Column-oriented NoSQL databases

NoSQL (Not only Structured Query Language) databases are databases that are used to store data in non-relational databases i.e. graphical, document store, column-oriented, key-value, and object-oriented databases (Sadalage & Fowler, 2012; Services, 2015). Column-oriented databases are perfect for sparse datasets, ones with many null values and when columns do have data the related columns are grouped together (Services, 2015).  Grouping demographic data like age, income, gender, marital status, sexual orientation, etc. are a great example for using this NoSQL database. Cassandra, which is a column-oriented NoSQL database focuses on availability and partition tolerance, this means that as an AP system it can achieve consistency if data can be replicated and verified (Hurst, 2010).

Cassandra has been assessed for performance evaluation against other NoSQL databases like MongoDB and Raik for health care data analytics (Weider, Kollipara, Penmetsa, & Elliadka, 2013).  In this study, NoSQL database demands for health care data were two-fold:

  • Read/write efficiency of medical test results for a patient X (Availability)
  • All medical professionals should see the same information on patient X (Consistency)

A NoSQL graph database did not have the fit to use for the above demands, thus wasn’t part of this study.

The architecture of this project: nine partition nodes, where three by three nodes were used to mimic three data centers that would be used by 100 global health facilities, where data is generated at a rate of 1TB per month and must be kept for 99 years.

The dataset used in this project: a synthetic dataset that has 1M patients with 10M lab reports, averaging at seven lab reports per person, but randomly distributed of from 0-20 lab reports per person.

In meeting both of these two demands, Cassandra had a significantly higher throughput value than the other two NoSQL databases. Cassandra’s EACH_QUORUM write and LOCAL_QUORUM read options are part of their datacenter aware system, providing the great throughputs results, using the three synthetic datacenters. Testing consistency, by using Cassandra’s ONE for its write and read options at an eventual rate (slower consistency) or strong rate (faster consistency), shows that throughput increases with the eventual system. The choice to use either rate rests with the healthcare stakeholders.

The authors concluded that for their system and their requirements Cassandra had the highest throughput regardless of the level of consistency rates (Weider et al., 2013).  They also suggested that each of these tests should be adjusted based on the requirements from key stakeholders in the healthcare profession and that a small variation in the data model could change the results seen here.

In conclusion of this post, NoSQL databases provide huge advantages to data analytics over traditional relational database management systems. But, NoSQL databases must fit the needs of the stakeholders, and quantitative tests must be thoroughly designed to assess which NoSQL database will meet those needs.

References

  • Hurst, N. (2010). Visual guide to NoSQL systems. Retrieved from http://blog.nahurst.com/visual-guide-to-nosql-systems
  • Sadalage, P. J., Fowler, M. (2012). NoSQL Distilled: A Brief Guide to the Emerging World of Polyglot Persistence, 1st Edition. [Bookshelf Online].
  • Services, E. E. (2015). Data Science and Big Data Analytics: Discovering, Analyzing, Visualizing and Presenting Data, 1st Edition. [Bookshelf Online].
  • Weider, D. Y., Kollipara, M., Penmetsa, R., & Elliadka, S. (2013, October). A distributed storage solution for cloud based e-Healthcare Information System. In e-Health Networking, Applications & Services (Healthcom), 2013 IEEE 15th International Conference on (pp. 476-480). IEEE.
Advertisements

Graphical NoSQL Databases

There is a lot of complicated connections between patients, their provider, their diagnoses, etc., and graphically representing this relationship data is one of the main highlights of using a NoSQL graph database (Park, Shankar, Park, & Ghosh, 2014). NoSQL (Not only Structured Query Language) databases are databases that are used to store data in non-relational databases i.e. graphical, document store, column-oriented, key-value, and object-oriented databases (Sadalage & Fowler, 2012; Services, 2015). Graph NoSQL databases are used drawing networks by showing the relationship between items in a graphical format that has been optimized for easy searching and editing (Services, 2015). Each item is considered a node and adding more nodes or relationships while traversing through them is made simpler through a graph database rather than a traditional database (Sadalage & Fowler, 2012). Some sample graph databases consist of Neo4j Pregel, etc. (Park et al., 2014).

