The Human Touch
By Rachel Podczervinski MS, RHIA, Senior Vice President, and Julie A. Pursley, MSHI, RHIA, CHDA, FAHIMA, Director of Industry Relations, Harris Data Integrity Solutions
Neither humans nor AI are infallible, but when it comes to person matching, humans have the edge. They may create identification errors, but humans are also integral to identifying, verifying, and correcting them.
Humans are adept at recognizing inconsistencies that AI-enabled technologies—such as MPI/EMPI with advanced algorithms, biometrics, machine learning models, and predictive analytics with augmented data—may overlook because AI cannot make contextual judgments and decisions based on nuanced considerations.
For example, humans can make distinctions based on judgment and decision-making, creativity, innovation and agility, emotional intelligence, and empathy. Further, the complexity of patient data, coupled with the similarity of demographic details among individuals, creates a distinct set of challenges that AI alone cannot fully address.
AI Governance Framework
To be successful, AI requires high-quality data that is managed within an AI governance framework that includes a human-in-the-loop component to support and elevate patient identification strategies. The governance framework should involve a variety of stakeholders responsible for the management of patient identities and comprehensive quality assurance protocols such as working duplicate queues daily, routinely measuring error rates, and conducting regular quality audits.
Data entrusted to AI must be of the highest quality and integrity and managed within an AI Governance Framework that includes The Human Touch and Caring Algorithms.
It’s crucial that the organization has both an AI Governance Specialist and an EMPI Analyst. These individuals should collaborate on addressing governance related to data entrusted to AI and adapting systems and processes to identify patients across diverse demographics. This could involve transparency in processes, raising awareness of the data lifecycle, and exhibiting a comprehensive understanding of where the data has originated from (provenance).
Misidentification Risk
The integration of AI in person identification and matching processes offers significant opportunities to enhance patient safety and care quality. The full realization of these benefits requires a balanced approach that combines the strengths of AI with the irreplaceable insights of the human touch.
By fostering a collaborative environment where technology and human expertise work within an AI Governance Framework, healthcare providers can ensure the safe and effective application of AI in identifying patients, family, and loved ones.
Read “The Human Touch,” which is the first installment of Entrusting Data in an AI World: A Framework for Accurate Healthcare Data, a 3-part white paper on the impact of AI on person matching.