Executive Roundtable: AI and Covid-19 | Straight Talk

SUBSCRIBE NEWSLETTER

The latest insights from your peers on the latest in Enterprise IT, straight to your inbox.

A Stanford University (virtual) conference offers perspectives on how technology can be harnessed to combat the pandemic.

By Pragati Verma, Contributing Editor, Straight Talk

Can artificial intelligence and data analytics help protect us from the coronavirus and other pandemics? Undoubtedly, according to AI and data scientists presenting at a virtual conference hosted by the Stanford Institute for Human-Centered Artificial Intelligence.  They outlined ways that technology can help monitor and fight pandemics, from tracking infections to stimulating the drug discovery and development process.

Tech to Flatten the Curve

A big part of the challenge is how the pandemic has strained the global healthcare system beyond its capacity. Xavier Amatriain, CTO and co-founder of startup Curai, described one of his company’s solutions: AI-driven assisted conversations to help patients describe symptoms, prompt answers doctors need to make an accurate diagnosis, and assist patients in understanding their condition and treatment. A covid-aware machine learning diagnostic model can answer people’s specific questions about their symptoms and condition via an online chat.

The aim is to help patients get care without exposing them to infectious diseases. “We only want people to go to hospitals if they are in a critical situation,” he said. The tool “will reduce the load on clinics” and thus help flatten the curve. Curai hopes to scale up its end-to-end medical platform to provide virtual primary care for everyone.

Pressures on healthcare systems are not new, according to Amatriain. “There were baked-in inefficiencies in healthcare even before Covid-19, but [the pandemic] has exposed the shortcomings,” he said. AI can help overcome these, but doesn’t provide a quick-and-easy solution. “AI can’t be dropped in the middle of old workflows and approaches,” he warned. “We need to integrate it in an end-to-end medical care system benefiting both patients and providers.”

Hunting the Virus with Data

AI experts are not alone. Data scientists are stepping up, too.  Lucy Li of the Chan Zuckerberg Biohub says her organization is developing a tool to estimate unreported infections. “The most obvious solution is to do mass testing of the population, but it can be very expensive and time-consuming and may not be doable when resources are already constrained,” she said.  

That’s why researchers use mathematical models to estimate the number of infections, based on when and where the infections occur. “However, there is a lot of uncertainty in modeling results, especially at the start of a pandemic,” she said. So Li and her team turned to viral genomic data — the number of virus mutations among identified cases. “If we know how quickly the virus mutates, we can infer how many missing transmissions could have occurred between the testing cases,” she explained, “We model the time-series of case counts at the start of the epidemic and the biogenomic diversity to quantify the number of undetected infections.”

At Harvard Medical School, pediatrician John Brownstein and his team have crowdsourced large volumes of data to create an online map, called Healthmap, to track the Covid-19 outbreak in real-time. “Because the data was heavily curated and publicly available, it has generated a whole set of publications aimed at forecasting and understanding the risks in different populations, especially in understanding the efficacy of different interventions,” Brownstein said. His team is also working closely with the CDC to build tools to understand the impact of social distancing policies.

Meanwhile, Ryan Tibshirani, associate professor of statistics and machine learning at Carnegie Mellon University, is building an epidemiological forecasting tool. Having modeled and forecast seasonal flu spread for the CDC since 2013, he said that models and tools used to predict influenza rates might not work “as is” for Covid-19. But he and his team are adapting their influenza-forecasting model to predict Covid-19 spread and help public health officials make policy decisions.

Time for “Open Science”

Researchers across disciplines underscored the need for “open science,” to eliminate barriers to sharing data and tools that could help track and better understand the disease. Mark Musen, Stanford professor of medicine and biomedical data science, noted that “[t]here’s a lot of data on the virus already, but it’s like a library with no good catalog.”

Good data is absolutely critical, agreed Nigam Shah, Associate Professor of Medicine (Biomedical Informatics) and of Biomedical Data Science at Stanford. “Without good quality data, all of these models will give exponentially bad outputs.” He suggested a new framework for sharing healthcare data during pandemics like Covid-19. “If a governor is allowed to legally command Sheraton to share its hotel beds as hospitals,” he said, “I believe the governor should be allowed to command a tech company to share their tracking infrastructure for public health for a limited period of time.”

The proposal to share data and tracking technology to slow the pandemic, however, raised a few concerns about government outreach. Michele Barry, Stanford University professor of medicine, worried that governments might not want to unwind the surveillance laws even after the emergency has passed. “It’s easy to roll out public health laws, but sometimes, especially in authoritarian countries, it is hard to unroll laws made in the times of public health emergencies,” she noted.

If there was one thing that tech as well as healthcare researchers and practitioners agreed on, it was that AI and data analytics projects to beat Covid-19 have accelerated data sharing and cooperation, despite privacy and security concerns. “There are real issues of privacy surrounding sharing of medical data,” Tibshirani, of Carnegie Mellon, acknowledged. “That being said, what we’ve seen in our group — and I think this is true across, probably, anybody who’s working on COVID modeling and forecasting — is that there’s been more movement and more accomplished in the last two weeks than I’ve seen in years in terms of data sharing.”

Pragati Verma is a writer and editor exploring new and emerging technologies. Previously, she was a business journalist and managed technology sections at India’s The Economic Times and The Financial Express.