Case Study: Graph Databases for large-scale healthcare systems: A framework for efficient data management and data services (Park et al., 2014)

Driver for data analytics needs: Finding areas for cost savings through anomaly detection algorithms, because currently there are a bunch of individual tables and non-normalized data that are replicated multiple times which is causing bottlenecks.

Problem: Understanding and establishing relationships between self-referrals and shared providers, which allows for the use of a collaborative filter.

System Needs: Data management needs error-tolerant and non-redundant database system, while data services need data retrieval, analytics queries, statistical data extraction and mining algorithms.

NoSQL Database used: Neo4J graph NoSQL Database using Cypher query to keep the data normalized and reduce the number of individual tables of data due to the advanced yet simple query capabilities

Methodology: Using the 3EG: 3NF Equivalent Graph Transformation algorithm to convert traditional relational database data into graph database data on realistic synthetic healthcare data.  The synthetic healthcare data consists of zip-codes, diagnosis of disease, available procedures, beneficiary, claim, and providers. The data when flattened can showcase 1 M beneficiaries to 100 K providers, but in a graphical format, that same data will have 51 M nodes and 257 M relationships.

Queries Ran on the NoSQL Database:

  • Shared providers between two beneficiaries
  • Shared providers between two beneficiaries through either actual visits or by referrals
  • List of shared diseases between two beneficiaries through their claim records
  • Any link between two beneficiaries à helps to direct further investigations/queries
  • Shared beneficiaries between two providers
  • Self-referred beneficiaries for a given provider
  • Similar claims based on diagnoses codes
  • Patient wants to switch to a new provider based on a referral by another provider

Using 50 random queries for each of the 8 cases above: the time it took to run the first three cases was faster in a MySQL query, but by less than 0.0X seconds, whereas the last 5 cases the NoSQL was faster ranging from 0.5-40 seconds.  As data size grew so did the processing time for the last five cases on MySQL grew.

Conclusions: The authors were able to show that with more highly advanced cases, MySQL takes more time than NoSQL. Thus, for big data analytics, NoSQL graph databases can help store dynamic relationship data as well as process more complex queries using fewer lines of code and faster than MySQL queries.  This style of storing data allows the end-user in the healthcare field to ask more complex questions and get those answers promptly.

References

  • Park, Y., Shankar, M., Park, B. H., & Ghosh, J. (2014). Graph databases for large-scale healthcare systems: A framework for efficient data management and data services. In Data Engineering Workshops (ICDEW), 2014 IEEE 30th International Conference on (pp. 12-19). IEEE.
  • Sadalage, P. J., Fowler, M. (2012). NoSQL Distilled: A Brief Guide to the Emerging World of Polyglot Persistence, 1st Edition. [Bookshelf Online].
  • Services, E. E. (2015). Data Science and Big Data Analytics: Discovering, Analyzing, Visualizing and Presenting Data, 1st Edition. [Bookshelf Online].

Diagnosis of illness via big data

Big data is defined as high volume, high variety/complexity, and high velocity, which is known as the 3Vs (Services, 2015). Using Machine Learning and Artificial Intelligence, they do well at analyzing patterns from frequent and voluminous amounts of data at faster speeds than humans, but they fail to recognize patterns in infrequent and small amounts of data (Goldbloom, 2016). Thus, the use of data analytic theories and techniques on big data rather than novel situations in healthcare is vital to understand.

Fayyad, Piatetsky-Shapiro, and Smyth (1996) defined that data analytics can be divided into descriptive and predictive analytics. Vardarlier and Silaharoglu (2016) agreed with Fayyad et al. (1996) division but added prescriptive analytics.  Thus, these three divisions of big data analytics are:

  • Descriptive analytics explains “What happened?”
  • Predictive analytics explains “What will happen?”
  • Prescriptive analytics explains “Why will it happen?”

