Monitoring the brain of the sick newborn infant in the NICU using electroencephalography (EEG) is essential in order to accurately detect seizures and assess the severity of neonatal encephalopathy (NE). Confirmation of encephalopathy with EEG monitoring can act as an important adjunct to other investigations and the clinical examination, particularly when considering treatment strategies such as therapeutic hypothermia for hypoxia ischaemia. Assessment of the EEG can also offer valuable early information regarding prognosis. However, EEG interpretation in the NICU can be difficult as expertise is not widely available, especially on a 24-hour basis. Exciting developments in machine learning and signal processing offer the exciting prospect of real-time decision support for neonates with encephalopathy in the NICU. Technology has advanced dramatically in the last few years, and there have been exciting developments in the field of Machine Learning. We have recently developed an algorithm for neonatal seizure detection called ANSeR (www.anserstudy.com/). We used a comprehensive set of 55 features of neonatal EEG to train a support vector machine (SVM) learning algorithm as a classifier [1Ahmed R., Temko A., Marnane W., Lightbody G., Boylan G. Grading hypoxic–ischemic encephalopathy severity in neonatal EEG using GMM supervectors and the support vector machine Clin Neurophysiol 2016 ;  127 : 297-309 [inter-ref]

Haga clic aquí para ir a la sección de Referencias]. A real-time clinical investigation of this algorithm was then undertaken in 8 sites across Europe by comparing EEG monitoring with algorithm support to EEG monitoring without algorithm support. Seizure detection was higher in the algorithm group and the results of this study will be discussed in this presentation. Signal processing for the assessment of the severity of encephalopathy in neonates is at an earlier stage but a number of studies have been published and these will be reviewed [2Awal M.A., Lai M.M., Azemi G., Boashash B., Colditz P. EEG background features that predict outcome in term neonates with hypoxic ischaemic encephalopathy: a structured review Clin Neurophysiol 2016 ;  127 : 285-296 [cross-ref]

Haga clic aquí para ir a la sección de Referencias, 3Temko A., Thomas E., Marnane W., Lightbody G., Boylan G. EEG-based neonatal seizure detection with Support Vector Machines Clin Neurophysiol 2011 ;  122 : 464-473 [cross-ref]

Haga clic aquí para ir a la sección de Referencias]. A consensus in neonatal EEG grading is urgently needed in order to accurately train, test and validate algorithms for neonatal encephalopathy. This is not a trivial task and grading schemes need to be correlated with both short and long-term outcome data. A multidisciplinary approach is now required to advance EEG interpretation in neonatal encephalopathy.