- Open Access
SOCR data dashboard: an integrated big data archive mashing medicare, labor, census and econometric information
© Husain et al. 2015
- Received: 12 April 2015
- Accepted: 1 June 2015
- Published: 17 July 2015
Intuitive formulation of informative and computationally-efficient queries on big and complex datasets present a number of challenges. As data collection is increasingly streamlined and ubiquitous, data exploration, discovery and analytics get considerably harder. Exploratory querying of heterogeneous and multi-source information is both difficult and necessary to advance our knowledge about the world around us.
We developed a mechanism to integrate dispersed multi-source data and service the mashed information via human and machine interfaces in a secure, scalable manner. This process facilitates the exploration of subtle associations between variables, population strata, or clusters of data elements, which may be opaque to standard independent inspection of the individual sources. This a new platform includes a device agnostic tool (Dashboard webapp, http://socr.umich.edu/HTML5/Dashboard/) for graphical querying, navigating and exploring the multivariate associations in complex heterogeneous datasets.
The paper illustrates this core functionality and serviceoriented infrastructure using healthcare data (e.g., US data from the 2010 Census, Demographic and Economic surveys, Bureau of Labor Statistics, and Center for Medicare Services) as well as Parkinson’s Disease neuroimaging data. Both the back-end data archive and the front-end dashboard interfaces are continuously expanded to include additional data elements and new ways to customize the human and machine interactions.
A client-side data import utility allows for easy and intuitive integration of user-supplied datasets. This completely open-science framework may be used for exploratory analytics, confirmatory analyses, meta-analyses, and education and training purposes in a wide variety of fields.
- Graphical Query
- Industry Type
- North American Industry Classification System
- Chart Type
- Medicare Service
State of open-science
Open-science refers to a new paradigm liberalizing access, advancement, control, accreditation and ownership of scientific knowledge, information resources (e.g., data), and decision-making instruments (e.g., software tools). In open-science settings, the entire community has open unrestricted access to resources incentivizing user participation at different levels (e.g., consumption, mashing, development), enabling social interactions where the collective outcome is more than the sum of its individual parts, novice learners and experts can choice, contribute to and debate concepts, algorithms, analytics, results and theoretic principles. This paradigm requires a critical mass of participation, commitment for trust, diversity, sharing and cooperation. Outcome products often take unexpected and innovative turns, looking at problems from different angles and employing expertise, methods and services, which initially may appear as not-interoperable. There are many examples of successful open-science initiatives. Two of these are the Polymath project  and the Mozilla Open-Science . Polymath1 problem identified by the Polymath community involved searching for a new combinatorial proof to the Hales–Jewett theorem . The project morphed into multiple independent threads, which led to a solution of the problem within several months, using constructive contributions from dozens of people. An international team of scientists and engineers participated in Mozilla Open-Science 52-h coding marathon to enhance open science lessons, learning materials, teaching tools, and software resources and establish minimal standards for open science education. This virtual activity used open-resources (e.g., GitHub) to establish reproducible research guidelines, develop and enhance teaching materials in a diverse array of scientific disciplines (bioinformatics, medical imaging, oceanography, social science, etc.)
There are now over 1 billion web-sites hosted on millions of Internet servers . This widespread availability of information provides web access to resources that can be efficiently shared with minimal barriers to content and data. Yet, there are significant barriers to effective open-science. Some of the challenges include lack of interoperability or compatibility of resources, licensing restrictions, federal, state and local regulations, mistrust in attribution, ineffective infrastructure, and discipline boundaries.
Challenges in managing, fusing, processing, servicing and understanding heterogeneous data
Big Data is ubiquitous in many, if not all, scientific disciplines, applied studies and research explorations. Its characteristics include heterogeneity, multiple-scales, incongruent space-time sampling, format complexity, privacy restrictions, and multi-source provenance. There are significant, unique and impeding challenges associated with Big Data modelling, handling, analytics, interpretation and progress of extracting information and gaining knowledge from it. Each step in the complete workflow from data acquisition, to data storage, servicing, archival, manipulation and processing present problems that need to be overcome to enable the course of information extraction and decision-making. Lack of standards, unstructured data formats and aggregation of data (elements and cases) inhibit the semantic content processing, search and holistic data understanding. Data volume is not the only bottleneck. In many applications, the data complexity and heterogeneity frequently constrain the use of established methods, techniques, software tools and services. One of the paramount feature of Big Data is its temporal dynamism. The value of models, inference, and findings presented using a stale data repositories rapidly depreciate with time . Just like the formulation of conditional probability and Bayesian inference  advanced our understanding of marginal and joint probabilities , data mashing, the process of fusion of information from disparate resources, has a potential to revolutionize our understanding of natural processes and complex phenomena.