(Fayyad et al., 1996; Vardarlier & Silahtaroglu, 2016). Depending on the goal of diagnosing illnesses with the use of big data analytics should depend on the theory/division one should choose.  Raghupathi & Raghupathi (2014), stated some common examples of big data in the healthcare field to be: personal medical records, radiology images, clinical trial data, 3D imaging, human genomic data, population genomic data, biometric sensor reading, x-ray films, scripts, and traditional paper files.

The use of big data analytics to understand the 23 pairs of chromosomes that are the building blocks for people. Healthcare professionals are using the big data generated from our genomic code to help predict which illnesses a person could get (Services, 2013). Thus, using predictive analytics tools and algorithms like decision trees would be of some use.  Another use of predictive analytics and machine learning can be applied to diagnosing an eye disease like diabetic retinopathy from an image by using classification algorithms (Goldbloom, 2016).

The study of epigenetics, which are what parts of the genetic code is turned on versus turned off, can help explain why will certain illnesses are more probable to occur in the future (What is epigenetics, n.d.).  Thus, the use of prescriptive analytics could be of some use in the study of epigenetics. Currently, clinical trials use descriptive analytics to help calculate true positives, false positives, true negatives, and false negatives of a drug treatment versus a placebo are commonly used.  Thus, depending on the goal of diagnosing illnesses and the problem, that should help define which theories and techniques of big data analytics to use. The use of different data analytics techniques and theories based on the problem and data can change how healthcare jobs in the next 30 years from today (Goldbloom, 2016; McAfee, 2013).

References

Using R and Spark for health care

Use of R with regard to healthcare field case study by Pereira and Noronha (2016):

R and RStudio have been used to look at patient health and diseases records located in Electronic Medical Records (EMR) for fraud detection.  Anomaly detection revolves around using a mapping code that filters data based on geo-locations.  Secondly, a reducer code which aggregates the data based on extreme values of cost claims per disease along with calculating the difference.  Finally, a code that analyzed the data that meets a 60% cost fraud threshold. It was found that as the geo-location resolution increased, the anomalies detected increased.

R and RStudio have been able to use big data analytics to predict diabetes from the Health Information System (HIS) which houses patient information, based on symptoms. For predicting diabetes, the authors used a classification algorithm (decision tree) with a 70%-30% training-test dataset split, to eventually plot the false positive rate v. True positive rate.  This plot showed skill in predicting diabetes.

Use of Spark about the healthcare field case study by Pita et al. (2015):

Data quality in healthcare data is poor and in particular that from the Brazilian Public Health System.  Spark was used to help in data processing to improve quality through deterministic and probabilistic record linking within multiple databases.  Record linking is a technique that uses common attributes across multiple databases and identifies a 1-to-1 match.  Spark workflows were created to help do record linking by (1) analyzing all data in each database and common attributes with high probabilities of linkage; (2) pre-processing data where data is transformed, anonymization, and cleaned to a single format so that all the attributes can be compared to each other for a 1-to-1 match; (3) record linking based on deterministic and probabilistic algorithms; and (4) statistical analysis to evaluate the accuracy. Over 397M comparisons were made in 12 hours.  They concluded that accuracy depends on the size of the data, where the bigger the data, the more accuracy in record linking.

References

  • Pereira, J. P., & Noronha, V. (2016). Anomalies Detection and Disease Prediction in Healthcare Systems using Big Data Analytics. Retrieved from http://www.aijet.in/v3/1608001.pdf
  • Pita, R., Pinto, C., Melo, P., Silva, M., Barreto, M., & Rasella, D. (2015). A Spark-based Workflow for Probabilistic Record Linkage of Healthcare Data. In EDBT/ICDT Workshops (pp. 17-26).

Big Data Analytics: Health Care Industry

Big data has influenced many industries. One area that has been greatly influenced is the health care industry. This post describes how big data is influencing personal genomics in the health care industry. This post also evaluates how analyzing an individual’s genomes can aid in the foundation of predictive and preventive medicine.

