AI is changing medical technology careers less by replacing clinicians than by creating hybrid roles that sit between the bedside, the data, and the device. Clinical informaticists, nursing informatics specialists, biomedical engineers, and healthcare data analysts now configure, validate, and monitor the algorithms clinicians rely on. The FDA has authorized more than 1,600 AI-enabled medical devices as of September 2026, and each one needs people who understand both the technology and the clinical setting it lands in.
Labor and delivery (L&D) is one of the clearest examples. The CDC’s National Center for Health Statistics counted 3,612,012 US births in 2025, and L&D units generate a continuous stream of fetal heart rate data, nursing documentation, and alarms that must be interpreted in real time. Because errors in that data can have lifelong consequences, obstetric safety has become a proving ground for clinical AI and for the professionals who manage it.
What is actually changing in medical technology careers?
Demand is shifting toward people who can work with clinical data, not just store it. The Bureau of Labor Statistics projects employment of health information technologists and medical registrars to grow 16% from 2025 to 2035, against 3% for all occupations, with a May 2025 median wage of $68,020.
The work itself is changing too. Hospital IT work in obstetrics once centered on keeping monitors connected and the EHR running. Today it increasingly means governing algorithms: validating outputs against local patient populations, tuning alert thresholds, monitoring for performance drift, and protecting the integrity of the record. Employers screening for these roles look for the digital skills healthcare employers are prioritizing, from analytics to cybersecurity awareness, layered on top of clinical fluency.
Why is fetal monitoring a test case for clinical AI?
The interpretation problem
Electronic fetal monitoring (EFM) records the fetal heart rate alongside uterine contractions. In 2008, the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), ACOG, and the Society for Maternal-Fetal Medicine published a workshop report standardizing EFM definitions and a three-tier interpretation system. Category I tracings are normal, Category II are indeterminate, and Category III are abnormal and call for prompt evaluation.
That framework gave clinicians a shared vocabulary. It also gave software developers something they could encode. Pattern-recognition tools can now measure baseline rate, variability, accelerations, and decelerations continuously, and FDA-cleared fetal heart rate analysis software is commercially available in the US.
Where AI fits in the workflow
Central fetal monitoring surveillance systems let one nurse or charge nurse view multiple patients’ tracings from a single station. AI-assisted early-warning layers sit on top of those systems and flag when fetal heart rate patterns, maternal vital signs, or labor progress cross limits the hospital has defined.
Consider a hypothetical night-shift L&D unit where a central surveillance system flags a Category III tracing on a patient whose nurse is managing two other labors. The alert does not make a clinical decision. Its value depends on whether it reaches the right person, on the right device, with enough context to act. That routing logic is built and maintained by informatics and IT staff, not by the algorithm vendor alone.
The systems lesson
Monitoring problems are often systems problems. Materials from the AHRQ Safety Program for Perinatal Care note that 34% of perinatal death or permanent disability cases reported to The Joint Commission between 1996 and 2004 cited inadequate fetal monitoring as a root cause, with communication problems among the most frequently identified issues. Those figures come from root cause analyses of 47 reported sentinel events, not from a measure of how birth injuries generally occur. Still, they point technologists toward the problems that matter most: handoffs and alert routing, and whether clinicians can trust the data in front of them.
Why do timing and record integrity matter for patient outcomes?
Every tracing, timestamp, and alarm log an L&D system stores is connected to outcomes that can last a lifetime. According to the CDC, congenital cerebral palsy accounts for 85% to 90% of cases, and birth complications such as placental detachment, uterine rupture, or umbilical cord problems can disrupt the baby’s oxygen supply and result in cerebral palsy. The CDC also notes that in many cases the cause of cerebral palsy is not fully known, and many listed risk factors, such as low birthweight, multiple births, and infections during pregnancy, are present before labor begins.
Hypoxic-ischemic encephalopathy (HIE), a brain injury linked to reduced oxygen and blood flow around birth, shows why timing data matters. The NICHD Neonatal Research Network’s randomized trial found that whole-body therapeutic hypothermia reduced death or moderate-to-severe disability among newborns with HIE. Because cooling has to begin within hours of birth, eligibility decisions depend on accurate timestamps, Apgar scores, and documentation captured in the first hours of life. If a fetal monitor’s clock drifts from the EHR’s, or a tracing fails to archive during an interface outage, that is a data problem with direct clinical stakes.
Birth Injury Justice Center, the US-based educational resource behind childbirthinjuries.com, documents how conditions such as hypoxic-ischemic encephalopathy, cerebral palsy, and Erb’s palsy can follow complications during labor and delivery, and pairs that guidance with nurse and patient-advocate support for affected families. Its condition explainers, read alongside CDC and NICHD material, show health IT professionals what the data they manage ultimately represents: a child’s diagnosis, therapy plan, and family’s understanding of what happened at birth.
Which hybrid roles are emerging in labor and delivery?
The roles below overlap, and titles vary by health system. What they share is a position where clinical workflow meets data, usually with regulated technology somewhere in the middle.
| Role | Core skills | Where it sits in L&D workflows |
|---|---|---|
| Clinical informaticist | Clinical background, EHR build, workflow analysis, change management | Translates obstetric protocols into Epic or Oracle Health order sets, flowsheets, and alerts |
| Nursing informatics specialist | RN licensure, NICHD EFM nomenclature, documentation design, staff training | Aligns charting with standardized definitions and trains nurses on surveillance tools |
| Perinatal data specialist / healthcare data analyst | SQL, statistics, quality measurement, data visualization | Builds dashboards for perinatal quality collaboratives, AIM safety bundle reporting, and Joint Commission perinatal care measures |
| Biomedical engineer | Device integration, HL7 interfaces, FDA regulatory literacy, risk management | Maintains fetal monitors, central surveillance, and archiving; supports local validation of AI devices |
| Clinical IT analyst | Interface engines, alert design, downtime procedures | Keeps monitor-to-EHR data flowing and tracings archived and retrievable |
Engineers weighing a move into this space can compare these positions with other engineering roles in healthcare, many of which share the same device and systems skill set.