Big data analytic infrastructure
Existing interactive data visualization platforms
The purpose of the Dashboard project was to design a webapp capable of seamlessly merging datasets from a wide variety of sources without the need for complex mathematical and statistical packages such as R and Mathematica. Two versions of the webapp were designed. The first one (Location-Anchored) provides an interface to fusing examples of available datasets (using a common FIPS or ZIP location mapping) to generate a mashed and munged archive accessible via an intuitive graphical user interface, as well as via a machine accessible API. The second version (Free-Anchored) enables users to provide their own data, map and assemble the data according to specific variables of interest (e.g., subject weight). By using data simulation and by using Crossfilter (http://square.github.com/crossfilter) to dynamically track changes to datasets by using indexing and bitfields, computation time is minimized, thereby providing a fast, scalable, and easy-to-use platform for integrating and comparing datasets from unrelated sources. By integrating a simulation algorithm, the Dashboard maintains data privacy without the need for encryption, making the webapp ideal for consumer or patient based datasets. The use of a non-relational database such as MongoDB allows for built in horizontal scalability, making the app just as proficient at manipulating traditional datasets as it is at true “Big Data” datasets. Data simulation and fusion reduces the overall data footprint of the webapp, resulting in an application that is small enough to deploy onto a smartphone or other portable device. Finally, real-time updating and auto integration of multiple datasets allow for quick and seamless operations without the time overhead associated with traditional RDBMS-style approaches . In order to demonstrate the functionality of the application we have chosen healthcare and neuroimaging data demonstrations. However, the Dashboard webapp can be used to mine, analyze and visualize structured data from virtually any research area.
Data sources used and rationale for selection
The data sources used to test the Dashboard webapp were selected from a variety of national government and nonprofit online repositories, including the Census Bureau (www.census.gov), Centers for Medicare and Medicaid services (www.cms.hhs.gov), and the Bureau for Labor Statistics (www.bls.gov). These datasets were selected based on completeness, accuracy, and consistent data formatting. Whilst not large enough to serve as true benchmark tests for a Big Data oriented application such as this, the datasets are diverse enough to demonstrate the webapp’s usefulness in cross-discipline applications. For tests regarding properties of the location referenced version of the webapp, data sets that did not contain data for the entire United States were excluded. In addition, data sets that did not display information on a small enough scale were also excluded. For instance, data that only had information on the state level were considered too sparse and were excluded. Particular attention was given to datasets contain information pertaining to the current state of healthcare and medicine in the United States. Furthermore, increased importance was given to datasets containing commonly used demographic identifiers such as race, gender, age group, income, etc. When multiple datasets were encountered containing data for different time periods, sets containing information close to the year 2010 were given priority, in order to match the most recent census year. In order to test the webapp version enabling location-free data references, datasets were chosen that were representative of those typically generated in biomedical research, in this demonstration neuroimaging data. Datasets from multiple studies were used in order to demonstrate the multi-source integration capabilities of the webapp. Listed below is a summary of each included dataset, as well as URLs to each source and rationale for their selection.
2010 Census (http://quickfacts.census.gov/qfd/download/DataSet.txt): Representing the most recent decade-level US census, this dataset provides many of the most commonly used variables for filtering and analysis including race, gender, age group, population density, and education level. This dataset also provides population levels for each county, which are used for a variety of internal dashboard calibrations and operations. This dataset is unique in that it must be included in order for the dashboard to function, and therefore cannot be excluded from the data pool in a custom build of the webapp.
Bureau of Labor Statistics Labor Force Data by County (2010) (http://www.bls.gov/lau/laucnty10.txt): This dataset contains employment statistics provided by the Bureau of Labor Statistics for the year of 2010. This dataset provides several important variables, including unemployment rate and viable labor force levels, and serves as an important indicator of economic health for a particular region.
