The GENIUS project aims at developing a machine learning-based model that can be used for the comprehensive and objective evaluation of medical training during the acquisition of skills, based on available data sources (e.g. behavioral data, neurophysiological signals, etc.). In particular, making errors could have important consequences in some operational environments, such as in microsurgery field by using robotic assistance. Robots for assisted surgery (i.e. da Vinci system) are far from being user-friendly and require extensive training. In this regard, it is clear as an effective training procedure and skills maintenance assessment is of high relevance to guarantee patient safety and avoid or at least minimize errors probability. Traditional training assessment methods rely on Instructor (i.e. expert physician) supervision and task performance evaluation of trainees during repeated practice. Nevertheless, although the results of different operators in terms of performance should be the same, cognitive demands could be not (e.g. less mental workload, less allocated attentional resources, improved situation awareness/decision making, etc.). This aspect highlights as nowadays one of the current limitations of the standard training assessment procedures is the lack of objective information regarding the actual amount of employed cognitive resources that could significantly affect expertise assessment during training. The aim of the GENIUS project is to develop a machine learning-based model to arise and track informative training-dependent features and, at the same time, to combine the available information to generate an objective and even real-time "Comprehensive Training Index".
In a long-term vision, the GENIUS model will be used as i) a prediction tool for the Instructor to plan and a better tailoring of medical training, ii) an evaluation instrument to assess surgeon skills maintenance over time.
Thanks to the great progresses in technology, it is possible to collect a huge amount of metrics of human performance training programs by using robotic surgical systems or simulators, thus big data analysis have been proposed as the proper approach to synthesize reliable measures of training level by a multitude of human performance metrics. However, the performance metrics have an intrinsic limit in evaluating the actual 'operational readiness' of an operator, since different operators could achieve the same performance but by using a different amount of cognitive resources. The possibility of having insights, in terms of measures, by a neurophysiological/cognitive perspective would provide the Instructor with a powerful tool. In fact, thanks to the GENIUS project it will be possible not only to answer the question "Is this operator able to work ensuring high performance standards?", but also to assess if such operator is able to ensure the same performance maintaining adequate mental conditions also in adverse situations, and over time.
Within neuroscientific literature, several works investigated the correlation between operators' cognitive processes and neurophysiological measures. Anyhow, just few of them exploited neurophysiological measurements during simple task learning (e.g. N-back task, Buschkuehl et al., 2014), moreover by using "lumbering" and really expensive instruments (i.e. functional Magnetic Resonance Imaging - fMRI). On the contrary, the proponent of the project boast a great expertise through several collaborations and projects, in particular in aviation and automotive domains regarding the possibility to assess cognitive progresses of trainees depending on learning. Such evidences represent the demonstration of the proposed GENIUS model potentialities. A multidimensional model based only on behavioral and performance metrics, as well as any other no-cognitive data, could ensure that an operator is able to guarantee the best performance, and also in simulated critical situations. But, for example, is he/she experiencing a great stress that could deteriorate his/her performance? Is he/she focusing too much his/her attention losing in situation awareness? Is he/she experiencing a cognitive closure that could affect his/her decision-making ability? Such kind of information becomes very important also for renewing the operator license, or for verifying their operational readiness after a certain period, or in any other occasions, different from the initial training course. In fact, after a particular period an operator could seem still able to operate with high performance standards, but actually his/her cognitive profile is changed. Up to now, there are no evidences of similar tools able to combine such a multitude of information (performance, behavioral and neurophysiological data), and provide a comprehensive objective measure of the training level, and in general, of the operational readiness of the operator.
Last but not least, in future scenarios the GENIUS model could be employed also to compare the effectiveness of different training programs, as well as tools used within the programs themselves or in real operational situations.
Bibliography:
Aricò et al., 2015. ATCO: Neurophysiological Analysis Of The Training And Of The Workload. Italian Journal of Aerospace Medicine 1.
Buschkuehl et al., 2014. Neural Effects of Short-Term Training on Working Memory. Cogn Affect Behav Neurosci 14, 147-160. doi:10.3758/s13415-013-0244-9
Diedrichsen et al., 2005. Neural Correlates of Reach Errors. J. Neurosci. 25, 9919¿9931. doi:10.1523/JNEUROSCI.1874-05.2005
Estes, 2014. Handbook of Learning and Cognitive Processes (Volume 4): Attention and Memory. Psychology Press.
Hart and Staveland, 1988. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research, in: Peter A. Hancock and Najmedin Meshkati (Ed.), Advances in Psychology, Human Mental Workload. North-Holland, pp. 139-183.
Kelly and Garavan, 2005. Human functional neuroimaging of brain changes associated with practice. Cereb. Cortex 15, 1089-1102. doi:10.1093/cercor/bhi005
McLeod et al., 2006. The acquisition of tacit knowledge in medical education: learning by doing. Medical Education 40, 146-149. doi:10.1111/j.1365-2929.2005.02370.x
Mutha et al., 2011. Critical neural substrates for correcting unexpected trajectory errors and learning from them. Brain 134, 3647-3661. doi:10.1093/brain/awr275
Pisella et al., 2000. An 'automatic pilot' for the hand in human posterior parietal cortex: toward reinterpreting optic ataxia. Nat Neurosci 3, 729-736. doi:10.1038/76694
Ritter and Scott, 2007. Design of a Proficiency-Based Skills Training Curriculum for the Fundamentals of Laparoscopic Surgery. Surgical Innovation 14, 107-112. doi:10.1177/1553350607302329