Storing, accessing and sharing electronic medical data with intelligent patient matching
In the age of smartphones, cloud storage, and the internet of things, we have come to expect the information we want to be at our fingertips in seconds. One notable exception to this rule is medical records. Individuals viewing test results, doctors accessing a new patient’s medical history, and health insurance agents corroborating claims all need to access the same highly sensitive data, but medical IT has been slow to progress because of concerns about keeping that data secure.
Rewards worth the risks
The HIPAA Privacy Rule sets strict standards for how individuals’ medical records and other personal health information is handled. HIPAA violations are extremely costly both in terms of monetary fines and reputation damage, leading many healthcare providers to be hesitant about embracing the convenience of cutting-edge technology.
As medical data becomes digitized, the risk of compromising HIPAA-protected data only grows, however, the value of electronic medical records (EMRs) is too significant to overlook. Patients receive better care when they and their doctors have instantaneous access to all the data they need.
In 2014, the Office of the National Coordinator for Health Information Technology (ONC), part of the US Department of Health & Human Services, released their ten-year vision to achieve an interoperable health IT infrastructure, identifying patient matching as an early milestone in the project to create a robust health information exchange system (HIE). The AHIMA foundation expands on the vital importance in their white paper “Patient Matching in Health Information Exchanges:”
Lack of a standard data set can lead to patient records not being linked to one another in the HIE, resulting in an incomplete health record being available to the provider for the patient being treated, thereby defeating the purpose of the HIE. Even more concerning is the potential for different patients being identified as the same, resulting in the possibility of improper care rendered on the basis of inaccurate patient information. In addition to patient care concerns, sharing inaccurate information also poses the risk of privacy breaches and erodes consumer confidence in the benefits of HIE.
Reliable patient matching then can both improve upon the value created by HIE systems, as well as mitigate the risk of compromising data that HIE creates.
Fuzzy name matching for medical records
Patient matching is not a simple search problem because names are the key data type, presenting a unique challenge for search tools. Names are shorter than documents, offering fewer opportunities to correctly identify a match, and a single name can have hundreds of variations including alternate spellings, nicknames, or transliterations from foreign languages.
Take for example the name Caitlin Murray Smithe. Legacy search tools would only find Caitlin’s name in a database if the search query is an exact match to the record. HIE systems must recognize the similarity of Katelyn M. Smithe, Cait Smith, Smithe, C. M., and more without relying on a simple list of potential variations that must be constantly updated and will inevitably have holes.
But how to connect these seemingly disparate but connected names? Approximate string matching, colloquially called fuzzy string matching or fuzzy matching, is a computer science technique of finding strings that match a pattern approximately rather than exactly, returning more results than a more strict search. A standard feature in any search tool, string matching is doubly important for names search, putting the solution’s deep knowledge of names to work. An accurate and powerful name matching solution must have both string matching capabilities and a deep understanding of names.
Powering patient search
Patient matching of the caliber that the Office of the National Coordinator is calling for requires intelligent, deep understanding of names, enabling a simple search query to find potential matches in the database regardless of errors commonly to medical records like nicknames, misspellings, names split inconsistently across database fields and more.
A database search tool with fuzzy name matching presents all likely matches for each search query based on the match score threshold set by system managers. Possible matches are ranked from most to least likely, allowing hospital staff to review the results and ask the appropriate follow-up questions to find the right patient record faster. This further protects against risks of misidentifying patient records as the final decision is made by a human rather than a machine.
Fuzzy name matching improves the accuracy of your records search, ensuring that staff finds the correct patient record on the first try, providing better care to the patient in question, preventing the accidental disclosure of HIPAA-protected data to the wrong party, and saving time and money.
Looking to learn more about fuzzy name matching and the challenges of patient identification? We’ve already done the research for you:
- Effective Healthcare Identity Management: A Necessary First Step for Improving U.S. Healthcare Information Systems
- Unpacking the costs of patient misidentification on a hospital’s bottom line
Ready to add fuzzy name matching to your database search tool? Babel Street’s Rosette text analytics combines string matching with phonetic comprehension, etymology, onomastics and advanced natural language processing to creates the ideal fuzzy name matching system for EMRs. Rosette’s trusted blend of machine learning and traditional name matching techniques improves name search for the world’s thorniest name matching challenges.