| | | |  By Casey Ross, Katie Palmer, Mario Aguilar, and Mohana Ravindranath | Good morning from Mario and Mohana! The health tech world's abuzz about the end of Amazon Care. We've got more on that below. | | | Amazon Care to shutter by year's end The retail giant shocked the health and business worlds this week when it revealed plans to wind down telehelath venture Amazon Care by the end of the year — less than a month after it announced its intent to buy primary care tech company One Medical. In a memo, Amazon Health Services' head Neil Lindsay told staff the company's efforts to build a national primary care service for employers “was not a complete enough offering for the large enterprise customers we have been targeting, and wasn’t going to work long-term.
While Amazon Care initially began in 2019 as a modest service targeting Amazon employees, it had expanded to provide virtual services to employers across the country. The nationwide rollout — along with Amazon’s Crossover Health collaboration and its joint venture, Haven — signaled Amazon’s big ambitions in primary care. But Amazon Care’s demise shows the company is not immune to the challenges that have hamstrung other tech giants as they move into medicine. | Pregnancy research has a big data problem New data sharing networks that cropped up during the pandemic could demystify issues surrounding pregnancy, which is woefully understudied. Covid-19 supercharged efforts to make health records more interoperable as government agencies, health systems and data companies open up their records while preserving patient privacy to share sensitive medical data for research. That could net researchers valuable, real-world data about pregnancy even beyond the pandemic, Katie reports. “The tolerance for risk is so low, and for that reason you do see very few studies — and when you do, it’s voluntary and at your own risk,” said Jose Figueroa, a physician at Brigham and Women’s Hospital who researches health policy and management. “So real world data is super important.” But the data’s still very messy, and researchers are scrambling to untangle it.
“You can have the best questions in the world, but if you don't have a good data set to answer those questions, there's only so much you can do,” said Figueroa. “The trick is, are we able to develop tools and technology to fill in and make the data better?” Katie has the full story. | wearables to tell you you’re stressed Consumer wearables can track your activity and your sleep and detect heart conditions. Now companies want you to use them to track your mood and stress levels. Fitbit yesterday announced its new lineup of wearable trackers, including the Sense 2, which has a new sensor that continuously monitors your electrodermal activity to detect when you might be feeling stressed. The Fitbit app combines this data along with information about your activity and sleep to calculate a stress management score, so users can follow trends over time. On the far-less-established end of the spectrum: A new company called, I’m not kidding, Happy Health, also yesterday announced it had raised $60 million to launch a Happy Ring, which uses similar electrodermal sensing. It’s basically an Oura ring, but for your mood. (Founder Sean Rad also created the dating app Tinder.) These aren’t the first efforts to make use of wearables to monitor mood — Amazon’s Halo wristbands, for example, can monitor tone of voice, which itself is a methodology being tackled by many other companies. But to be more than marketing gimmicks, all of these experiments must show that such mood tracking is reliable and that surfacing the information to users has utility besides, well, stressing people out. | RCTs with smaller control arms help more patients receive experimental treatments for HD and ALS Our latest case study shows how TwinRCTs reduce control arm sizes by up to 22% in Phase 3 trials in Huntington’s Disease and Amyotrophic Lateral Sclerosis. Smaller control arms provide trial patients with a greater opportunity to receive the experimental treatment. Running smaller trials also accelerates timelines and helps to bring new therapies to patients sooner. Download the case study. | A guide to minimizing bias in radiology AI There are about a dozen ways researchers can try to steer machine learning systems away from bias in radiology from the get-go. In the first of three-part series in the Radiological Society of North America’s journal, researchers sounded the alarm about missteps in thee early stages of model development, including data collection and exploratory analysis, that could predispose AI systems to systematically miss certain signals. Machine learning teams should also have members with data science and clinical expertise, said radiologist Bradley Erickson, director of Mayo Clinic’s AI Lab. “If these systematic biases are unrecognized or not accurately quantified, suboptimal results will ensue, limiting the application of AI to real-world scenarios,” he said.
Faulty image processing could also lead to bias. If a deep learning system necessarily crops a lung image and leaves out an important feature — denoted by the red arrow — and then feeds it to a subsequent classifier, that classifier will miss it, the Radiological Society of North America warned. | Funding and other industry news - Digital Diagnostics raised $75 million in Series B funding to advance its artificial intelligence tools. The round was led by KKR.
-
Bicycle Health, which provides virtual treatment for opioid use disorder, has been added to Evernorth’s behavioral health network, making Bicycle’s services available to the Evernorth clients as well as Cigna customers. Bicycle recently raised $50 million to expand its services. - Fair Square Medicare, which is developing a service to help seniors navigate Medicare, announced a $15 million Series A round led by Define Ventures with participation from Slow Ventures, and YCombinator.
- Folx Health, which provides health care services tailored to the queer and trans communities, announced it would be offering consultations around monkeypox through its burgeoning virtual care platform.
| Personnel file -
Therapeutic chatbot developer Woebot Health hired Robbert Zusterzeel to head up regulatory science and strategy. He most recently served a one-year stint at health data giant IQVIA. Before that he spent nearly nine years at the Food and Drug Administration. -
Amazon head of executive development Patty Bedard jumped ship to health care workforce marketplace platform CareRev, where she’ll serve as chief people officer. -
Johnson & Johnson has named former CVS president and CEO Larry Merlo a non-executive chair designate for its planned consumer health company. Johnson & Johnson said last year it was separating the consumer line out from the rest of its health care R&D business. | | | Thanks for reading! More next week,  | | | |
No comments