Though many healthcare organizations have recently launched artificial intelligence chatbots to take on some of the burden of triaging potential COVID-19 cases, those chatbots are producing vastly different findings, Stat reports.
To test out several of these AI tools, healthy Stat reporters entered COVID-19 symptoms into eight chatbots. The solutions tested came from the CDC, Cleveland Clinic, Renton, Wash.-based Providence, Google sister company Verily, Buoy Health and more.
After entering the same symptoms into each, the chatbots' responses ranged from estimating a "low" risk of a COVID-19 infection to recommending immediate in-home isolation, according to Stat.
Furthermore, these conclusions were reached by very different paths: While some bots asked only a handful of questions, others asked more than 30; some asked for far more detail than others about specific symptoms; and the countries included in travel history questions varied widely.
Ultimately, as an AI researcher at the Boston-based Harvard T.H. Chan School of Public Health's epidemiology department told Stat, these AI chatbots show major promise in diverting some of the burden from already overworked clinicians to triage potential cases. However, until the technology is more standardized, it is unwise to rely too heavily on AI for triage in a crisis as widespread and fast-moving as the ongoing coronavirus pandemic.
"My sense is that we should err to the centralized expertise of public health experts instead of giving people 1,000 different messages they don't know what to do with," Andrew Beam, PhD, said. "I want to take this kind of technology and integrate it with traditional epidemiology and public health techniques."
Read more here.