Introduction: Sensory prediction (SP) is the ability to anticipate future stimulations on the basis of previous sensory inputs. It is related to repetition suppression (RS), the reduction of activity in the brain when a stimulus is repeated or becomes irrelevant. According to this theory, the first stage of sensory processing takes place in the primary sensory cortex where the RS would be observed, the second is frontal and corresponds to the establishment of SP based on primary cortex inputs. Oddball protocols, in which a rare deviant stimulus appears randomly in a sequence of frequent, repeated stimuli, and stimulus omission protocols in which an expected stimulus is omitted, are used to highlight these skills. Recent studies indicate that children born preterm may have altered SP and RS. Touch, the first sensory modality to develop, and the foundation of motor, cognitive and affective development, has already been used to evaluate the ability of preterm neonates to form predictions. The aim of the current study is to investigate the reliability of EEG and fNIRS to assess somatosensory prediction and its spatial distribution in preterm neonates. We hypothesize that preterm neonates will present RS in the somatosensory cortex in both EEG and fNIRS, and stronger early somatosensory and late frontal responses to deviance. We hypothesize that in fNIRS, preterm neonates will have an increase of HbO and a decrease of Hb during stimulus omission. Methods: We measured 21 preterm neonates so far out of 90 planned participants, born before 35 weeks of gestation. We built an 18 minutes vibrotactile oddball-omission protocol designed to generate SP and RS. 290 trials were 200ms-long vibrations on the forearm, with 3.5±0.2s interstimulus intervals. The first (Familiarization) and last (Control) 40 standard stimulations were used to quantify RS. In between, rare deviants (14%) and omissions (14%) were pseudo-randomly interspersed among standard trials to quantify SP. The brain activity was recorded in the patient’s room at the hospital at 35 weeks of corrected gestational age using simultaneous 128-channels EEG (Magstim EGI) and fNIRS (Imagent, ISS) at 690nm and 830nm with 4 sources located between the electrodes (2 placed over the somatosensory cortex and 2 placed over the frontal cortex) and 1 detector placed over frontocentral areas. The EEG data are preprocessed using a custom-made Matlab script while the fNIRS data are preprocessed using both Homer3 software and Matlab. In EEG, we compare the amplitude of the N140 in the somatosensory cortex and changes after 500ms in the frontal cortex between conditions. In fNIRS, we compare HbO and Hb concentration changes after standard trials for the 40 first vs 40 last stimuli for RS, and after deviant and omission trials for SP. Results: Preliminary analysis shows that for RS, there is a decrease of neuronal activity at P300 in the somatosensory cortex, as well as an increase of HbO and a decrease of Hb in the frontal cortex during the control phase, but no change was observed in the somatosensory cortex. For deviants, there is a late frontal negativity coupled with an increase of HbO and a decrease in Hb in the frontal cortex. Omission were associated with an increase in HbO and a decrease of Hb in the frontal cortex as well, without visible EEG changes. Conclusion: We found partially consistent results with EEG and fNIRS during a tactile prediction paradigm in preterm neonates. As suggested by the predictive coding theory, late frontal activation was found in EEG associated with frontal hemodynamic changes during prediction trials. For RS, EEG is consistent with theory in the primary somatosensory area, but frontal changes only were observed using fNIRS. As we include more participants, we wish to further investigate the complementarity of electrophysiology and optical imaging for the study of neonatal cognition.