CMS Hospital Inpatient Data (https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Inpatient_Data_2011_CSV.zip): Released in June 2014 as part of the Obama administration’s initiative to increase healthcare transparency, this dataset provides various statistics, including average Medicare payment, average withdrawal, and total number of cases. All provided statistics are grouped by inpatient procedure type, as well as by hospital .
CMS Hospital Outpatient Data (https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Outpatient_Data_2011_CSV.zip): Similar to the CMS Hospital Inpatient dataset, this set provides statistics on all outpatient procedures performed by each hospital. Reported statistics are the same as those for the inpatient dataset. This dataset allows for determination of correlations and links between outpatient procedures and socioeconomic factors .
2010 Census Economic Survey (ftp://ftp.census.gov/econ2010/CBP_CSV/cbp10co.zip): Provides various economic statistics per county, as well as per industry type, based on North American Industry Classification System (NAICS) specification. Variables reported include number of establishments per industry type, average payroll per industry type, and number of establishments per industry type per size class.
CMS Physician Data (http://download.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Medicare-Physician-and-Other-Supplier-PUF-CY2012.zip?agree=yes&next=Accept): Released April 2014 as part of the White House’s initiative to improve healthcare transparency, this report gives provides information and statistics on individual practitioners, as well as group practices. Statistics reported include provider credentials, provider gender, specialization, average Medicare withdrawal, and patient count. In addition, specific information is given for each type of condition treated.
Biomedical Data: We used clinical, imaging and demographic data from a Parkinson’s disease study [26, 27], a nutrition and obesity study [28, 29], and an Alzheimer’s study [30, 31]. These datasets were chosen to test and validate the performance of the Dashboard app for several reasons. First, these datasets represent large, heterogeneous, multi-source, incomplete and longitudinal Big Data. Second, the translational research projects managing the data provided powerful driving biomedical challenges of interrogating and mining the complex data to visually identify patterns trends and associations. And third, they did not explicitly include FIPS/ZIP code meta-data to allow seamless integration with the default Dashboard datasets, which provided the opportunity to validate the webapp interface with non-standard data and establish indirect mapping anchoring cases by alternative meta-data (e.g., subject weight).
Data conversion and quality control
Data representation and manipulation was done via simulation of a 1/1,000th population sample. Qualitative properties (Race, Gender, etc.) were assigned via simple random sample, using probability distributions calculated from county-level data variables. Quantitative variables (income, unemployment rate, etc.) were set equal to state or county averages, with no distribution or simulated deviation from the mean. By simulating data instead of performing direct filtering and comparison operations, correlations and statistical operations can be performed without revealing any information about the true raw data points. This allows for a layer of anonymity between data and statistics, thereby allowing for increased privacy in applications involving personal or confidential information. Furthermore, the simulation layer results in a fused dataset much smaller in size than the sum of each of its corresponding raw files, thereby enhancing the portability of the webapp.
Geo Choropleth Chart
In addition, several “widgets”, such as a table viewer, are provided in order to further enhance the effectiveness and usability of the webapp.
Once generated, charts and other widgets can be freely moved around the dashboard in accordance with the end-user’s needs. In addition, selection a certain value or range of values causes the population to be filtered to only display samples with the selected value (s). Since all charts and widgets share the same population pool, any filters or operations applied to a chart are reflected on all other charts. For instance, selecting the male category in a Gender pie chart will cause all other charts to filter data so that only male samples are included. This demonstrates the visual graphic based, rather than SQL based, query of the mashed data archive.
Loading the dashboard webapp in the web-browser instantiates the downloading of the summary file from the MongoDB webserver. This allows the webapp to populate the list of all available variables which can be used by the user to initiate the graphical query process. Upon selection and creation of a chart, the webapp queries and downloads the specific data for the required variable (s), and integrates the data into the simulated population dataset. The simulated population pool consists of approximately 300,000 individual “samples”, with each sample representing 1,000 individuals. In order to integrate a new variable, a probability distribution is generated for each possible value, and each sample is assigned a value using a simple random sample based on the calculated probability distribution (e.g. a county with 3,000 males and 5,000 females would be represented as 8 samples, of which on average 3 will be male, and 5 will be female).
