CareCredit - Women's Health April 2024

Google Claims Healthcare AI More Empathetic, Accurate Than Real-Life Doctors

At the heart of healthcare workflows sits the physician.

And at the center of care delivery lies the physician-patient dialogue, the initial engagement point where diagnosis and care management begin.

At least, that is how healthcare and medicine have traditionally operated. But with the advent of generative artificial intelligence (AI), new opportunities for care delivery and streamlined provider-patient engagements could be around the corner.

A new research paper from Google scientists titled “Towards Conversational Diagnostic AI” claims that a healthcare-specific AI system trained to conduct medical interviews was able to match and in some cases, surpass, the performance of human doctors when conversing with simulated patients as well as when listing possible diagnoses on the basis of the patients’ medical history.

The study included 149 case scenarios from clinical providers in Canada, the U.K. and India. Per the paper, the AI chatbot system, named the Articulate Medical Intelligence Explorer (AMIE), demonstrated “greater diagnostic accuracy and superior performance on 28 of 32 axes according to specialist physicians.”

“While further research is required before AMIE could be translated to real-world settings, the results represent a milestone towards conversational diagnostic AI,” said the researchers, noting that the exercise had “several limitations and should be interpreted with appropriate caution.”

The research paper has not yet been peer reviewed, and the AMIE chatbot was tested on actors trained to portray people with medical conditions for the paper, not on people with real health problems.

But the AMIE chatbot, based on a large language model (LLM) developed by Google — which has been focused on deploying AI solutions within the healthcare sector — was more accurate than the physicians in diagnosing, among other ailments, both respiratory and cardiovascular conditions. Compared to the doctors, the AI system also scored higher on empathy.

Read more: How Gen AI Helped Healthcare Get Smarter 

Modernizing the Physician-Patient Dialogue

Google’s healthcare AI ranked higher than human physicians across 24 of 26 conversational axes, according to the patient actors in the study, who reported that AMIE outperformed across areas like politeness, coming across as honest, explaining the condition and treatment, and expressing care and commitment.

Of course, the study was performed using text-based chat interfaces — an area where AI LLMs shine, and a domain where human physicians may not have as much experience interacting with patients.

As PYMNTS Intelligence found, the generative AI healthcare market is projected to reach $22 billion by 2032, offering a number of possibilities for better patient care, diagnosis accuracy and treatment outcomes.

Still, despite the potential benefits of AI in healthcare, PYMNTS Intelligence also found that over half of adults (60%) remain uncomfortable with the idea of AI-driven healthcare decisions. Surveyed concerns range from biases in AI algorithms to fears that AI may lead to worse outcomes.

For now, at least, the human touch — and a history of real-world experience — is one of the most important qualities care providers can offer.

“The notion that AI would completely replace a physician or another healthcare provider, to me, is a long way off when I look at the scenarios,” Tom O’Neil, managing director at Berkeley Research Group and former chief compliance officer at Cigna, told PYMNTS.

And as Erik Duhaime, co-founder and CEO of data annotation provider Centaur Labs, told PYMNTS, “AI for healthcare has never been about replacing doctors, but doctors who use AI might end up replacing those physicians who don’t.”

Read moreNew Healthcare AI Models Highlight Role of Specially Trained Point Solutions

Democratizing Access to Healthcare

Still, an AI system capable of diagnostic dialogue is not an advance that should be written off lightly.

AI’s capacity to increase the accessibility, consistency, and quality of care delivery are among the most promising use cases for the technology.

And they are eminently scalable.

“Worldwide, very few people have access to doctors — and the opportunity to have an AI doctor, even if they have just 30%, 50% of an average provider’s knowledge and capability, is still a massive value add,” Beerud Sheth, CEO at conversational AI platform Gupshup, told PYMNTS in an interview posted in November.

As PYMNTS’ Karen Webster wrote, AI’s greatest potential is in creating the knowledge base needed to equip the workforce — any worker in any industry — with the tools to deliver a consistent, high-quality level of service, quickly and at scale – and this is particularly true within healthcare.

After all, the healthcare system encompasses a vast array of symptoms and ailments, many of which can be diagnosed and treated without the need for a specialist. By applying AI at critical care delivery junctures, healthcare systems can reserve valuable provider staffing resources for more urgent and strategic cases.