02-05-2025
How AI is improving patient care and boosting health care efficiency
There's no corner of the health care system that hasn't been — or will soon be — augmented by AI.
That was the takeaway at a recent health care symposium co-hosted by the San Francisco Business Times, Silicon Valley Business Journal, and Sacramento Business Journal on April 10 in San Francisco.
Intersect 2025: AI-Health Care brought together health care leaders, clinicians, researchers and technologists for peer-led learning and networking. It also served as a celebration for Northern California's pivotal role as a hub for innovation in AI and health care.
AI is already facilitating stronger patient-provider relationships, noted David Joyner, CEO of Hill Physicians. 'This is the first time in a very long while that I've seen new technology being embraced by physicians,' Joyner said during a panel conversation. 'Most of the time, new technology that results in more clicks and data entry is greeted by physicians with groans.'
To illustrate, Joyner conducted an informal poll of the audience: 'How many of you use MyChart or similar to message your doctor or your care team?'
He was met with a sea of raised hands.
'This is an avalanche of new work that providers don't get reimbursed for. The AI tool that Epic has created is helping to reduce that burden," Joyner commented.
Joyner pointed to the rise of AI scribe technology, which generates the visit chart notes in real time, as an example of technology that healthcare providers are genuinely excited about. 'The doctor-patient relationship and the documentation are dramatically improved,' he said.
Dr. David Klein, Chief Executive Officer of MarinHealth, agreed. 'AI scribes allow our providers to spend more time with patients and less time on paperwork,' he remarked.
AI is also making medicine more accessible to people of diverse backgrounds, the panelists observed. 'Most of the healthcare system today offers English or Spanish as your choices of language,' said Joyner. 'AI tools have enabled an explosion of not just languages, but also reading levels, that can be delivered to patients.'
Joyner referenced an AI scribe tool that Hill Physicians is piloting, which can summarize patient encounters conducted in dozens of languages.
Richard Milani, Chief Clinical Innovation Officer at Sutter Health, reported that Sutter is building a health education program using AI avatars that can speak to patients in more than 100 languages. The technology is personalized to the patient's health and linguistic literacy level, he said, meaning it can provide content at a level that the patient can understand. If the patient's literacy is below a third-grade level, it can even deliver content in the form of a cartoon.
'The idea is, how can we personalize health education based on a variety of attributes that cut across populations,' Milani said.
MarinHealth's Klein was enthusiastic about the predictive power of AI to improve efficiency in hospitals. 'The complexity of staffing a hospital is enormous,' he said. 'You have patients coming in, undergoing surgery in the ER, moving to beds and getting discharged. There's a whole series of healthcare providers, including physicians, nurses and techs who organize and orchestrate that.'
'These predictive AI models are going to help save us time and be more efficient,' Klein continued. 'I see it as an opportunity to help us triage in a resource-constrained environment to strategically focus our teams.'
The prospects among the panelists for predictive medicine, which quantifies patient risk so care teams can intervene before a patient's condition deteriorates, were equally promising.
'Our in-house data team has built a proprietary model for predicting patients who are at high risk for heart failure,' said Dr. Hemal Kanzaria, chief of performance excellence at Zuckerberg San Francisco General Hospital and Trauma Center.
'As a result of their efforts, we've driven down readmission rates for our patients with heart failure. We've been meeting pay-for-performance benchmarking that allows for financial ROI. And we've driven down the mortality rate for Black and African American patients,' who, according to the American Heart Association, have the worst clinical outcomes for heart failure.
ZSFG reported that as of last year it has closed the 5.4% readmission gap for its Black and African American heart failure patients.
Mulani described a similar AI model that's under development at Sutter. 'We asked ourselves, 'What if we could predict hours in advance that somebody's likely to have one of these arrests or rapid deteriorations?' And now we know with pretty good confidence if something might happen within four hours.'
Malani said that by using the model to direct emergency medicine and critical care resources, Sutter was able to reduce codes by 40% in trials.
'To predict hours in advance, the model has to absorb everything in the medical record and update it every five minutes,' Milani said. 'There's no human on earth that could make that computation.'
