AI to restore power of human connection to medicine

eHealthNews.nz: AI & Analytics

AI to restore power of human connection to medicine

Tuesday, 21 March 2023

Share |

NEWS - eHealthNews.nz editor Rebecca McBeth

Use of AI tools can restore humanity in medicine by liberating clinicians from their computer keyboards, world renowned cardiologist and digital health researcher Eric Topol told an Auckland conference.

Topol, founder and director of the Scripps Research Translational Institute and professor at Auckland University’s Faculty of Medical and Health Sciences, presented at the Hack Aotearoa Conference in Auckland on March 17.

He said the AI landscape is exploding and the rapid uptake of AI tools such as ChatGPT are “like nothing we have seen before in history”.

Topol said AI tools can be used to generate clinical documentation in real time through voice recognition, and can make recommendations or request basic tasks such as lab tests, as well as nudge patients.

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month.

“This gives doctors time to think. We have to restore the power of touch and human connection, which is the fundamental deficiency in medicine today,” he told the Auckland audience.

“We can give back the gift of time to restore the patient doctor relationship and that will happen if we use these tools in the best way possible.”

Topol said everyone keeps talking about precision medicine, but what we need is accuracy in medicine and that is where AI can make a difference.

He said a US study of 11 hospitals in Boston found that 37 percent of admissions had medical errors or serious adverse events that could have been prevented, “so the first thing for AI to do is improve accuracy”.

Topol said deep neural networks can be trained to detect things that human eyes cannot see and while studies show that clinicians are more likely to miss things on scans in the afternoons, machines never get tired.

He discussed some of the extraordinary things that AI can do, such as tell the age and sex of a patient from an ECG reading and not only detect cancer, but tell where it is originating from in the body.

Topol also briefly described the many innovations in AI that enable virtual care, such as smartphone ultrasound technology that allows self-imaging in people’s homes.

“This is an exciting time: to think that in the future, unless a person is so critically ill that need to be admitted to an ICU or go to an Emergency Department, that person should be at home in their bedroom… which is far less expensive than hospital care,” he said.

Leo Celi, principal research scientist, Massachusetts Institute of Technology, also spoke at Hack Aotearoa and said we have to address problems of bias in the data before leveraging AI tools such as Large Language Models, as they are often trained on data that is not representative of the populations of most of the world.

He said these models cannot assess the quality of information or bias in the data, so “you can’t just take studies at face value, put them in a machine and expect accurate answers”.

“Health equity should be the focus of health AI,” Celi said.

Picture: Eric Topol, founder and director of the Scripps Research Translational Institute, presenting at Hack Aotearoa