By Jacob Boye Hansen, Founder & CEO CareCom A/S
CareCom has been working in the clinical terminology space for many years, and in the last 3 years we have seen a consistent interest in analytics. Interest is coming from not only terminology services clients, but health care technology organizations in general. Many healthcare software companies have implemented analytics taking many forms – everything from population health analysis to big data analytics.
CareCom’s terminology server HealthTerm®, is a state of the art enterprise terminology server, with many large organizations implementing our solution. Some clients are using our application for development and maintenance of terminology/classifications such as subsets, mapsets and language translations. Other clients are using the power of the enterprise terminology server for heavy lookup services in a run-time environment, with use cases such as:
- Lookup description for a code for any code system (SNOMED CT, LOINC, RxNORM etc.)
- Crosswalk from a local/custom code system to a standard code system
- Apply content of a specific type of codes in a dropdown list (allergy for example)
- Auto-mapping local term to a standard code system, using cognitive intelligence to find the mapping for large amount for terms to be mapped to standards
- Lookup sensitive codes across a patient record
- Group diagnoses, procedures and medication that are related
- Lookup medication details and groupings
So, how can we add value with analytics?
First, let’s separate web-page analytics such as Google Analytics and big data analytics such as IBM Analytics.
Is it interesting to know who visit which page in our web-based maintenance tool? It might be for some clients, but most want more detail. All objects in HealthTerm have history logged for all changes, which includes documented date, by whom, publish date and by who and valid from date. HealthTerm has already logged all the information needed for transparency – and analytics of the same information would probably deliver limited value. I agree it would be nice to know who did most mappings or translations of the best quality, but HealthTerm already has statistical reports for this function and brings more detailed information than page analytics.
Big Data Analytics
For all our clients who use our API for lookup or decoration of data, I do see value in big data analytics. Some of my ideas are:
- For an HIE client: It could be valuable to know which participating organization contributes codes that are not map to standard; or which organization still reports using ICD-9 codes.
- For a hospital accessing HealthTerm through a browser: Which terms are being searched, and does it deliver a result? Is there missing interface terminology content?
- From an operations perspective: How fast are lookups and when is the peak access time(s)?
- From a clinical perspective: Which diseases are most often observed together?
- From a reimbursement perspective: Which measures are achieved by which organizations?
- From a financial perspective: Which costly drugs are used by which organizations?
To answer the call of our current and prospective client analytics use cases, HealthTerm added support for IBM Analytics tool. This new feature logs all web-service calls with input and output parameters as well as performance data.
I imagine that there are many more valuable use cases, especially when we apply Natural Language Processing (NLP) indexing on free text in clinical notes. (You can read more about this on my next blog). In the meantime, let me know more about your high value use case(s) utilizing terminology analytics. What do you think is the highest value use case for big data analytics?