Since its inception 25 years ago, the human genome project has been trying to sequence its first 3B base pair of the human genome over a 13 year period (Green, Watson, & Collins, 2015).  This 3B base pair is about 100 GB uncompressed and by 2011, 13 quadrillion bases were sequenced (O’Driscoll, Daugelaite, & Sleator, 2013).  With the advancement in technology and software as a service, the cost of sequencing a human genome has been drastically cut from $1M to $1K in 2012 (Green et al., 2015 and O’Driscoll et al., 2013).  It is so cheap now that 23andMe and others were formed as a consumer drove genetic testing industry that has been developed (McEwen, Boyer, & Sun, 2013).  At the beginning of this project, the researcher was wondering what insights the sequencing could bring to understanding decease, to the now explosion of research dealing with studying millions of other genomes from biological pathways, cancerous tumors, microbiomes, etc. (Green et al., 2015 and O’Driscoll et al., 2013).  Storing 1M genomes will exceed 1 Exabyte (O’Driscoll et al., 2013).  Based on the definition of Volume (size like 1 EB), Variety (different types of genomes), and Velocity (processing huge amounts of genomic data), we can classify that the whole genomic project in the health care industry as big data.

This project has paved the way for other projects like sharing MRI data from 511 participants, (exceeding 18 TB) to be shared and analyzed (Poldrak & Gorgolewski, 2014).  Green et al. (2015) have stated that the genome project has led to huge innovation in tangent fields, not directly related to biology, like chemistry, physics, robotics, computer science, etc.  It was due to this type of research that a capillary-based DNA sequencing instruments were invented to be used for sequencing genomes (Green et al., 2015).  The Ethical, legal and Social Implication project, got 5% of the National Institute of Health Budget, to study ethical implications of this data, opening up a new field of study (Green et al., 2015 & O’Driscoll et al., 2013).  O’Driscoll et al. (2013), suggested that solutions like Hadoop’s MapReduce would greatly advance this field.  However, he argues that current java intensive knowledge is needed, which can be a bottleneck on the biologist.   Luckily, this field is helping to provide a need to create a Guided User Interface, which will allow scientist to conduct research and not learn to program.  O’Driscoll et al. (2013), also state that the biggest drawback of using Hadoop MapReduce function is that it reduces data line by line, whereas genomic data needs to be reduced in groups.  This project, should, with time improve the service offering of Hadoop to other fields outside of biomedical research.

In the medical field, cancer diagnosis and treatments will now be possible due to this project (Green et al., 2015).  Green et al. (2015) also predict that a maturation of the microbiome science, routine use of stem-cell therapies could result from this.  These predictions are not far from becoming reality and are the foundation of predictive and preventative medicine.  This is not so far into the future that McEwen et al. (2013) have stated what are the ethical issues, for people who have submitted their genomic data 25 years ago, and they found data that could help the participants take preventative measures for adverse health conditions.  Mostly because clinical versions of this data are starting to become available like from companies like 23andMe. This information so far has yield genealogy data, a few predictive medical measures (to a certain confidence interval).  Predictive and preventative medical advances are still primary and currently in the research phase (McEwen et al., 2013).  Finally, genomics research will pave the way for metagenomics, which is the study of microbiome data of as many of the ~4-6* 10^30 bacterial cells (O’Driscoll et al., 2013).

From this discussion, there is no doubt that genomic data can fall under the classification of big data.  The analysis of this data has yielded advances in the medical fields and other tangential fields.  Future work, to expanding the predictive and preventative medicine is still needed; it is only in research studies, where the participants can learn about their genomic indicators that may lead them to certain types of adverse health conditions.

Resources:

  • Green, E. D., Watson, J. D., & Collins, F. S. (2015). Twenty-five years of big biology. Nature, 526.
  • McEwen, J. E., Boyer, J. T., & Sun, K. Y. (2013). Evolving approaches to the ethical management of genomic data. Trends in Genetics, 29(6), 375-382.
  • O’Driscoll, A., Daugelaite, J., & Sleator, R. D. (2013). ‘Big data,’ Hadoop and cloud computing in genomics. Journal of biomedical informatics, 46(5), 774-781.
  • Poldrack, R. A., & Gorgolewski, K. J. (2014). Making big data open: data sharing in neuroimaging. Nature neuroscience, 17(11), 1510-1517.