The dashboard webapp also allows for the ability to aggregate multiple variables using simple operations. By selecting the “Customize” tile, users are able to add, multiply, or subtract variables, allowing for creation of more complex combined variables.
The current implementation of the dashboard webapp allows for an integration and manipulation of a maximum of 30 different charts at a time, where each chart can contains between 1 and 6 variables. However, we have observed speed and performance impacts as the number of charts doubles (e.g., from 8 to 16 charts).
In addition, the webapp was designed to be able to accommodate a near-limitless number of datasets and variables without suffering any significant performance issues. Currently, approximately 700 different variables are available, though more are expected to be added in the near future as we expand the integrated archive to include complementary information. Finally, we plan to allow for mashing of user-submitted datasets, in order to further enhance the versatility, usability and customization of the webapp.
Filtered and modified data can be exported from the application in a variety of ways, in order to allow further data analysis via other programs and applications. The full combined data file, as well as all source code for the webapp can be found at the project’s GitHub page at https://github.com/SOCRedu/Dashboard-Lab. Processed datasets can be downloaded in zip format at http://socr.umich.edu/data/SOCR_DataDahboard_Dataset_V1.1.zip.
In order to facilitate machine data retrieval, two data API’s have been provided. A raw data REST API, located at http://socrdev.nursing.umich.edu:8080/users/var/[VARNAME] allows for retrieval of unfiltered quantitative variable data points. Qualitative variable data points can similarly be retrieved from the REST API at http://socr-dev.nursing.umich.edu:8080/users/super/[VARNAME].
A second data API, located at http://socr-dev.nursing.umich.edu:8080/api/request, was implemented in order to allow for customized data filtration procedures. By utilizing a Node based server-style version of CrossFilter , the API allows for data integration and filtration with results identical to the GUI Dashboard webapp. The webapp utilizes a query format for requests, wherein filtered variables are specified by setting the variable to the requested filtered variable, and requested variables are indicated by specifying the value of the reqVar variable. For example, in order to request data points for the Average Medicare Payment for Extracranial Procedures for patients with income greater than 60000 and are either White or Black, the API would be queried as http://socr-dev.nursing.umich.edu:8080/api/request?income_per_capita[min]=60000&race=White&race=Black&reqVar=EXTRACRANIAL_PROCEDURES_W-O_CC-mcc_avg_charge.
Besides the machine accessible REST API’s, data from the dashboard can also be exported using the Dashboard web application itself. At any time during operation of the webapp, the Export tile (button) can be selected in order to download all data to local storage. Only data points that pass all currently applied filters are exported, and only variables which are currently displayed on the Dashboard are represented in the exported data file. For example, clicking the Export tile after selecting ‘Male’ on a Gender pie chart and selecting income ranges from $40,000 to $60,000 on an income histogram will generate a CSV data file containing income levels and gender for all Males with income between $40,000 and $60,000.
Data interrogation examples
Below we include three example use-cases for the dashboard webapp, which highlight the major features of the application, as well as its practicality in performing cross-dataset comparisons. The webapp can be accessed online at http://socr.umich.edu/HTML5/Dashboard/.
Upon startup of the webapp, select the ‘Add’ tile.
In the ‘Data Source’ dropdown, select 2010 Census
In the Data Variable dropdown, select Highest Level of Education
Press the Continue Button
In the chart selection window, select the bar chart.
Press the Continue button
Once again, select the ‘Add’ tile
In the ‘Data Source’ dropdown, select 2010 Census
In the Data Variable dropdown, select Income per Capita
Press the Continue Button
In the chart selection window, select the histogram
Press the Continue button
It is now possible to observe relations between the two datasets by applying filters on one or both of the charts. For example, select the ‘Dropout’ bar on the Highest Level of Education bar chart. There should be an observable decrease in the Income per Capita histogram, indicating a correlation between level of education and income. Pressing the export button will download the resulting dataset as a csv file. The resulting file is the CSV equivalent of the result from an API call to http://socrdev.nursing.umich.edu:8080/api/request?reqVar=income_per_capita&Highest_Level_of_education=Dropout, Fig. 5.
