AI flags people who may develop complications if COVID-19 attacks
Israel’s biotechnology major, Maccabi Healthcare Services, has joined hands with AI developer Medial EarlySign to come up with an algorithm that could identify people who are more likely to contract severe complications if infected by the Coronavirus. The system is country-specific and currently works on 2.4 million members it covers within Israel – of which around 40,000 has already been flagged as high-risk.
Once identified, such people can be prioritized for Coronavirus testing. The AI tool can also assist in formulating a suitable treatment strategy based on risk evaluation — home isolation, external quarantine detention, or hospitalization. It could also prove helpful in isolating high-risk people once lockdowns are lifted, thus minimizing their chances of coming in contact with asymptotic friends or relatives who may unknowingly infect them.
The tool has been developed by refining an existing AI system that could diagnose people most susceptible to flu infections using an extensive 27-year database from Maccabi records. The new AI system draws on a range of medical data like age, BMI, pre-existing health conditions, and past records of hospitalization.
Maccabi is exploring avenues to collaborate with major US healthcare firms. But,although AI-based early detection of vulnerability can save many lives, there are complications in using this tool in a different country. The algorithm can only work when composite data sets are available from an entire population which, in this case, means exhaustive healthcare records from all or most citizens of a country – the reason why the tool is country-specific. Experts say, the biggest challenge is to craft a single data by collating dispersed patient information across all hospitals and healthcare providers. For example, in the US, such medical records are locked up inside many different healthcare systems.
However, COVID-19 will possibly coax decision-makers to introduce new guidelines to allow data transfer between different hospitals – thus facilitating centralized monitoring. As more patient data becomes accessible, diagnostic AI tools would be more accurate, and easy to implement too.
