Join the event trusted by corporate leaders for almost two decades. VB Transforma connects people building AI Real Enterprise. Learn more
How would it be to talk to medical documentation as it could be with ChatgPT?
Initially posted by a medical student, this question caused the development of Chatehr in Stanford Health Care. Now, in the production, the tool accelerates the reviews of charts to accept in an ambulance, improves the summary of patients’ transfer and synthesizes information from elaborate medical stories.
In early pilot results, clinical users have significantly accelerated information; In particular, rescue doctors noticed 40% shortened chart review time during critical messages, Michael A. Pfeffer, SVP Stanford and Information and Digital Director, said today Fireside chat at the chat in Fireside na VB Transform.
This helps reduce burnout of doctors while improving patient care, and the decades of work that medical facilities perform to collect and automate critical data.
“This is such an exciting time in healthcare, because we spent the last 20 years of digitization of healthcare data and we put it in electronic health documentation, but not transforming it,” said Pfeffer in chat with the VB Matt Marshall editor. “Thanks to the new large language model of technology, we are actually starting to do this digital transformation.”
As Chatehr helps to shorten “pajamas” time, return to real face interaction
Doctors spend up to 60% of time for administrative tasks, not direct care for the patient. They often put in significant “Time of pajamas“Sacrifice Personal and family hours to perform administrative tasks outside regular working hours.
One of the biggest goals is to improve work flows and limit these additional hours so that clinicians and administrative staff can focus on more crucial work.
For example, a lot of information comes through patient portals online. AI now has the opportunity to read messages from patients and are looking for answers that a person can then view and confirm sending.
“It’s a bit like a starting point,” he explained. “Although it does not necessarily save time, which is interesting, it actually reduces cognitive burnout.” What’s more, he noticed, the messages are more patient’s warm because users can instrument the model to exploit a specific language.
Going to agents, Pfeffer said they are a “quite new” concept in healthcare, but they offer promising opportunities.
For example, patients with cancer diagnoses usually have a team of specialists who view their registers and determine the next stages of treatment. However, preparation is a lot of work; Clinics and employees must review the entire patient record, not only his EHR, but imaging pathologies, sometimes genomic data and information on clinical trials to which patients can adapt well. Pfeffer explained that they all have to meet so that the team can create a timeline and recommendations.
“The most important thing that we can do for our patients is to make sure they have adequate care, and requires a multidisciplinary approach,” said Pfeffer.
The goal is to build agents in Chatehr, which can generate a summary and schedule and give recommendations regarding the review of clinicians. Pfeffer emphasized that he does not replace, prepares “simply amazing recommendations in a multimodal summary.”
This allows medical teams for “actual patient care”, which is critical among a doctor and a shortage of nursing.
“These technologies will change the time that doctors and nurses issue administrative tasks,” he said. In combination with AI ambient scribes that take over the duties of notification, medical staff focus on patients.
“This direct interaction is simply priceless,” said Pfeffer. “We’ll see how AI is more translated to the interaction of clinicians-patient.”
“Amazing” technologies combined with a multidisciplinary team
Before Chatehr, the PFEFFER team introduced Securens to all Stanford medicine; A sheltered portal has 15 different models that everyone can tinker with. “What is really powerful in this technology is that you can really open it for so many people to experiment,” said Pfeffer.
Stanford adopts a diverse approach to the development of artificial intelligence, building its own models and uses a sheltered and private mix of ready (such as Microsoft Azure) and Open Source models in appropriate cases. Pfeffer explained that his team “is not completely specific” for one or the other, but rather with what will probably work best in the case of a specific case of exploit.
“There are so many amazing types of technology now that if you can break them down the right way, you can get solutions such as what we’ve built,” he said.
Another recognition for Stanford is the multidisciplinary team; Unlike the AI director or AI group, Pfeffer collected the main scientist of data, two IT specialists, medical information director and nursing information director as well as their CTO and CISO.
“We combine computer science, data and traditional IT and wrap it in architecture; you get this magical group that allows you to perform these very complex projects,” he said.
Ultimately, Stanford perceives artificial intelligence as a tool from which everyone should know how to exploit, emphasized Pfeffer. Various teams must understand how to exploit artificial intelligence so that when they meet with business owners and invent ways to solve the problems: “AI is simply part of how they think.”