Snow Owl MQ: a semantic platform for healthcare data processing

Snow Owl® Meaningful Query (MQ) is a scalable Big Data software platform for

  • Searching and browsing health records
  • Grouping patients that share the same characteristics into cohorts
  • Inspecting records to identify trends and correlations
  • Statistical analysis of patient cohorts to test and verify clinical hypotheses

The platform covers the entire data processing workflow such as data import, real-time exploration, interactive querying, and production analysis including sophisticated machine learning algorithms – while providing access to the semantics captured in the healthcare data at every step of the process. The unified platform enables data professionals to use a single system or software developers to utilize a single set of APIs in order to implement their own semantic healthcare data processing applications.

Clinical terminologies

The foundation for future-proof analytics.
Learn more

Patient cohort builder

Self-service patient cohort analysis.
Learn more

Healthcare analytics

Create, document, and publish analytics findings.
Learn more
0
Unleashed Terminologies
0
Unlocked patient records
0+
Countries
Try Snow Owl MQ Now!

Clinical terminologies

Use terminology subsets and health outcomes of interest created and curated by dozens of governments and organizations collaborating to improve the health of humankind.

Our terminology browser and semantic search interface creates terminology filters to retrieve subsets of concepts based on various criteria. Search methods can be combined (e.g. lexical, subsumption, semantic, subset, mapping) to simplify concept retrieval for different domains such as pharmaceuticals, findings, procedures, and observations.

Cross-terminology filters can be created based on their medical significance, public health implications, or historical associations with drug toxicities. They can be saved and shared with the community.

Explore cross-terminology health outcomes of interest from the Foundation for the National Institutes of Health (USA), Johnson & Johnson, IMS Health, Merck Research Laboratories, Janssen Research and Development, Reagan-Udall Foundation for the Food & Drug Administration (USA), Health, AstraZeneca, Siemens Health Services, Pharmaceutical Research Manufacturers of America, Oracle Health Sciences, and others.

SNOMED CT is considered the most comprehensive, multilingual clinical healthcare terminology in the world. Its use is mandated for healthcare information exchange in over 20 national healthcare systems and recommended in dozens more. This has led to exponential growth of SNOMED CT encoded health records.

In contrast to other clinical terminologies, SNOMED CT is a formally defined ontology. This allows identifying concepts based on clinical meaning in addition to textual labels or pre-defined categorizations. Concepts may have multiple parents, allowing Viral pneumonia to be categorized as both a Lung disease and an Infectious disease. Mappings to other terminologies like ICD-9/10, LOINC, ATC, etc. permit even more categorization and grouping options.

Snow Owl MQ unleashes SNOMED CT, supporting extensions from the International Health Terminology Standards Development Organization (SNOMED International), National Library of Medicine (USA), National Health System (UK), Kaiser Permanente (USA), Ministry of Health (Singapore), National IT Institute for Health care in the Netherlands, Ministry of Health and Family Welfare (India), National E-Health Transition Authority (Australia), Canada Health Infoway, National Agency for e-Government and Information Society (Uruguay), Secretary for Food and Health (Hong Kong), and many more.

Support for over 70 standard terminologies, including SNOMED CT, ICD-10, LOINC, ATC, RxNorm, dm+d, ICD-9, MedDRA, CPT, along with regional extensions such as ICD-10-CM.

Include your own terminologies, local code systems, mappings, and value sets. Create your own queries by specifying a set of rules and rest assured that new content from the above authorities automatically appears correctly in your subset. The friendly user interface requires no expertise knowledge but provides the power of an ontological foundation based on formal description logic.

Patient cohort builder

Create patient cohorts by querying electronic health records for patients that meet particular criteria, including demographic traits, drug exposures, clinical findings, procedures, and observations.

Snow Owl MQ supports clinical research focusing on evidence-based healthcare through observational studies:

  • Cohort based: comparing similar patient groups with the exception of an exposure
  • Case-control: find patient groups with a problem (case) and without a problem (control) and see if the case is more likely present in the control than a particular population
  • Patient selection for experimental studies, such as randomized clinical trials

The cohort builder allows combining semantic queries with temporal criteria and demographic data to explore patient populations. Temporal criteria are expressed in relation to a baseline event such as a drug exposure or a condition. Baseline exposures and multiple inclusion and exclusion criteria can be defined and combined with logical operators. Matching results are displayed immediately as the cohort is created.

Visualize individual patient records as a sequence of events displayed on a timeline. The observation period is automatically zoomed to a level that displays the baseline exposure along with all events relevant to the selected inclusion and exclusion criteria. The observation period can be increased or decreased to expand or reduce the timeframe. The view can be toggled between relevant events only or to include all record details.

Conditions, procedures, and drug exposures are interactive, allowing hypotheses testing by selecting and browsing random sets of patients. Display of condition and drug eras indicate changes in health or prescribed medications.

Use your own patient records with existing plug-ins and tooling that simplifies ETL.

Or analyze 1 billion existing patient records from Clinical Practice Research Datalink (CPRD), Truven MarketScan Commercial Claims & Encounters (CCAE), Multi-State Medicaid (MDCD), Medicare Supplemental Beneficiaries (MDCR), Lab Supplemental (MSLR), GE Centricity Medical Quality Improvement Consortium, Premier, National Health and Nutrition Examination Survey (NHANES), OptumInsight ClinFormatics Data Mart, Healthcare Cost and Utilization Project (HCUP), Indiana Network for Patient Care, CMS Limited Data Sets, SAFTINet (Scalable Architecture for Federated Therapeutic Inquiries Network), and more.

Healthcare analytics

Interactively query and visualize healthcare data. Determine which patient cohorts are at risk for adverse events. Identify the most effective markets for new products. Find the population segment most affected by an outbreak. Perform casemix analysis. The sky’s the limit.

Notebook-style development provides an exploratory approach to data analysis. Notebooks are composed of code blocks called cells, which contain statistical languages like R and Scala or rich-text documentation. The cells can be independent or act together. This creates a discovery-based approach to analytics, where an analyst experiments in one cell, and then continues in subsequent cell based on results from the previous cell.

Organizing your data in rich-text notebooks lets you share your findings and visualizations with multiple users. You can even collaborate with others in real time.

Analytics notebooks run everywhere without changes, whether on a local server, a private cluster, or in the cloud. Support for both batch and real-time data stream processing and mining. Analytics notebooks can be executed immediately or scheduled to run as a job. This allows compute-intensive jobs to run on a lower budget on cloud services like Amazon EC2 Spot Instances.
Analyze patient data with R, Scala, or Python; or export data for use in SAS and other systems. Combine R’s mature collection of machine learning libraries with distributed, massively parallel computing capabilities to process datasets beyond the ability of standalone R programs. We provide simple APIs for operating on large datasets, including over 100 data transformation operators. Standard libraries include support for over 70 healthcare terminologies, SQL queries, machine learning, and graph processing.

Easily include external analytics packages like OHDSI’s patient-level prediction and population-level estimation methods libraries. Visualize your results with any visualization package such as ggplot2, ggvis, rcharts, googleVis, matplotlib, and D3.js.