Leveraging Open Electronic Health Record Data and Environmental Exposures Data to Derive Insights Into Rare Pulmonary Disease.

Research on rare diseases has received increasing attention, in part due to the realized profitability of orphan drugs. Biomedical informatics holds promise in accelerating translational research on rare disease, yet challenges remain, including the lack of diagnostic codes for rare diseases and privacy concerns that prevent research access to electronic health records when few patients exist. The Integrated Clinical and Environmental Exposures Service (ICEES) provides regulatory-compliant open access to electronic health record data that have been integrated with environmental exposures data, as well as analytic tools to explore the integrated data. We describe a proof-of-concept application of ICEES to examine demographics, clinical characteristics, environmental exposures, and health outcomes among a cohort of patients enriched for phenotypes associated with cystic fibrosis (CF), idiopathic bronchiectasis (IB), and primary ciliary dyskinesia (PCD). We then focus on a subset of patients with CF, leveraging the availability of a diagnostic code for CF and serving as a benchmark for our development work. We use ICEES to examine select demographics, co-diagnoses, and environmental exposures that may contribute to poor health outcomes among patients with CF, defined as emergency department or inpatient visits for respiratory issues. We replicate current understanding of the pathogenesis and clinical manifestations of CF by identifying co-diagnoses of asthma, chronic nasal congestion, cough, middle ear disease, and pneumonia as factors that differentiate patients with poor health outcomes from those with better health outcomes. We conclude by discussing our preliminary findings in relation to other published work, the strengths and limitations of our approach, and our future directions.

A Biomedical Knowledge Graph System to Propose Mechanistic Hypotheses for Real-World Environmental Health Observations: Cohort Study and Informatics Application.

BACKGROUND: Knowledge graphs are a common form of knowledge representation in biomedicine and many other fields. We developed an open biomedical knowledge graph-based system termed Reasoning Over Biomedical Objects linked in Knowledge Oriented Pathways (ROBOKOP). ROBOKOP consists of both a front-end user interface and a back-end knowledge graph. The ROBOKOP user interface allows users to posit questions and explore answer subgraphs. Users can also posit questions through direct Cypher query of the underlying knowledge graph, which currently contains roughly 6 million nodes or biomedical entities and 140 million edges or predicates describing the relationship between nodes, drawn from over 30 curated data sources. OBJECTIVE: We aimed to apply ROBOKOP to survey data on workplace exposures and immune-mediated diseases from the Environmental Polymorphisms Registry (EPR) within the National Institute of Environmental Health Sciences. METHODS: We analyzed EPR survey data and identified 45 associations between workplace chemical exposures and immune-mediated diseases, as self-reported by study participants (n= 4574), with 20 associations significant at P<.05 after false discovery rate correction. We then used ROBOKOP to (1) validate the associations by determining whether plausible connections exist within the ROBOKOP knowledge graph and (2) propose biological mechanisms that might explain them and serve as hypotheses for subsequent testing. We highlight the following three exemplar associations: carbon monoxide-multiple sclerosis, ammonia-asthma, and isopropanol-allergic disease. RESULTS: ROBOKOP successfully returned answer sets for three queries that were posed in the context of the driving examples. The answer sets included potential intermediary genes, as well as supporting evidence that might explain the observed associations. CONCLUSIONS: We demonstrate real-world application of ROBOKOP to generate mechanistic hypotheses for associations between workplace chemical exposures and immune-mediated diseases. We expect that ROBOKOP will find broad application across many biomedical fields and other scientific disciplines due to its generalizability, speed to discovery and generation of mechanistic hypotheses, and open nature.

Visualization Environment for Federated Knowledge Graphs: Development of an Interactive Biomedical Query Language and Web Application Interface.

