WHO launches Sarah, an artificial intelligence avatar on health issues

More than half of the population consults health topics on the Internet, according to a survey by the Societies of Family and Community Medicine (semFYC), to find explanations for symptoms, discomforts, diagnoses, medications, vaccines, treatments and lifestyles, among other things. Technology companies, which already have artificial intelligence medical assistants, want to exploit this vein. But these services, more focused on professionals, leave information in the hands of free resources about which María del Campo, a member of semFYC, highlights the importance of "there being mechanisms for searching for refined medical information." The World Health Organization (WHO) has tried, with many shortcomings, according to early experiences, to do its part with an artificial intelligence health chat, Sarah, which it has just launched.

Sarah stands for

Smart AI Resource Assistant for Health

(Intelligent AI Resource Assistant for Health) and is an avatar that was already tested during the pandemic under another name (Florence), but that reemerges with a new language and technology model and, for now, in eight languages, including the Spanish.

Sarah answers very general questions with minimal empathy and always recommends going to the doctor. It is assumed capable of providing information on the main health topics, but in the tests carried out in these first steps, it has failed to provide links to more specific medical information and has limited itself to offering very general recommendations or a basic list. of symptoms linked to some ailments. She also can't view images.

But, despite the deficiencies in this launch, the WHO does not want to miss the train of artificial intelligence in its corresponding field or leave it to companies with economic and commercial interests. “The future of health is digital and supporting countries to harness the power of digital technologies for health is a priority,” Tedros Adhanom Ghebreyesus, director general of the WHO, explained in a note.

“Sarah gives us an idea of ​​how artificial intelligence could be used in the future to improve access to health information in a more interactive way,” he says, admitting the deficiencies of the current system. In this sense, the director of the WHO asks “the research community for help to continue exploring how this technology could reduce inequalities and help people access up-to-date and reliable health information.” Like any artificial intelligence system, Sarah aspires to grow with human interaction.

Del Campo, during the self-care week two years ago, highlighted that the high frequency of searches for health topics “indicate that there is a concern that we must respond to,” but he also warned of the importance of “accompanying the patient in information screening.”

Nor does the WHO avoid the difficulties of this type of tools to ensure equitable access, privacy, security, precision, the absence of bias and data protection.

“Developers, policymakers and healthcare providers must address these ethical and human rights issues when developing and implementing AI to ensure that all people can benefit from it,” says the WHO.

More information

Technology companies prepare the assault on primary care

The warning from the global organization refers to the assault by large technology companies on consultations to bring their conversational machines to the field of primary care, but always, they say, as a support tool, co-pilot for the doctor and possible solution to the collapse of care. . 81.4% of women and 72.3% of men go to these services at least once a year, according to the National Institute of Statistics.

IBM has Watson Health, a conversational robot available at any time, which collects basic information and alerts of possible changes that require additional attention.

In the professional field, Microsoft develops Azure Health Bot, a conversational system based on medical information, classification protocols and language models trained to understand clinical terminology.

Google has also entered this market with the family of models integrated into MedLM. Greg Corrado, head of Artificial Intelligence for Health at the multinational, highlights the radiographic image analysis tools and AMIE, an application “optimized for diagnostic reasoning and conversation that emulates interactions between the patient and the caregiver.”

The avatar launched by the WHO has been created by Soul Machines with the support of Rooftop and, although it warns that it cannot access images, it requires access to the microphone and camera "to improve the conversational experience." The organization ensures that “all data collected is anonymized and complies with current privacy practices and regulations.”

The WHO warns that “answers may not always be accurate because they are based on patterns and probabilities in the available data.” In this sense, the organization warns that it is not responsible for any conversation content created by generative AI nor does it “represent or understand the views or beliefs of the WHO.” The final warning to the user is decisive: “You understand and accept that you should not rely on the responses generated as the sole source of truthful or factual information, nor as a substitute for professional advice.”

You can follow

EL PAÍS Tecnología

on

Facebook

and

X

or sign up here to receive our

weekly newsletter

.