One of the most reliable approaches used for the functional assessment of the integrity of the young nervous system is the Prechtl assessment of general movements. It is based on the visual Gestalt perception of normal versus abnormal movements that comprise the entire body and manifest themselves in variable versus less variable sequences of arm, leg, neck, and trunk movements in preterm, term, and young infants.[1] Apart from being non-intrusive and cost-effective, the assessment is easily learned and results in a 92% accuracy rate in differentiating between normal and abnormal general movements after completing a mere 4- to 5-days training course.[2] Experienced observers consistently achieve high interrater agreements, ranging from 89 to 93%.[1] Regardless of these data demonstrating its objectivity and reliability, critics often label the Gestalt assessment of general movements as being too subjective. Moreover, they note it requires well-trained and experienced observers.[3-5] But which diagnostic tool does not require training and expertise?

Admittedly, Gestalt perception is vulnerable to the observer's tiredness and/or to the omission of proper recalibration with criterion standards for each age-specific general movements pattern assessed. Furthermore, Gestalt perception is a particularly powerful tool when it comes to the analysis of complex phenomena but has its limitations when the observer focuses on details.[1, 2]

During the last few years a number of computer-based movement assessment tools using optical flow meters[3, 6] or electromagnetic tracking systems[4, 7] have aimed to perform quantitative analyses of general movements, and thereby to discriminate more objectively between normal and abnormal movement patterns. Two-dimensional representations of general movements over time were derived from reflective marker signals[4, 5, 7] or – more convenient for the infant – directly from the monitor after placing infants on a special mattress during the recording.[3, 6]

Most of the analyses focused on tracking multivariate interactions (or the lack thereof) in the kinematic chains of the upper and lower limbs.[4, 7, 8] Also, Kanemaru et al.[5] proposed that a high correlation between limb velocities of at least two limbs might indicate a lack of normal variability. Indeed, lack of variability is the key feature of abnormal general movements. From term age until the end of the second month after term, abnormal general movements are called ‘poor repertoire general movements’ if the sequence of movement components is monotonous, and the amplitude, speed, and intensity lack normal variability. ‘Cramped-synchronized’ general movements, however, are also monotonous but appear rigid as the limb and trunk muscles contract almost simultaneously and relax almost simultaneously.[1] Unfortunately, computerized analyses of general movements at this age[5] could not yet differentiate between these two abnormal general movement patterns, which would be essential as cramped-synchronized general movements have a much higher predictive value for adverse neurodevelopmental outcome than poor repertoire general movements.[1] On the other hand, Kanemaru et al. found lower average velocities of legs and higher kurtosis of accelerations of arms in infants at term age who were scored as being developmentally delayed (but not necessarily neurologically abnormal) at age 3.[5]

Others applied computer-based general movement analyses at the age of 3 to 4 months,[3, 4, 6-8] when general movements appear as ‘fidgety movements’. The Prechtl assessment of fidgety movements revealed a sensitivity of 94% and a specificity of 82 to 100% when predicting neurodevelopment.[1] Interestingly, computer-based analyses of general movements at this age also identified infants who later developed cerebral palsy with a sensitivity of 85 to 90% and a specificity of 88 to 96%.[7, 8]

So far, there are only data on five infants published where both methodological approaches were compared: the mean accuracy between visual Gestalt perception and computer analysis was 0.77.[4] More comparison studies are certainly needed.

For the time being, computer-based analyses of quantitative general movement aspects cannot replace the visual Gestalt assessment of the quality of general movements. Both methodological approaches should rather complement each other, and for optimal effect would be combined with structural and other functional assessment methods to delineate comprehensively the integrity of the developing nervous system.


  1. Top of page
  2. References