Alison Stribling has learned a lot about infectious diseases since switching to the Covid-19 response in the health department in Contra Costa County, near San Francisco. One of her discoveries: how important fax machines are to the response to the US pandemic.
Across the country, labs and healthcare providers are reporting new Covid-19 cases to local health departments. At Contra Costa Health Services, officers use the data to initiate contact tracking or, in certain cases, send additional help, such as in a nursing home or to an infected healthcare provider.
On a typical day in Contra Costa, only about half of these reports come in electronically; the rest, hundreds, flows in through the fax line, creating a Sisyphean reading list. “It can be a very long day, especially during peak pressures,” said Stribling, a public health programs specialist. “It’s that ‘I can never get on top of this’ feeling.”
Now Contra Costa’s first responders have a little high-tech help. Just before Thanksgiving, the department launched the Covid Fast Fax software, developed in hasty collaboration with researchers from Stanford University. It marks the most urgent new faxes using machine learning algorithms. When Stribling and others in the fax fellowship returned to work after the holidays, they had hundreds of faxes backlog to read – but at least they knew where to start. “That was great timing,” said Stribling.
Like many pandemic response topics in the US, the project highlights the cracking down on the country’s health system. It’s also another example of creative minds revamping it with hasty innovation, after skilled auto workers create face shields or homemade hand sanitizers. In 2020, such projects can be life-saving. Contra Costa employees at Stanford have now released their code and methodology for other researchers or health departments to use.
Contra Costa got his AI helper after Amit Kaushal, a Stanford professor and practicing physician working on integrating machine learning into healthcare, offered the department his skills this spring. Kaushal suggested collaborating on a grant he had received to stop the spread of the virus with a contact tracking app that uses Bluetooth signals. Officials got even more excited when he ditched the idea of an AI-enhanced fax line.
Health officials in Contra Costa struggled with more than just the number of incoming case reports. Faxes appear as PDFs on a server, not as piles of paper – this is the 21st century, after all. But it’s tricky to spot and assess a Covid-19 case at a glance. Cases can be reported on various forms, which are also used for other illnesses, are often hand-scribbled and not typed, and sometimes end up in a jumble of other messages or records. On an ordinary day, two public health specialists would be assigned to read and prioritize incoming faxes. “Few faxes are alike, and it takes a lot of attention to detail and training to know what you’re looking at,” said Stribling, who for a time led the team that handled incoming case data. “That can be difficult for eight hours or more.”
Kaushal and fellow Stanford researchers tried to tame the problem using machine-learning software that analyzes images – technology that medical researchers tend to target on tumors, not faxes.
To avoid the need to process sensitive medical data, Kaushal has recruited some fellow doctors to fill out disease reporting forms with randomly generated patient data in authentic doctor’s doodles. The fake forms were sent to a fax line to create authentic-looking sample data. Grad student Adam Lavertu used that data to train software to classify whether a page of an incoming fax contains a new Covid-19 case report or is something else, such as a medical record or tuberculosis report.