BACKGROUND: Efforts are underway to semantically integrate large biomedical knowledge graphs using common upper-level ontologies to federate graph-oriented application programming interfaces (APIs) to the data. However, federation poses several challenges, including query routing to appropriate knowledge sources, generation and evaluation of answer subsets, semantic merger of those answer subsets, and visualization and exploration of results. OBJECTIVE: We aimed to develop an interactive environment for query, visualization, and deep exploration of federated knowledge graphs. METHODS: We developed a biomedical query language and web application interphase-termed as Translator Query Language (TranQL)-to query semantically federated knowledge graphs and explore query results. TranQL uses the Biolink data model as an upper-level biomedical ontology and an API standard that has been adopted by the Biomedical Data Translator Consortium to specify a protocol for expressing a query as a graph of Biolink data elements compiled from statements in the TranQL query language. Queries are mapped to federated knowledge sources, and answers are merged into a knowledge graph, with mappings between the knowledge graph and specific elements of the query. The TranQL interactive web application includes a user interface to support user exploration of the federated knowledge graph. RESULTS: We developed 2 real-world use cases to validate TranQL and address biomedical questions of relevance to translational science. The use cases posed questions that traversed 2 federated Translator API endpoints: Integrated Clinical and Environmental Exposures Service (ICEES) and Reasoning Over Biomedical Objects linked in Knowledge Oriented Pathways (ROBOKOP). ICEES provides open access to observational clinical and environmental data, and ROBOKOP provides access to linked biomedical entities, such as "gene," "chemical substance," and "disease," that are derived largely from curated public data sources. We successfully posed queries to TranQL that traversed these endpoints and retrieved answers that we visualized and evaluated. CONCLUSIONS: TranQL can be used to ask questions of relevance to translational science, rapidly obtain answers that require assertions from a federation of knowledge sources, and provide valuable insights for translational research and clinical practice.

Clinical Annotation Research Kit (CLARK): Computable Phenotyping Using Machine Learning.

Computable phenotypes are algorithms that translate clinical features into code that can be run against electronic health record (EHR) data to define patient cohorts. However, computable phenotypes that only make use of structured EHR data do not capture the full richness of a patient's medical record. While natural language processing (NLP) methods have shown success in extracting clinical features from text, the use of such tools has generally been limited to research groups with substantial NLP expertise. Our goal was to develop an open-source phenotyping software, Clinical Annotation Research Kit (CLARK), that would enable clinical and translational researchers to use machine learning-based NLP for computable phenotyping without requiring deep informatics expertise. CLARK enables nonexpert users to mine text using machine learning classifiers by specifying features for the software to match in clinical notes. Once the features are defined, the user-friendly CLARK interface allows the user to choose from a variety of standard machine learning algorithms (linear support vector machine, Gaussian Naïve Bayes, decision tree, and random forest), cross-validation methods, and the number of folds (cross-validation splits) to be used in evaluation of the classifier. Example phenotypes where CLARK has been applied include pediatric diabetes (sensitivity=0.91; specificity=0.98), symptomatic uterine fibroids (positive predictive value=0.81; negative predictive value=0.54), nonalcoholic fatty liver disease (sensitivity=0.90; specificity=0.94), and primary ciliary dyskinesia (sensitivity=0.88; specificity=1.0). In each of these use cases, CLARK allowed investigators to incorporate variables into their phenotype algorithm that would not be available as structured data. Moreover, the fact that nonexpert users can get started with machine learning-based NLP with limited informatics involvement is a significant improvement over the status quo. We hope to disseminate CLARK to other organizations that may not have NLP or machine learning specialists available, enabling wider use of these methods.

ROBOKOP KG and KGB: Integrated Knowledge Graphs from Federated Sources.

A proliferation of data sources has led to the notional existence of an implicit Knowledge Graph (KG) that contains vast amounts of biological knowledge contributed by distributed Application Programming Interfaces (APIs). However, challenges arise when integrating data across multiple APIs due to incompatible semantic types, identifier schemes, and data formats. We present ROBOKOP KG ( http://robokopkg.renci.org ), which is a KG that was initially built to support the open biomedical question-answering application, ROBOKOP (Reasoning Over Biomedical Objects linked in Knowledge-Oriented Pathways) ( http://robokop.renci.org ). Additionally, we present the ROBOKOP Knowledge Graph Builder (KGB), which constructs the KG and provides an extensible framework to handle graph query over and integration of federated data sources.

ROBOKOP: an abstraction layer and user interface for knowledge graphs to support question answering.

SUMMARY: Knowledge graphs (KGs) are quickly becoming a common-place tool for storing relationships between entities from which higher-level reasoning can be conducted. KGs are typically stored in a graph-database format, and graph-database queries can be used to answer questions of interest that have been posed by users such as biomedical researchers. For simple queries, the inclusion of direct connections in the KG and the storage and analysis of query results are straightforward; however, for complex queries, these capabilities become exponentially more challenging with each increase in complexity of the query. For instance, one relatively complex query can yield a KG with hundreds of thousands of query results. Thus, the ability to efficiently query, store, rank and explore sub-graphs of a complex KG represents a major challenge to any effort designed to exploit the use of KGs for applications in biomedical research and other domains. We present Reasoning Over Biomedical Objects linked in Knowledge Oriented Pathways as an abstraction layer and user interface to more easily query KGs and store, rank and explore query results. AVAILABILITY AND IMPLEMENTATION: An instance of the ROBOKOP UI for exploration of the ROBOKOP Knowledge Graph can be found at http://robokop.renci.org. The ROBOKOP Knowledge Graph can be accessed at http://robokopkg.renci.org. Code and instructions for building and deploying ROBOKOP are available under the MIT open software license from https://github.com/NCATS-Gamma/robokop. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The impact of time on seizure prediction performance in the FSPEEG database.

