At a time when AI is all the rage, a new survey from GE Healthcare has highlighted a significant level of distrust and skepticism around its use in medical settings.
The Reimagining Better Health poll of 5,500 patients and patient advocates and 2,000 clinicians found that the majority of doctors believe that AI has the potential to transform healthcare. At the same time, many feel that the technology is not ready yet — and remains marred by roadblocks such as biases.
The findings come as a number of healthcare giants continue to look at and experiment with AI models, including generative technologies like ChatGPT and conversational AI, to improve patient experience and outcomes, automate tasks and enhance productivity.
AI is here but concerns remain
Today, whenever anyone talks about AI, they mention how the technology is revolutionizing patient care, be it via drug discovery or predicting an individual’s best treatment plan. In the GE survey, clinicians iterated similar benefits, with 61% saying the technology can help with decision-making, 54% saying it enables faster health interventions and 55% suggesting it can help improve operational efficiencies.
The possibilities are endless, but many remain concerned about the risks associated with the adoption of AI in the field. Specifically, 55% of survey respondents said AI technology is not yet ready for medical use and 58% implied that they do not trust AI data. For clinicians with more than 16 years of experience, the skepticism level was even higher, with 67% lacking trust in AI.
Clinicians indicated that the biggest reason for this distrust is the potential for algorithms to produce unfair or discriminatory outcomes due to various factors such as incomplete training data, flawed algorithms or inadequate evaluation processes. As many as 44% of the respondents said the technology is subject to built-in biases.
Secondly, clinician awareness on the technologies involved is not often up to the mark. The study found that only 55% of surveyed clinicians feel they get adequate training on how to use medical technology.
How to build confidence?
As GE Healthcare CTO Taha Kass-Hout points out, a thoughtful, data-driven approach — where efforts are made to ensure data quality and transparency — is the key to building confidence among clinicians who are on the fence about AI technology.
“We pay special attention to where data sets come from and the characteristics of the population sampled,” Kass-Hout told VentureBeat. “We also evaluate the algorithms that classify and organize data and look at the AI formulation itself and clinicians’ feedback when updating these algorithms.”
To get the ball rolling, the CTO said, companies should drive training/education programs where clinicians are guided on all things AI, starting from how it works to how it can augment their work.
“As an industry, we need to build clinician understanding of where and how to use it and when it can be trusted fully versus leaning on other tools and human expertise,” said Kass-Hout. “I refer to this as ‘breaking the black box of AI’ to help clinicians understand what is in the AI model.”
This includes what data it comprises — age, gender, lab results, remote monitoring, medical history, genetic variant or biomarker, lesion progression in subsequent images — so clinicians can better understand what is influencing the AI output.
“Transparency on what influences the model and how it can be adjusted with a consistent feedback loop over time is critical to building confidence in AI technology among clinicians,” he noted.
As healthcare systems around the world face extreme pressures, clinicians are burning out and considering leaving the industry. In fact, according to the World Health Organization, there could be a shortage of 10 million health workers by 2030, when 1.4 billion people will be 60 or more.
In such scenarios, AI-driven systems could come in and eliminate repetitive low-level tasks to help workers focus solely on patients’ care, said Kass-Hout.
“There are places where technology can help reduce administrative tasks, better allocate resources and reduce burnout,” he said.
GE HealthCare’s Command Center is a great example of this, he said. The platform is helping hospitals use real-time utilization data to better allocate resources. “Using AI technology, hospitals can redirect ambulatory services to bring patients to facilities with lower utilization — helping to reduce burnout,” Kass-Hout said.
In another example, Hyro, a company providing plug-and-play conversational AI assistants for the healthcare industry, is automating tasks like patient registration, routing, scheduling, IT helpdesk ticketing and prescription refills, which constitute roughly 60-70% of inbound calls and messages into health systems.
“While we are still in the early stages of seeing the true impact of these technologies, with appropriate human supervision, AI can help to reduce the burden of data query and analysis on clinicians so that they can be focused on what really matters: Improving patient outcomes,” Kass-Hout noted.
The post Lack of trust slowing down AI revolution in medical settings: GE Healthcare report appeared first on Venture Beat.