Clinical Fellow 2026 - 2028

The Norman Beischer Medical Research Foundation is delighted to announce Dr Deb Karmakar as the 2026–2028 NBMRF Fellow.

The Fellowship will provide $660,000 over three years in funding, enabling Deb to continue his groundbreaking work focused on improving outcomes for mothers and babies through advanced technologies in fetal monitoring.  The project will co-design AI-enabled fetal monitoring decision support to improve perinatal outcomes, ensuring better detection of early signs of fetal distress and guiding safer, faster decision-making during childbirth. This continues work previously supported by NHMRC and Norman Beischer Medical Research Foundation Research Innovation grant funding.

By integrating clinicians and patients into the design process, this initiative seeks to make cutting-edge technology both effective and practical for real-world maternity care.

Cardiotocography (CTG) is a standard tool in maternity care.  It records a baby’s heart rate alongside the mother’s uterine contractions during labour and late pregnancy.  CTG has been used for decades, yet it remains difficult to interpret. Readings can vary between clinicians, and early warning signs of trouble are often subtle. Sadly, this means some babies still experience oxygen deprivation (asphyxia) even when their CTG appears “normal”.  In Australia, moderate to severe intrapartum asphyxia occurs in around 2–4 per 100 births, and rates are higher in low-resource settings. Up to 50-60% of stillbirths occur during labour. These events can lead to serious complications, including brain injury, long-term disability, or death. A substantial proportion of such devastating outcomes are potentially preventable through earlier detection. Poor recognition of asphyxia has also contributed to rising caesarean section rates.

Deb’s recent research has shown that certain patterns of CTG signal loss – previously dismissed as artifacts – are in fact linked to higher risk of asphyxia.  Recognising and interpreting these patterns offers the chance to detect fetal compromise earlier and intervene before harm occurs – while also reserving intervention for when it is truly needed.

In this Fellowship, Deb will build on that work to further develop and test a real-time, explainable artificial intelligence (AI) system called REALM™ (Real-time Evaluation and Alerting using Longitudinal Modelling). REALM™ continuously analyses CTG data in rolling 30 second windows, detecting subtle but important changes and providing clear, visual explanations for each alert. These explanations are transparent—clinicians can see exactly which patterns triggered an alert and why – helping to build trust in AI and ensuring decisions are based on interpretable information rather than a “black box”.  Upon development and testing across around 53000 labour episodes this system has detected asphyxia twice as often as clinicians in our proof-of-concept work.

Previous attempts at AI- based CTG analysis have relied on “cleaned” data segments for model developments development.  While this can improve technical performance, it strips away the messy reality of clinical monitoring and may explain why no system to date has reduced infant mortality or severe morbidity in randomised trials. Such approaches also limit generalisability to different hospital settings, geographies, and CTG machines. Dr Deb’s proof-of-concept work has instead used minimally processed CTG signals, achieving detection rates that exceed the state of the art, and validating results on diverse external datasets. A unique feature of this project is its strong focus on consumer co-design. The team will work closely with parents, midwives, doctors, and Aboriginal Community Controlled Health Organisations to shape how alerts are presented, ensuring the system is user-friendly, culturally safe, and trusted. This is particularly important for rural and regional hospitals, where timely access to specialist expertise can be limited. Deb is a co-inventor (alongside Associate prof Fiona Brownfoot) on two patents, related to this project, for integration of the algorithm with fetal monitoring hardware, supported by the University of Melbourne

The project will evaluate REALM™ on datasets from multiple Australian and international hospitals, comparing its performance to current CTG interpretation. The project goal is to demonstrate earlier and more accurate detection of fetal compromise, supporting regulatory approval as a certified medical device. By making fetal monitoring interpretation faster, more accurate, and transparent, this work can reduce preventable harm and improve birth outcomes.

Every family hopes for a safe delivery. Deb aims to give every baby the best start in life and every parent confidence that no warning sign will be missed.