ARPA-H plans to start combining new fetal monitoring signals in June 2027

Kate Arnold said the childbirth monitoring program will move from finding better signals to figuring out how to present them in one result clinicians can use.

Author
HDM Staff
Save
Share
Illuminated HLTH lettering and event entrance arch
Event branding at the entrance to HLTH USA 2026 in Las Vegas. Photo credit: HLTH USA 2026

    The Advanced Research Projects Agency for Health plans to begin testing how to combine new fetal monitoring signals into a single result for delivery teams in June 2027, program manager Kate Arnold said Wednesday. The planned transition is a milestone for a four-year effort to give clinicians clearer information about a fetus during labor.

    The work is part of Making Obstetrics Care Smart, or MOCS. Arnold said ARPA-H is directing most of its funding toward programs developed by its managers, giving teams a defined clinical problem to address and milestones for establishing whether the approach can work.

    Arnold spoke with host Ian Khan during The Futurist Physical AI Summit on the Monaco Stage at HLTH USA 2026 in Las Vegas. The work she described is meant to change how medicine is practiced and make a difference for patients over roughly two to five years. She gave no dollar figure for the share going to program-led efforts.

    For organizations interested in federal support for medical robotics or another form of physical AI, the distinction affects when to approach the agency and how to read a call for proposals. A program begins with an objective and a set of gaps the manager thinks can be addressed. Teams interested in joining need to see where their work fits that objective and what evidence they could deliver during the program.

    A program begins with a problem someone will own

    Arnold described ARPA-H hiring program managers to build programs around the ideas they bring. She said the aim is to change practice on a relatively short horizon. The time frame makes it important to define what progress would look like before a team commits to building a technology.

    Arnold said “we’re putting most of our funding now towards these program efforts.” The agency’s description of the role gives program managers responsibility for developing programs, evaluating offers to do the work and building partnerships that can help translate the results into use. That structure makes the proposed outcome, timeline and milestones part of the decision about what to fund.

    Arnold’s route to the agency began in clinical practice. The agency biography identifies her as an obstetrician-gynecologist and describes her leadership at a federally qualified health center in Oklahoma. She said the experience left her looking for a way to have a wider effect than she could reach through one clinical role. Her account gave the audience an example of how a problem seen in care can become the starting point for an agency program.

    Fetal monitoring shows how a clinical problem becomes a program

    Arnold said she was working as a hospital-based obstetrician when ARPA-H asked her what problem in healthcare needed a better technology. She described repeatedly reviewing fetal heart rate patterns that were hard to interpret with certainty and then discussing that uncertainty with patients. That recurring decision in the delivery room gave her a specific place to start.

    MOCS seeks better information during labor and delivery. In its June award announcement, ARPA-H described a four-year, $90.7 million program, with up to $34 million in its first year. Arnold rounded the commitment to about $90 million in her remarks. The agency’s published figure describes the overall program, including research yet to be done. Arnold did not present patient outcome results during this part of the discussion.

    The program’s premise is that a clinician needs information that helps distinguish a fetus in danger from one whose monitoring pattern is hard to interpret but may be reassuring. That distinction affects decisions teams make while labor is in progress. ARPA-H’s announcement described multiple teams pursuing different ways to measure fetal status. The common question is whether the information will improve a decision at the bedside.

    Arnold described AI as integral to the fetal monitoring work. The starting point was the clinical information a team lacked. The program still has to establish which signals can be measured reliably, how a result would be shown to staff and whether it can be interpreted in time to support a decision. Those questions help explain why the program spans measurement, analysis and how clinicians use the result.

    Arnold said the first year is aimed at determining whether teams can obtain new signals. The integration work would start in June 2027, the program’s second year, as teams look at using AI and machine learning to turn the new information into what she called a “single output for the care provider.” She said an earlier approach to improving monitoring had offered additional information without integrating it for the clinician. A team in a delivery room would then have to work out what another signal means while also caring for the patient.

    Where robotics and AI fit

    Khan asked whether the phrase physical AI covered robots, sensors or other devices. Arnold described a broad range of work across the agency. She distinguished programs in which AI is central to the problem being studied from programs that use it as a component. The fetal monitoring project illustrates the second case, where software has to work with a way to make useful measurements during care.

    For a startup seeking support, that distinction changes the proposal. A sensor company may have a way to collect better information. A software team may have a way to interpret it. A credible program proposal needs to show what decision the information would improve and what evidence would help establish that. It also needs to explain where other teams or institutions would have to contribute expertise.

    The timelines Arnold described leave limited room for projects that treat the eventual care setting as a question to resolve late. For a technology intended for use during labor, evaluations have to account for who uses it, when they need the information and what other parts of care are happening at the same time. The device and its software can be assessed in relation to those needs as the teams work toward clinical evidence.

    Arnold said the agency thinks about what happens after federal support ends when it evaluates a program. She described wanting a path to continued investment so a technology can keep moving toward the bedside. In practical terms, a team needs to show what the program can establish and what a hospital, manufacturer or another funder would still need before taking on the next phase. Arnold also said she is expected to pitch at least one program each year, making conversations with scientists another way for her to identify problems that could become future programs.

    Arnold’s example ties a startup’s proposed contribution to a specific uncertainty or delay in care. A proposal can then show what the technology would establish and how its evidence would help solve the problem the program manager has committed to address.