Cardiotocography (CTG) is the continuous recording of the fetal heart rate (FHR) using ultrasound transducers placed on the mother's abdomen. The signal obtained exhibits characteristic patterns that allow for the assessment of the fetus's well-being, behavioral states, and nervous system development. Information on the gestational age (GA) is crucial in fetal state and development evaluation. The objective of this work is to train a model to predict the GA from the FHR in healthy pregnancies, and assess whether the same model can identify deviations from normal development due to Intra Uterine Growth Restrictions (IUGR). The chosen AI model is based on a modified ResNet architecture and was trained with FHR signals from healthy subjects. The model is then tested on both Healthy and IUGR fetuses. This paradigm offers advantages such as eliminating the need for class balancing and opens up potential applications in various pathological conditions. The use of these AI methods was enabled by the NAPAMI database, from which we used 63342 signals over more than 80000 available. Preliminary results (R2: 0.50 in healthy and R2: 0.16 in IUGR) show that the proposed method is capable of detecting deviations from normal fetal development. We are working to improve robustness, interpretability, and performance to contribute to its applicability in real-world clinical settings.