existing methods, MIKGI attained the most robust performance with accuracy the highest or near the highest across all tasks.The proposed MIKGI algorithm can effectively integrate incomplete summary data from biomedical text and EHR data to generate harmonized embeddings for EHR codes for knowledge graph modeling and cross-institutional translation of EHR codes.

The increasing availability of electronic health record (EHR) systems has created enormous potential for translational research. However, it is difficult to know all the relevant codes related to a phenotype due to the large number of codes available. Traditional data mining approaches often require the use of patient-level data, which hinders the ability to share data across institutions. In this project, we demonstrate that multi-center large-scale code embeddings can be used to efficiently identify relevant features related to a disease of interest. We constructed large-scale code embeddings for a wide range of codified concepts from EHRs from two large medical centers. We developed knowledge extraction via sparse embedding regression (KESER) for feature selection and integrative network analysis. We evaluated the quality of the code embeddings and assessed the performance of KESER in feature selection for eight diseases. Besides, we developed an integrated clinical knowledge map combining embedding data from both institutions. The features selected by KESER were comprehensive compared to lists of codified data generated by domain experts. Features identified via KESER resulted in comparable performance to those built upon features selected manually or with patient-level data. The knowledge map created using an integrative analysis identified disease-disease and disease-drug pairs more accurately compared to those identified using single institution data. Analysis of code embeddings via KESER can effectively reveal clinical knowledge and infer relatedness among codified concepts. KESER bypasses the need for patient-level data in individual analyses providing a significant advance in enabling multi-center studies using EHR data.

Disk I/O is a major bottleneck limiting the performance and scalability of data intensive applications. A common way to address disk I/O bottlenecks is using parallel storage systems and utilizing concurrent operation of independent storage components; however, achieving a consistently high parallel I/O performance is challenging due to static configurations. Modern parallel storage systems, especially in the cloud, enterprise data centers, and scientific clusters are commonly shared by various applications generating dynamic and coexisting data access patterns. Nonetheless, these systems generally utilize one-layout-fits-all data placement strategy frequently resulting in suboptimal I/O parallelism. Guided by association rule mining, graph coloring, bin packing, and network flow techniques, this paper proposes a general framework for adaptive parallel storage systems, with the goal of continuously providing a high-degree of I/O parallelism. Evaluation results indicate that the proposed framework is highly successful in adjusting to skewed parallel access patterns for both hard disk drive (HDD) based traditional storage arrays and solid-state drive (SSD) based all-flash arrays. In addition to the storage arrays, the proposed framework is sufficiently generic and can be tailored to various other parallel storage scenarios including but not limited to key-value stores, parallel/distributed file systems, and internal parallelism of SSDs.

The process of identifying a cohort of interest is a very challenging task. It requires manually inspecting many patient records of complex structure that might include medical coding errors and missing data. This paper presents a computational pipeline for refining the process of cohort selection based on medical concepts recorded in the electronic health records (EHRs). The pipeline extracts EHR data for a given cohort and normalizes this data using standard vocabularies. Then a stacked denoising autoencoder is used to embed the normalized patient vectors in a low dimensional space, where the patients are subsequently clustered into sub-cohorts. The goal is to represent the cohort in a standard format and abstract variants of sub-populations. As a use-case, we applied the pipeline to 1.8 million Veterans diagnosed with major depressive disorder (MDD), and identified four meaningful sub-cohorts using the features learned by the autoencoder. Then, each sub-cohort was explored using a set of keywords for interpretation.

The Atmospheric Radiation Measurement (ARM) Climate Research Facility (www.arm.gov) provides atmospheric observations from diverse climatic regimes around the world. Currently, ARM archives over 22 million user assessable data files, primarily stored in NetCDF file format, with total data volumes close to one Petabyte. In this paper, we will discuss how ARM is currently storing, distributing, cataloging and visualizing such large volumes of multi-dimensional climate observations and model data and also describe their future plan.

Storage performance bottlenecks are one of the major threats limiting the scalability of I/O intensive applications. Parallel storage systems have the potential to alleviate I/O bottlenecks through concurrent operation of independent storage components if a parallelism-aware data layout can be continuously guaranteed. Existing systems use one-layout-fits-all data placement strategy that frequently results in sub-optimal I/O parallelism. Guided by association rule mining, graph coloring, bin packing, and network flow techniques, this paper proposes a general framework for self-optimizing parallel storage systems, with the goal of continuously providing a high-degree of I/O parallelism that is robust to changes in the parallel access patterns of applications and the coexistence of applications with different parallel access characteristics. Evaluation results indicate that the proposed framework is highly successful in adjusting to skewed parallel access patterns for both traditional hard disk drive (HDD) based storage arrays and solid-state drive (SSD) based all-flash arrays. In addition to the storage arrays, the proposed framework is sufficiently generic to be tailored to various other parallel storage scenarios including but not limited to key-value stores, parallel/distributed file systems, and internal parallelism of SSDs.

As a result of immense growth of digital data in the last decade, energy consumption has become an important issue in data storage systems. In the US alone, data centers were projected to consume $4 billion (40 TWh) yearly electricity in 2005. This cost had reached to $10 billion (100 TWh) in 2011, and expected to be around $20 billion (200 TWh) in 2016 by doubling itself every 5 years. In addition to the economic burden on companies and research institutions, these large scale data storage systems also have a negative impact on the environment. According to the EPA, generating 1 KWh of electricity in the US results in an average of 1.55 pounds of carbon dioxide emissions. Considering a projected 200 TWh energy requirement for 2016, energy-efficient data storage systems can have a huge economic and environmental impacts on society. This project exploits replication and heterogeneity existing in modern multi-disk storage systems and proposes an energy-efficient and performance-aware replica selection technique to reduce the energy consumption of data storage systems without negatively affecting their performance. Our proposed technique exploits the difference between active and idle energy consumption in heterogeneous disks holding the same replica and selects replicas by balancing energy and performance.