Select the Add tile
In the ‘Data Source’ dropdown, select Inpatient Charge Data (Total discharges)
In the Data Variable dropdown, select Major Cardiovascular Procedures (Total Discharges)
Press the Continue Button
In the chart selection window, select the scatterplot
Press the Continue button
In the ‘Data Source’ dropdown for the Y axis, select Inpatient Charge Data (Avg medicare pmnt)
In the Data Variable dropdown for the Y axis, select Major Cardiovascular Procedures (Avg medicare pmnt)
In the ‘Data Source’ dropdown for Color, select Inpatient Charge Data (Avg covered charge)
In the Data Variable dropdown for Color, select Major Cardiovascular Procedures (Avg covered charge)
In the resulting scatterplot, there is a fairly strong observable negative correlation between number of discharges and per-patient charge for hospital treatments of Major Cardiovascular Procedures. However, if the above steps are repeated for Disequilibrium, there is instead a positive correlation.
Select Free-Anchored Version
For the Cohort Name, enter select cerebellum_Volume
Ensure that both datasets contain a column with a header named cerebellum_Volume
Select the cohort range to be from 0 to 300000
Select number of cohorts to be 300
Import both datasets by selecting the ‘Import’ tile and navigating to the dataset .csv files
The data upload and conversion feature of the dashboard was analyzed to gauge dashboard performance for various datasets. Testing was done Using Chrome v.42.0.2311.152 m browser running on a Windows 8.1 Operating System, Intel Core i5 processor, 8 GB of RAM, 320 GB SATA hard drive, and 100 MB network connection. Randomly generated datasets were created using a predetermined number of variables and entries. Dataset upload and graph generation times were obtained using the Google Analytics User Timing library (analytics.js).
Data upload and graph generation times based on number of entries per variable
Number of entries
Mean upload time (ms) (n = 10)
Std deviation (ms)
Mean graph generation time (ms) (n = 10)
Std deviation (ms)
Data upload and graph generation times based on number of variables
Number of variables
Mean upload time (ms) (n = 10)
Std deviation (ms)
Mean graph generation time (ms) (n = 10)
Std deviation (ms)
The era of Big Data deluge presents unique challenges and opportunities for innovation to enhance, expedite and simplify the processes of data exploration, query and discovery. Interactive and device-agnostic tools and services facilitating such data munging and interrogation are necessary for the next generation of powerful analytics-driven modeling, prediction and decision-making. In this manuscript we report on the construction of an interesting mash of multi-source data that could be used to explore intricate associations between variables, population strata, or clusters of data elements, which may not be easy to untangle by independent inspection of the individual data archives. In addition, we present a new platform and device agnostic tool (Dashboard webapp) for querying, navigating and exploring the multivariate associations in complex heterogeneous datasets.
We chose to illustrate the core functionality and service-oriented infrastructure supporting this application using healthcare data. Specifically, the current version of the webapp used US data from the 2010 Census (Demographic and Economic surveys), Bureau of Labor Statistics, and Center for Medicare Services (Hospital Inpatient and Outpatient, Physician Data). This dashboard platform is continuously expanded to include additional data elements and can be customized to manage diverse types of datasets with varying applications. This framework could be used for exploratory analytics, confirmatory analyses, meta-analyses, as well as for education and training purposes.
This entire framework is developed under open-science principles and facilitates the collaboration, refactoring, improvement and sustainability of the data, software tools, web-services and computational infrastructure developed to harvest, pre-process, munge, fuse, query, analyze and interrogate the integrated archive. The complete dataset is available online (http://socr.umich.edu/data/SOCR_DataDahboard_Dataset_V1.1.zip), the webapp can be openly accessed on a public server (http://socr.umich.edu/HTML5/Dashboard/) and the complete software infrastructure is on GitHub (https://github.com/SOCRedu/Dashboard-Lab and https://github.com/SOCR).