How can you evaluate whether an AI clinical tool is FDA-authorized?
Anyone selecting, implementing, or supporting clinical AI should be able to answer five questions before go-live:
- Is it on the FDA’s list? Search the FDA AI-enabled device list and the relevant 510(k), De Novo, or PMA database for the product and its authorization number.
- What is the authorized intended use? Confirm the population, gestational age range, and setting match how your unit plans to use it.
- Is there a predetermined change control plan? Know how and when the model can be updated after authorization.
- How was it validated locally? Compare performance on your own patient population and monitor for drift over time.
- Who acts on the output? Define the human in the loop, the escalation path, and the audit trail before the first alert fires.
Which skills transfer from general IT into clinical informatics?
More than most career-switchers expect. HL7 and FHIR interface work, data quality management, HIPAA-aligned security practice, and alert and interface design all carry over directly. So does change management, which matters in units where nurses are asked to trust a new system during an emergency.
The gaps are usually clinical vocabulary and quality frameworks. Learning EFM terminology, Apgar scoring, and the structure of AIM patient safety bundles, developed by the Alliance for Innovation on Maternal Health with a national team at ACOG, makes a technologist far more useful to an obstetric team. Given the staffing pressure described in Apollo Technical’s look at healthcare hiring in 2026, candidates who combine both skill sets are well placed.
For the next wave of obstetric AI, algorithm accuracy is only part of the job. Someone has to integrate each tool with fetal monitors and the EHR, route its alerts to the right clinician, validate it against local patients, and keep its data intact for years. Those tasks sit with informaticists, analysts, and engineers, and they are the roles health systems will need to fill as more AI tools reach the bedside.
Frequently asked questions
How does AI support fetal monitoring during labor?
AI tools analyze fetal heart rate patterns continuously and flag tracings that may indicate distress, giving clinicians an additional review layer rather than replacing their judgment. Missed distress has real consequences. Educational resources such as Birth Injury Justice Center note that oxygen deprivation during delivery is linked to conditions including HIE and cerebral palsy. The CDC likewise lists birth complications that disrupt oxygen supply among the risk factors for cerebral palsy.
What does a perinatal informatics role actually involve?
Most of the work is configuration and governance: building L&D documentation in the EHR, mapping monitor data into the record, tuning alerts, and training staff. Analysts in these roles also produce the data that perinatal quality collaboratives and hospital safety committees review.
Who regulates AI tools used in labor and delivery?
The FDA regulates AI software that meets the definition of a medical device, typically through 510(k) clearance, De Novo classification, or premarket approval. Hospitals remain responsible for local validation, governance, and how the tool is used in practice.
What qualifications do health information technologists need?
BLS reports that education requirements vary: some health information technologist and medical registrar roles typically require an associate’s degree, while others require a bachelor’s degree or higher. Employers may also prefer or require certification. Informatics roles with clinical decision-making authority usually require a clinical license as well.
When does fetal monitoring data matter after delivery?
Archived tracings, alarm logs, and EHR documentation are used in quality review, clinical follow-up, and family records long after discharge. That turns record integrity into an engineering requirement: synchronized clocks across monitors and the EHR, archives that survive system migrations and vendor changes, and audit trails showing who viewed or edited a record and when. Because tracings may need to be retrieved years later, retention policies and storage formats have to be planned for that timeline from the start.
Who should own an AI alert when it fires?
Ownership should be defined before go-live: which clinician receives it, on which device, what the escalation path is, and how the response is documented. Informatics teams typically design this with nursing leadership, physicians, and risk management.
What is the NICHD three-tier fetal heart rate classification?
It is the system described in the 2008 NICHD, ACOG, and SMFM workshop report that sorts tracings into Category I (normal), Category II (indeterminate), and Category III (abnormal). Many EHR documentation templates and fetal monitoring software tools are built around this vocabulary.
Outbound links
- FDA: Artificial Intelligence-Enabled Medical Devices — Primary source for the 1,600+ authorized AI-enabled device count.
- CDC NCHS Data Brief No. 571: Births in the United States, 2025 — Primary source for the 3,612,012 US births figure.
- BLS Occupational Outlook Handbook: Health Information Technologists and Medical Registrars, Job Outlook — Primary source for the 16% growth projection and $68,020 median wage.
- The 2008 NICHD Workshop Report on Electronic Fetal Monitoring (Obstetrics & Gynecology) — Primary source for EFM nomenclature and the three-tier classification.
- AHRQ Safety Program for Perinatal Care: Electronic Fetal Monitoring — Federal source for the 34% root-cause figure from Joint Commission sentinel event data.
- CDC: Risk Factors for Cerebral Palsy — Primary source for the 85%–90% congenital CP figure and birth-complication risk factors.
- NICHD Neonatal Research Network: Whole-Body Hypothermia for Neonatal HIE (PMC) — Peer-reviewed NICHD trial supporting the therapeutic hypothermia statement.
