| | | |  | Good morning — STAT's health tech team here with the details you need on Verily's leadership shakeup and more industry news. | | | Verily announces layoffs and a leadership shakeup Verily, in a bid to narrow its sprawling focus, is laying off 200 people and cutting several programs so it can “advance fewer initiatives with greater resources.” In a memo, CEO Stephen Gillett said the reorganization will help Verily become "the data and evidence backbone for precision health." The company also confirmed to STAT's Matthew Herper that co-founder Jessica Mega and another leader, former health system executive Vivian Lee, have left the company. Former FDA official Amy Abernethy, meanwhile, has been promoted to president of product development and chief medical officer. Read more on Verily's reorganization here. | The problem with a common FDA device pathway 
The FDA clears thousands of medical devices every year through what's known as the 510(k) pathway, which lets manufacturers show simply that a new product is "substantially equivalent" to one that's already on the market. But in a pair of new studies, researchers examined the pathway and found that devices approved based on prior products that had undergone a Class 1 recall — meaning they could cause serious health problems or death — were six times more likely to then be recalled themselves than similar devices.
“This regulatory loophole allows safety issues to propagate for years and years, placing patient safety at risk,” said Harvard's Kushal Kadakia. The FDA, for its part, said it "takes steps to prevent or discontinue the marketing of a device that relies on a recalled predicate where the root cause of the recall has not been adequately addressed in the new device." Read more.
| Funding, acquisitions, and legal drama - Covid-19 vaccine maker BioNTech announced plans to acquire the machine learning firm InstaDeep for about $440 million. BioNTech has partnered with the London-based company for the past three years. CEO Uğur Şahin told STAT'S Jonathan Wosen that AI could play a powerful role in drug development and manufacturing. "This is the work of tomorrow based on personalized medicine, where we do things which are complex with assistance from AI tools," he said.
- Apple lost an early round in its battle with Masimo, which alleges the tech giant has infringed on its pulse oximetry patents with certain models of the Watch. An International Trade Commision judge agreed earlier this week with Masimo's claims. Apple is still locked in a fight with AliveCor over its heart monitoring patents.
- Telesair, which makes respiratory care products for hospital-to-home treatment, raised $22 million in a Series A Round led by Pasaca Capital.
| In-depth analysis of biopharma and the life sciences Sign up for STAT+ to access in-depth analysis of biopharma, inside intelligence from Capitol Hill, the latest on medicine tech, and more. Subscribe today to start your free 30-day trial. | A commitment to combating bias in health tech In Arizona, the American Statistical Association is undertaking a campaign to build anti-racism into statistical models and machine learning tools. Casey reports that at its health policy conference this week, the organization held a special session on efforts to audit algorithms for bias, ensure more diversity in teams building AI and statistical models, and identify mathematical approaches that are particularly vulnerable to implicit bias. “We don’t anticipate a comfortable journey,” executive director Ron Wasserstein said, noting the structural and historical inequities within statistical applications in health care and other industries. “But we are committed to moving forward.”
The association has hired the Nova Collective, a diversity and inclusion consulting firm, to help implement recommendations surfaced by task forces. Wasserstein called the work the “most important thing” the organization is doing and emphasized the need to ensure follow-up and track progress over time. Casey closed the conference by delivering a joint keynote with University of California, Berkeley's Ziad Obermeyer. The session focused on the many different forms of bias in health care decision making, and how machine learning tools used in the care of conditions such as sepsis can exacerbate or counteract those biases. | What doctors in training don't learn about AI Despite the growing role of AI in health, medical school training on the subject has been haphazard at best. Faculty members often don’t have the expertise to incorporate new technology into classes, leaving many students to direct their own specialized study of machine learning — like Erkin Ötleş, who is getting his MD/PhD at the University of Michigan. Along with professors and educators at the school, he argues that needs to change, and fast.
“We're going to be at a point where we're not going to be able to catch up and be able to call out the technology defects or flaws,” he said. In a conversation with Katie, Ötleş and former Michigan school of medicine dean Jim Woolliscroft discuss the barriers to changing the status quo, and a proposed framework that could help make AI as central to training as pharmacology and physiology. | What to read around the web today - Microsoft + OpenAI: Inside tech's hottest romance, The Information
- Upstart Element ratchets up race for cheaper DNA sequencing with a $200 genome, STAT
- LeanTaaS, Hospital IQ join forces to address hospital operational optimization, Fierce Healthcare
| | | |
No comments