In a new study published by Nature, the Southern California Permanente Medical Group shares results on the Kaiser Permanente Intelligent Navigator (KPIN), a new platform using artificial intelligence natural language processing to augment the patient portal by reducing care delays, recommending appropriate care pathways and generating alerts for high-acuity cases.
This study provides large-scale, real-world evidence of an AI-driven system successfully completing patient digital interactions – be it by accessing services, communicating with providers or self-scheduling appointments – rather than simply redirecting to call centers, which is an important step forward in digital first care.
THE CHALLENGE
Healthcare organizations are evolving to meet the expectations of a consumer-focused environment, where patients increasingly seek more personalized care and greater convenience, said Dr. Khang Nguyen, assistant executive medical director for care transformation, Southern California Permanente Medical Group, and chief medical officer for care navigation, The Permanente Federation.
"Online patient portals are now a mainstay for accessing services, communicating with providers and self-scheduling appointments," he said. "However, most healthcare portals provide patients with a list of options that may not adequately meet their specific care needs. We saw an opportunity to better understand what members and patients wanted and to enhance the quality-of-care delivery described in their own words.
"Kaiser Permanente is a value-based, integrated care delivery system that provides high-quality, affordable healthcare for our patients and the communities we serve," he continued. "Core to this model is a commitment to optimizing both the member experience and care outcomes across our system."
A key component of fulfilling this mission is the digital portal. Smartphones have become the critical channel for navigating nearly every aspect of life, including health.
"As a result, we developed the KPIN for online appointments in our Southern California region," Nguyen explained. "It has been rolled out to 4.9 million Southern California Kaiser Permanente patients beginning in October 2024.
"This isn't just another digital enhancement, it is a reimagining of how patients navigate care on a massive scale," he continued. "While most healthcare portals redirect patients to call centers, the KPIN system actually completes the patient's digital encounter by producing a set of recommended offerings best suited to the individual case."
PROPOSAL
Patients prefer digital experiences that will understand their requests and guide them to the right place to receive the care they need at the right time, Nguyen said.
"The AI NLP technology addresses this preference by allowing patients to describe their care needs in their own words when booking online," he offered. "The system can then offer more relevant options for convenient care resolution.
"High-acuity cases that need immediate clinical triage can be identified upstream and patients directed to the appropriate clinician," he added.
MEETING THE CHALLENGE
Kaiser used AI NLP to better understand patient requests using free text and offer patients appropriate care choices. Traditional models require patients to scroll through numerous pre-set choices or menus that may not meet their needs.
"Specifically, NLP is integrated into the KPIN tool that we developed in our Southern California region to optimize patient navigation within our value-based care delivery model," Nguyen said. "KPIN works as a front door that allows patients to describe their needs in their own words. This system allows us to better understand the patient's requests, to personalize their experiences, and to guide them to the appropriate care team.
"We have a success rate of getting patients to the right appointment one out of every two times," he continued. "The industry standard is one out of every three times."
RESULTS
With KPIN, Kaiser detected urgent medical cases accurately 97.7% of the time, Nguyen reported.
"KPIN takes patient-submitted free text and detects high-acuity cases as part of its clinical alert system," he explained. "It then recommends timely and appropriate booking options. In addition, appropriate care pathways are recommended accurately 88.9% of the time. This drives improvement for patient outcomes by reducing delays in care delivery and enhances personalized care experience.
"To optimize performance, custom few-shot transformer models were implemented as the foundation for both CAS and CNS," he continued. "A transformer is a type of AI model that learns patterns in data, especially in text. It excels at understanding the relationships between words, sentences or other pieces of information. Humans do the same when guessing what others are about to say."
This machine learning model reduces any time delays from patient input to providing an appropriate output. This enables rapid decision making for real-time clinical applications. This approach ensures both rapid inference and high accuracy, he added.
ADVICE FOR OTHERS
"At Kaiser Permanente, we believe AI tools and systems can automate tasks, but they will not replace the essential role of our doctors and care teams," Nguyen said. "This approach guides our integration of AI tools.
"In practice, the success of these efforts depends on the expertise and leadership of our clinical teams," he continued. "A critical success factor for KPIN in our Southern California region has been the role of physician subject matter experts. Their domain knowledge guided our training models."
KPIN performance is continually optimized by a multidisciplinary team of physicians, nurses and other clinical staff. This ensures alignment with booking and care standards.
"Frontline physicians along with operational staff also provide feedback," he concluded. "This creates an end-to-end feedback loop for continuous performance improvement. Patient informed, clinically led and multidisciplinary by design, KPIN ensures clinical safety while improving care experience."
Follow Bill's health IT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.
WATCH NOW: EHR transitions: How to ensure adoption and satisfaction
