Alison Stribling has has learned a lot about infectious diseases since his transfer Covid-19 response to the Contra Costa County Health Department, near San Francisco. One of his findings: How critical fax machines are to the US pandemic response.
Across the country, laboratories and health providers are reporting new cases of Covid-19 to local health departments. At Contra Costa Health Services, managers use the data to get started contact follow-up or send additional help in some cases, such as to a nursing home or to an infected health worker.
On a typical day in Contra Costa, only about half of these reports arrive electronically; the rest, up to the hundreds, poured in over the fax line, creating a Sisyphean playlist. “The day can be very long, especially during power surges,” says Stribling, a public health program specialist. “It’s that feeling of ‘I can never get over this.’ ‘
Now the first responders to Contra Costa’s fax machine have a little high-tech help. Just before Thanksgiving, the department launched software called Covid Fast Fax, developed in hasty collaboration with researchers at Stanford University. It marks the most urgent new faxes using machine learning algorithms. When Stribling and other members of the fax fraternity returned to work after the holidays, they had a backlog of hundreds of faxes to read – but at least knew where to start. “It was a great time,” says Stribling.
Like many other responses to the American pandemic, the project highlights the weakness of the country’s health system. It’s also another example of creative minds fixing it with hasty innovation, after skilled auto workers make face shields, or homemade hand sanitizer. In 2020, such projects can save lives. Contra Costa employees at Stanford have now published their code and methodology for other researchers or health departments.
Contra Costa got his AI help after Amit Kaushal, a Stanford professor and a practicing physician who works on integrating machine learning into healthcare, offered the department his skills this spring. Kaushal suggested collaborating on a grant he received to fight the spread of the virus with a contact tracing app using Bluetooth signals. Managers became more enthusiastic when he threw out the idea of an AI-enhanced fax line.
Contra Costa health officials were grappling with more than just the volume of incoming case reports. Faxes appear as PDF files on a server, not piles of paper – it’s the 21st century after all. But it’s difficult to spot and assess a Covid-19 case at a glance. Cases can be reported on several different forms, which are also used for other illnesses, are often hand-scribbled untyped, and sometimes arrive in a jumble of other messages or records. On a typical day, two public health specialists would be responsible for reading and prioritizing incoming faxes. “Very few faxes are the same, and it takes a lot of attention to detail and training to know what you’re looking at,” says Stribling, who for a time led the inbound case data processing team. “It can be difficult to do for eight hours or more.”
Kaushal and his fellow researchers at Stanford sought to tame the problem using machine learning software that analyzes images, a technology more commonly targeted by medical researchers. tumors, no faxes.
To avoid tampering with sensitive medical data, Kaushal recruited fellow doctors to fill out disease report forms with randomly generated patient data in authentically doctor scribbles. The fake forms were sent to a fax line to create authentic data samples. Graduate student Adam Lavertu used this data to train software to categorize whether a page of an incoming fax contains a new Covid-19 case report or something else, such as a medical record or a tuberculosis report.