In this manuscript we report on the design, implementation and testing of a new platform, SOCR Data Dashboard, for exploratory querying of heterogeneous and multi-source datasets. The Dashboard architecture enabled graphical navigation and discovery of subtle associations between data elements, sub-population strata, or clusters that may be obfuscated during traditional protocols for data inspection. The platform is open-source and openly disseminated as source-code and as service-oriented infrastructure. We tested the Dashboard using complex data from the 2010 US Census, Bureau of Labor Statistics, Center for Medicare Services, and various neuroimaging studies of neurodegeneration. We use continuous-development and extreme-programming practices to rapidly design, implement, test, update, and distribute the data archive and the dashboard human and machine interfaces. The entire computational and data science community is encouraged to employ, extend and support the Dashboard platform for research and training in exploratory analytics, confirmatory analyses and meta-analyses.
The SOCR data dashboard infrastructure was developed with partially supported from NSF grants, 1023115, 1022560, 1022636, 0089377, 9652870, 0442992, 0442630, 0333672, 0716055, and by the NIH Grants P20 NR015331, U54 EB020406, P50 NS091856, R34MH100494, and P30 DK089503. This work was supported in part by the HC Prechter Research Fund at the University of Michigan. Many colleagues have provided contributions including ideas, pilot testing, improvement suggestions and other help aiding the development and validation of these resources. Constructive editorial critiques and reviewer recommendations identified gaps and suggested improvements to the Dashboard webapp and the manuscript.
- Cranshaw J, Kittur A. The polymath project: lessons from a successful online collaboration in mathematics; 2011. ACM. pp. 1865-1874. http://dl.acm.org/citation.cfm?id=1979213
- Hayden EC (2013) Mozilla plan seeks to debug scientific code. Nature 501:472–472View ArticleGoogle Scholar
- Polymath D (2009) A new proof of the density Hales-Jewett theorem. arXiv preprint arXiv:09103926. http://arxiv.org/abs/0910.3926
- WashingtonPost (2014) http://www.washingtonpost.com/news/theintersect/wp/2014/09/22/there-are-now-officially-a-billionweb-sites-on-the-internet-we-think/
- Dinov I (2014) www.aaas.org/news/big-data-blog-part-v-interview-dr-ivo-dinov. AAAS.
- Box GE, Tiao GC (2011) Bayesian inference in statistical analysis: John Wiley & Sons. https://books.google.com/books?id=T8Askeyk1k4C
- Dinov I, Kamino S, Bhakhrani B, Christou N (2013) Technology-enhanced Interactive Teaching of Marginal, Joint and Conditional Probabilities: The Special Case of Bivariate Normal Distribution. Teaching Statistics 35(3):131–139Google Scholar
- White T (2009) Hadoop: the definitive guide: the definitive guide: “ O’Reilly Media, Inc.”. https://books.google.com/books?id=drbI_aro20oC
- Chodorow K (2013) MongoDB: the definitive guide: O’Reilly. https://books.google.com/books?id=uGUKiNkKRJ0C
- Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P et al (2009) The WEKA data mining software: an update. ACM SIGKDD Explorations Newsl 11:10–18View ArticleGoogle Scholar
- Athey BD, Braxenthaler M, Haas M, Guo Y (2013) tranSMART: An Open Source and Community-Driven Informatics and Data Sharing Platform for Clinical and Translational Research. AMIA Summits on Translational Science Proceedings 2013:6–8Google Scholar
- Berthold MR, Cebron N, Dill F, Gabriel TR, Kötter T et al (2008) KNIME: The Konstanz Information Miner. In: Preisach C, Burkhardt H, Schmidt-Thieme L, Decker R (eds) Data Analysis, Machine Learning and Applications. Springer, Berlin Heidelberg, pp 319–326View ArticleGoogle Scholar
- Dinov ID, Petrosyan P, Liu Z, Eggert P, Zamanyan A et al (2013) The perfect neuroimaging-genetics-computation storm: collision of petabytes of data, millions of hardware devices and thousands of software tools. Brain Imaging Behav 8:311–322Google Scholar
