Can machine learning read a baby's heart-rate trace in labour?
A scoping review of 36 studies found promise for predicting fetal hypoxia from cardiotocography, but weak benchmarks and small, repeated datasets.
The question
During labour, cardiotocography (CTG) tracks the baby’s heart rate to spot a lack of oxygen. Reading it is hard and varies between clinicians. Could machine learning help?
What we did
We searched PubMed, EMBASE and IEEE Xplore and reviewed 36 studies that used signal processing and machine learning on CTG.
What we found
Studies used many ways to extract signals and many algorithms. Most used one open-access database and treated fetal acidosis as the benchmark for hypoxia, with differing pH cut-offs, which we flagged as a concern. That database has few subjects for machine learning.
Why it matters
ML shows potential, but progress to the clinic needs more diverse data, standard definitions of hypoxia and better algorithms and features.