We demonstrate evidence that high discriminability between preictal and interictal intracranial electroencephalogram (iEEG) recordings [1,2] of the Freiburg database (FSPEEG) may be due to the amount of time that occurred between recordings, as opposed to the underlying seizure state, i.e., preictal or interictal. After replicating published classification methods and results, we performed two experiments. In the first experiment, almost perfect discriminability between discontinuous interictal recordings and almost perfect discriminability between discontinuous preictal recordings were observed as the amount of time between recordings increased. Further, a second experiment demonstrated that the classification performance for patients with large time gaps between preictal and interictal recordings was noticeably higher than the classification performance for patients with contiguous preictal and interictal files. These results provide evidence that time likely plays a major role in the discriminability of the iEEG features considered in this study, regardless of the underlying seizure state. Feature nonstationarity is present and may, under certain conditions, lead to overestimation or underestimation of the probability of seizure occurrence.

Moving Away From Error-Related Potentials to Achieve Spelling Correction in P300 Spellers.

P300 spellers can provide a means of communication for individuals with severe neuromuscular limitations. However, its use as an effective communication tool is reliant on high P300 classification accuracies ( > 70%) to account for error revisions. Error-related potentials (ErrP), which are changes in EEG potentials when a person is aware of or perceives erroneous behavior or feedback, have been proposed as inputs to drive corrective mechanisms that veto erroneous actions by BCI systems. The goal of this study is to demonstrate that training an additional ErrP classifier for a P300 speller is not necessary, as we hypothesize that error information is encoded in the P300 classifier responses used for character selection. We perform offline simulations of P300 spelling to compare ErrP and non-ErrP based corrective algorithms. A simple dictionary correction based on string matching and word frequency significantly improved accuracy (35-185%), in contrast to an ErrP-based method that flagged, deleted and replaced erroneous characters (-47-0%) . Providing additional information about the likelihood of characters to a dictionary-based correction further improves accuracy. Our Bayesian dictionary-based correction algorithm that utilizes P300 classifier confidences performed comparably (44-416%) to an oracle ErrP dictionary-based method that assumed perfect ErrP classification (43-433%).

Projected accuracy metric for the P300 Speller.

The P300 Speller brain-computer interface (BCI) is a virtual keyboard that allows users to type without requiring neuromuscular control. P300 Speller research commonly aims to improve the system accuracy, which is typically estimated by spelling a small number of characters and calculating the percent spelled correctly. In this paper we introduce a new method for estimating the long-term ("projected") accuracy, which utilizes all available flash data and a probabilistic model of the Speller system to produce an estimate with lower variance and lower granularity than the standard measure. We apply the new method to 110 previously-collected P300 Speller runs to confirm its consistency, and simulate spelling runs from real subject data to demonstrate lower variance on the accuracy estimate for any given amount of data.

Mandarin Chinese tone identification in cochlear implants: predictions from acoustic models.

It has been established that current cochlear implants do not supply adequate spectral information for perception of tonal languages. Comprehension of a tonal language, such as Mandarin Chinese, requires recognition of lexical tones. New strategies of cochlear stimulation such as variable stimulation rate and current steering may provide the means of delivering more spectral information and thus may provide the auditory fine-structure required for tone recognition. Several cochlear implant signal processing strategies are examined in this study, the continuous interleaved sampling (CIS) algorithm, the frequency amplitude modulation encoding (FAME) algorithm, and the multiple carrier frequency algorithm (MCFA). These strategies provide different types and amounts of spectral information. Pattern recognition techniques can be applied to data from Mandarin Chinese tone recognition tasks using acoustic models as a means of testing the abilities of these algorithms to transmit the changes in fundamental frequency indicative of the four lexical tones. The ability of processed Mandarin Chinese tones to be correctly classified may predict trends in the effectiveness of different signal processing algorithms in cochlear implants. The proposed techniques can predict trends in performance of the signal processing techniques in quiet conditions but fail to do so in noise.