- Goecks J, Nekrutenko A, Taylor J, Team TG (2010) Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol 11:R86View ArticleGoogle Scholar
- Johnson GT, Hertig S (2014) A guide to the visual analysis and communication of biomolecular structural data. Nat Rev Mol Cell Biol 15:690–698View ArticleGoogle Scholar
- Viegas FB, Wattenberg M, Van Ham F, Kriss J, McKeon M (2007) Manyeyes: a site for visualization at internet scale. Vis Computer Graph, IEEE Trans 13:1121–1128View ArticleGoogle Scholar
- Bostock M, Ogievetsky V, Heer J (2011). D3 data-driven documents. Visualization and Computer Graphics, IEEE Transactions on 17(12): 2301-2309.Google Scholar
- Ono K, Demchak B, Ideker T (2014) Cytoscape tools for the web age: D3. js and Cytoscape. js exporters. F1000Research 3:143–145Google Scholar
- Smoot ME, Ono K, Ruscheinski J, Wang P-L, Ideker T (2011) Cytoscape 2.8: new features for data integration and network visualization. Bioinformatics 27:431–432View ArticleGoogle Scholar
- Nandeshwar A (2013) Tableau data visualization cookbook: Packt Publishing Ltd. https://books.google.com/books?id=IKu_oD_fBiIC
- Kandel S, Paepcke A, Hellerstein J, Heer J. Wrangler: Interactive visual specification of data transformation scripts; 2011. ACM. pp. 3363-3372. http://dl.acm.org/citation.cfm?id=1979444
- Gudivada VN, Rao D, Raghavan VV. NoSQL Systems for Big Data Management; 2014. IEEE. pp. 190-197. http://ieeexplore.ieee.org/xpls/icp.jsp?arnumber=6903264
- Medicare CF, Medicaid Services H (2013) Medicare program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and Fiscal Year 2014 rates; quality reporting requirements for specific providers; hospital conditions of participation; payment policies related to patient status. Final Rules Federal Register 78:50495Google Scholar
- Gerhardt G, Yemane A, Apostle K, Oelschlaeger A, Rollins E, et al. (2014) Evaluating Whether Changes in Utilization of Hospital Outpatient Services Contributed to Lower Medicare Readmission Rate. Medicare & Medicaid research review 4(1):1–7Google Scholar
- Dinov ID, Petrosyan P, Liu Z, Eggert P, Hobel S, et al. (2014) High-throughput neuroimaging-genetics computational infrastructure. Front. Neuroinform 8:41. doi: 10.3389/fninf.2014.00041
- Lebedev A, Westman E, Simmons A, Lebedeva A, Siepel FJ, et al. (2014) Large-scale resting state network correlates of cognitive impairment in Parkinson’s disease and related dopaminergic deficits. Frontiers in Systems Neuroscience 8(45):23–28Google Scholar
- IglayReger HB, Peterson MD, Liu D, Parker CA, Woolford SJ et al (2014) Sleep duration predicts cardiometabolic risk in obese adolescents. J Pediatr 164:1085–1090, e1081View ArticleGoogle Scholar
- Rothberg AE, McEwen LN, Fraser T, Burant CF, Herman WH (2013) The impact of a managed care obesity intervention on clinical outcomes and costs: A prospective observational study. Obesity 21:2157–2162View ArticleGoogle Scholar
- Apostolova LG, Akopyan GG, Partiali N, Steiner CA, Dutton RA, et al. (2007) Structural correlates of apathy in Alzheimer’s disease. Dementia and Geriatric Cognitive DisordersGoogle Scholar
- Apostolova LG, Dinov ID, Dutton RA, Hayashi KM, Toga AW et al (2006) 3D comparison of hippocampal atrophy in amnestic mild cognitive impairment and Alzheimer’s disease. Brain 129:2867–2873View ArticleGoogle Scholar
- Wei-ping Z, Ming-Xin L, Huan C. Using MongoDB to implement textbook management system instead of MySQL; 2011. IEEE. pp. 303-305.Google Scholar
- Al-Aziz J, Christou N, Dinov I (2010) SOCR Motion Charts: An Efficient, Open-Source, Interactive and Dynamic Applet for Visualizing Longitudinal Multivariate Data. JSE 18:1–29Google Scholar
- Sarkar D (2008) Lattice: multivariate data visualization with R: Springer Science & Business Media. https://books.google.com/books?id=gXxKFWkE9h0C
- Hand D, Mannila H, Smyth P (2001) Principles of data mining: MIT press. https://books.google.com/books?id=SdZ-bhVhZGYC
- Phaltane S, Nimbalkar O, Sonavle P, Vij SR (2013) Apache Web Server Monitoring. International Journal of Scientific & Engineering Research 4(7):2195-2199.Google Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.