By Daniel Cidon, CTO of NextGate
Twitter: @NextGate
Twitter: @dcidon
Since the dawn of the digital revolution in healthcare, providers have been in search of tools to connect disparate systems and deliver timely and secure patient data directly to the point of care. Our public health crisis is a sobering example of just how desperately seamless information sharing capabilities are needed.
While EHRs have become commonplace, the disjointed, competitive nature of IT systems contributes to a proliferation of duplicate and incomplete patient records. The inability to consistently match patients to their data can have dire consequences, leading to inappropriate medications being dispensed, incorrect diagnoses, erroneous test results and increased risk from redundant medical procedures.
The problem not only endangers patient safety but contributes to needless waste and inefficiencies across the care continuum. The financial impact is substantial—duplicate records cost the U.S. healthcare system more than $6 billion annually and individual hospitals $1.5M a year.
With focus on solving patient matching challenges deepening, hospitals and healthcare companies are exploring use of artificial intelligence (AI) to support identity management. As AI continues to mature, the potential to optimize the technology for patient matching is immense.
EMPI and Referential Matching
Despite widespread adoption, the disjointed and competitive nature of EHR systems contributes to a flood of duplicate and disparate medical records. Data contained in EHRs is notoriously inconsistent, since disparate systems capture patient demographic elements in different ways. For instance, some providers or EHR systems will include hyphens, apostrophes and suffixes in last names; others don’t.
A study led by researchers at The Pew Charitable Trusts, found EHR matching rates within facilities as low as 80 percent‑—meaning one out of five patients may not be completely matched to his or her record. When exchanging records outside the organization, match rates can be far lower, at just 50 percent—even when the providers are running the same vendor EHR.
To combat the data fragmentation, EHRs and other health IT systems must communicate with each other to support the accurate and efficient exchange of clinical data. Integrating an Enterprise Master Patient Index (EMPI) is one way to ensure that patients only have one accurate record in a health system.
Many healthcare organizations are already leveraging EMPIs to identify and link patient data spread amongst multiple systems. EMPIs use algorithms to compare demographic data – name, address, etc. – from two patient records to determine if they belong to the same person, and also flag any potential duplicates.
Health providers can take this a step further by augmenting their EMPIs with AI-powered tools, like referential matching, to achieve new levels of patient matching accuracy. Using AI and machine learning, referential matching compares more than the often-outdated demographic data within hospital records. Instead, it applies data-driven technology to match patient demographic data to a comprehensive, constantly updated reference database of identities.
Combining these two approaches supports inter-organizational interoperability and can augment healthcare facility data with information from public records and proprietary data—like utility records, credit reports, etc.—to build an accurate “golden” record for every patient.
Clinical Trial Recruitment
Working in concert with EHR and patient matching technologies, AI can also serve to mine medical records and accelerate the identification and recruitment of patients for clinical trials.
Clinical trial enrollment has been a long-standing problem in healthcare. From the complications of raising awareness among clinicians to the time-consuming process of screening patients based on invalid or outdated EHR data, issues with clinical trial recruitment delay late-stage trials, increase operational costs, and disconnect potential patients from potentially life-changing care.
To access accurate, real-world data for suitable patient identification, health officials have turned to AI for enrichment of the patient matching process. Integrating intelligent technology like machine learning into your EHR or EMPI improves clinical trial matching, allowing doctors to essentially filter out the best candidate for the trial and avoiding instances of unqualified candidates, patients out of region, and those who have been improperly registered.
Machine learning can rapidly assess multiple texts, graphs, and other data simultaneously, and faster than any health information management (HIM) professional. By seamlessly extracting patient medical data like symptoms, diagnoses, and test results history from verified patient health records, companies can aggregate data points to develop a clinical profile of registered patients.
These clinical profiles are utilized, along with EMPI technology, to find and compare populations and individual patients that meeting user search criteria across numerous systems and locations.
AI also lessens cases of limited enrollment due to patients and doctors being unaware of existing clinical trials. Using Natural Language Processing (NLP) to read doctors’ reports, diagnoses, and recommendations, this technology can deliver a 360-degree view of patient data that expedites the matching process and puts clinicians and other health professionals in touch with the right patients.
The faster clinicians match patients to the right trials, the faster they can develop, test, and get cures out to market.
Geocoding
Because patient matching is often based on demographic elements such as name and address, pinpointing geographic locations through the use of latitude and longitudinal coordinates are key for geo-spatial data analysis.
Using real-time location intelligence and machine learning, geo-spatial data can add a new dimension to EMPI capabilities, helping organizations target social determinants of health (SDOH), as well as the spread of COVID-19, by providing insights into geographical and social risk factors.
As an individual’s precise location becomes more important in helping organizations identify communities at risk for disease, natural disaster, poverty, or poor water purity, location data can be useful for targeted interventions of at-risk populations. Public health officials fighting COVID-19, for example, can utilize geo-spatial data to predict patterns and trace hotspots of potential new outbreaks.
Building the EMPI of the Future
Correctly and consistently linking patient data remains a significant challenge for hospitals and HIEs around the world.
Health systems are increasingly making technology investments not just to reduce costs or improve efficiencies but also because today’s patients deserve the best care available. In order to accomplish that goal and remain competitive in the market, it’s important for healthcare organizations to embrace the advancement of AI in the healthcare industry as a symptom of progress and evolution.
This article was originally published on the NextGate Blog and is republished here with permission.