Inside the failed Bellerophon Therapeutics trial

Biotech company Bellerophon Therapeutics is abandoning a study on a pulmonary fibrosis treatment after a study using an innovative digital endpoint — a measurable outcome indicating whether a therapy works — showed no improvement for patients.

This was the first major trial to use an FDA-endorsed endpoint incorporating data from wearable devices, according to Bellerophon. Other biotech and pharma companies were closely watching the trial as they consider new tech in their own clinical trials, Mario Aguilar writes.

In a randomized clinical trial, patients who got the treatment — nitric oxide, inhaled through a cannula — didn't show improvement compared to a placebo group when it came to the primary goal of moderate to vigorous physical activity measured by an ActiGraph wearable. (They actually fared worse, showing a reduction in minutes of physical activity compared to placebo.)

Read more on the implications for digital biomarkers in clinical trials, which proponents have argued could speed up drug development and cut costs.

Artificial intelligence

Carbon's already using AI for patients notes

Last week, I visited a Carbon Health clinic for a look at their new souped-up health record software, which records and types up patients' visits using AWS Medical Transcribe and succinctly summarizes it using GPT-4. (For the purposes of a demo, I pretended to be a 44-year-old patient named Daisy Duck and rattled off some semi-fictional medical concerns — and I admit I was impressed at how quickly, accurately, and concisely the system summarized them in an electronic record.)

Just a few months after it began developing the system, Carbon's already trotting it out to hundreds of doctors and their patients, and it's already in talks to sell it to outside health systems. Caesar Djavaherian, Carbon's chief clinical innovation officer, told me Carbon's providers say the tech frees doctors from their administrative and documentation burdens, leaving more time and energy for deeply connecting with patients.

But I still have questions: How often do these systems introduce medical errors, which doctors might be inclined to overlook especially if they're infrequent? What details might the tech deem unimportant and omit from the summary, only for doctors to discover were medically relevant later? If you have thoughts, I welcome them! And read more on the tech here.

Medical devices

A wearable "vest" to treat lung cancer

A device made by Novocure that creates electrical fields in lungs through wearable skin patches was found to extend lung cancer patients' survival in a clinical trial, my colleague Adam Feuerstein reports from Chicago today.

The research findings could lead to an approval for the device beyond its FDA marketing clearance to treat a type of brain cancer.

While the study demonstrated an impact on patients with lung cancer that progressed after initial chemotherapy, Novocure and the device, called Optune, face challenges: Most study participants didn't get initial treatment with an immune checkpoint inhibitor such as Keytruda, Adam writes.

"It's quite impressive that it's a positive trial. I thought it would be negative. But with that said, these patients don't exist in my practice in 2023," said NYU cancer specialist Joshua Sabari. Read more.

Apple

#WWDC23 tidbits: Mental health tracking, daylight sensors

Apple played up a couple health-related features — though none particularly groundbreaking — at its hotly anticipated developer conference yesterday. Among them: An ambient light sensor on the Apple Watch that can sense the amount of time spent in daylight, a bid to prevent nearsightedness, and a mental health log and tracking feature that lets users "see how your state of mind may correlate with lifestyle factors like time spent in daylight, sleep, exercise, and mindful minutes," the site says. Still, Apple Watchers agree that health took a backseat at yesterday's event, especially compared to previous years marked by bold new features like heart monitoring.

Social determinants

What it costs to address patients' social needs

Waymark — an a16z, NEA and Lux-backed public benefit company that uses tech to coordinate community-based health care — led a new study estimating the cost of social needs interventions detected during patients' primary care visits. Lead author Sanjay Basu, Waymark's co-founder and clinical head, told me the study starts to shed light on the back-end problems that make it harder to address those social needs.

In analyzing 3 years of National Center for Health Statistics data, they concluded that the average cost of food, housing, transportation, and care coordination assistance was about $60 per member per month, less than half of which was covered through federal programs. Screening patients for social needs cost about $5 a month.

"For us on the health care end, we're screening and referring a lot of people to resources that don't exist — it's the bridge to nowhere problem," Basu said.

What does that mean for tech companies trying to tackle this issue? Basu said tech companies doing screening for social services, for instance, aren't focused on the real problem. "Some of the challenges are actually getting proper data on who's getting funded, to what extent, and what's the proper supply to demand ratio," he said.

More around STAT

Check out more exclusive coverage with a STAT+ subscription

Read premium in-depth biotech, pharma, policy, and life science coverage and analysis with all of our STAT+ articles.

What we're reading

More than 400 Grail patients were incorrectly told they may have cancer, Financial Times
Black men were likely underdiagnosed with lung cancer because of biased software, Associated Press
AI chatbots lose money every time we use them, The Washington Post

Thanks for reading! More on Thursday - Mohana

Mohana Ravindranath is a Bay Area correspondent covering health tech at STAT and has made it her mission to separate out hype from reality in health care.