Índice de desempeño para tareas de atención sostenida

Abstract

Through this work, we explore the idea of a translational attentional index that can be utilized in cognitive evaluation practice as well as in other areas in the biomedical space. With the increase of virtual assistants developed for in-house rehabilitation session monitoring – such as wearables and computer vision (CV) based AI assistants –, the standardization of an index based on behavioral indicators is imperative to give them the ability to monitor the users’ engagement level in the given task to procure a safety and effective execution. We begin by exploring the indicators of attention obtained from a Continuous Performance Test (CPT) – a standardized tool utilized in cognitive evaluation – where we detect shortcomings that cannot be translated to daily life scenarios where this index is intended to be used. We propose then an index that is computed based on common indicators that can be found in daily tasks (Correct Responses and Reaction Times) by weighting their contributions equally and bidirectionally penalizing variability on responses recorded. We close with presenting the information the new index can group from common indicators and that can be easily interpreted in a quick sight by different professionals in the biomedical space compared with the typical analysis of attentional levels obtained from a CPT.