Accurate person identity data is one of the most difficult challenges in healthcare today. The stakes are high; merge the wrong records, and you create a privacy risk; fail to link the right ones, and critical patient history gets lost. Health Tech organizations risk loss of productivity by up to 30% according to McKinsey when they reallocate staff dedicated to product innovation to activities not related to roadmap priorities, like data cleansing, instead of leveraging automation. The loss of trust when poor quality data causes care or compliance issues is detrimental to businesses and can be costly – even leading to serious lawsuits.
Historical person identity-matching solutions have faced challenges, often falling short when faced with inconsistencies like name variations, outdated addresses, or inconsistent birthdates. Records are flagged for manual review when that happens, requiring data stewards to step in and make the final call, which can take them up to three minutes per record.
To address these challenges, Rhapsody introduced Autopilot, a machine-learning enhancement to its Enterprise Master Person Index (EMPI) – the next generation of this highly effective technology. Autopilot AI is designed to replicate the decisions a human would make, at a scale and speed that’s impossible with manual workflows. The automation reduces up to 90% overhead costs because it can process 50,000 hours’ worth of data in only 80 minutes.
Why record matching is so complex
The primary function of an EMPI is to manage demographics and ensure that a single person record is accurately represented across various data sources. This isn’t always straightforward. If a new record enters the system, the EMPI has to determine whether it belongs to an existing patient or an entirely new one.
If the details match exactly, it’s easy to determine the record belongs to the same patient. But in real-world scenarios, discrepancies are common – names change, addresses are outdated, and data entry errors occur. Most EMPIs rely on fixed matching criteria, meaning if a record doesn’t meet certain thresholds, it’s flagged for manual review. This creates a backlog of potential duplicates, which grows significantly as data volumes increase, forcing data stewards to intervene manually. Manual intervention is:
- Slow: Reviewing each flagged record takes minutes, adding up to weeks of backlogged work for engineering teams. Or, if the issues are not resolved, it can lead to support issues downstream that get back to the engineering team to sort out.
- Expensive: Manual reconciliation and issue resolution consumes engineering resources that could be dedicated to product innovation instead.
- Inconsistent: Different data analysts may apply slightly different logic to resolve conflicts, leading to inconsistencies in person data matching.
How Rhapsody EMPI with Autopilot improves record matching
Rather than rely solely on conventional person data-matching algorithms, Rhapsody EMPI with Autopilot uses machine learning combined with referential matching to analyze patterns in data and make identity-matching decisions the same way a human would in an AI mimicking human brain patterns in a neural network. Instead of treating every slight discrepancy as a red flag, it evaluates the full context of the record.
For example, traditional algorithms might flag the discrepancy for manual review if a patient’s name is reversed and one of the records has a nickname, e.g. “Flynn Robert” in one record and “Bob Flynn” in another. On the other hand, Rhapsody EMPI with Autopilot understands that these variations are common and determines whether the records belong to the same person.
The fact that Autopilot leverages these technologies in tandem means it can adapt to the differences in a variety of healthcare populations.
Balancing automation with human oversight
Concerns around AI in healthcare often revolve around accuracy, reliability, and control. If an automated system merges records incorrectly, protected health information (PHI) could be exposed. If it fails to link related records, clinicians may lack access to a person’s entire medical history, risking inaccurate or insufficient care due to an incomplete picture, not to mention regulatory non-compliance.
To mitigate these risks, the Rhapsody EMPI with Autopilot does not replace human oversight – it enhances it.
- It processes the record automatically if it’s highly confident in a match. If there’s uncertainty, it flags the record for human review.
- Every match maintains a full audit trail, providing transparency into who merged records and when – whether it was a human or AI – and what system the data in the records came from
- The system complies with HIPAA, ISO 27001, SOC 2, HiTrust e1, and other global regulatory requirements
Laura Campomizzi, Director PMO and IT Operations at Rochester RHIO, recognized this shift in thinking firsthand:
“AI is new, and people think it will replace jobs. But you have to teach it the logic and monitor it. It will streamline your work, but people are still required. Now, these people can focus on other important items, high-value work.”
By automating person matching, AI ensures that human experts focus on the most complex cases, reducing errors and improving efficiency.
The business case for AI-driven person matching
Duplicate or inaccurate records contribute to costly inefficiencies, from non-compliance fines to manual reconciliation efforts that require dedicated staff. In the US, HIPAA compliance risks grow when incorrect matches expose sensitive person data, with potential fines of up to $1.5 million per incident.
Manual reconciliation is resource-intensive, requiring a team of data stewards to review flagged records individually. Using Rhapsody EMPI with Autopilot, health tech organizations can scale their person data management operations without increasing headcount through automation. This ensures person data is managed efficiently, accurately, and in compliance with regulations.
Before implementing Rhapsody EMPI, Geisinger Health, a large integrated health system, directed staff to manually clean about 150 records per day, or 36,000 annually. Using Rhapsody automation, Geisinger saved more than $78,000 per year simply on data cleansing — while freeing up the time of five FTEs to work on more important matters.
Many health tech organizations start out managing person identity data manually, using homegrown solutions or algorithms. While this works on a small scale, it quickly becomes unsustainable when additional data sources are connected. As healthcare databases grow, the number of uncertain matches increases, requiring either a larger team or a more advanced EMPI solution.
Cameron Kerber, VP of application development at Monogram Health, a health technology solution provider, explains how he thought through the option of building and maintaining a homegrown EMPI solution: “Rhapsody has the expertise, and we had to ask ourselves: do we want to continue to maintain, not only our internal software, but do we also want to do the work around all the downstream challenges (duplications, processes)? Who can support that?” For him, the decision was clear to trust Rhapsody when person data matching was not Monogram’s core competency… not a revenue driver for the business.
Reducing risk with the right EMPI is a no-brainer
AI-driven person data matching isn’t just about efficiency – it’s about reducing risk. Rhapsody EMPI with Autopilot enables organizations to scale without adding headcount, all while increasing data accuracy and enhancing compliance. For health tech teams focused on innovation, managing person data manually is an unnecessary burden. Instead of dedicating resources to maintaining homegrown solutions, organizations can rely on trusted industry experts. Learn how the Rhapsody EMPI with Autopilot can improve operational efficiency and reduce risk for your organization.
