Machine Learning for Working Memory Performance Prediction

This project addresses the need to characterize and predict the performance of Working Memory (WM), a cognitive system crucial for learning and reasoning, whose inter-subject variability and underlying neural mechanisms still require extensive investigation. The study focuses on how selective attention, modulated by features such as spatial cueing, color, or shape, influences WM performance, especially under novel paradigms where the cue is presented after the memory set. The methodology is structured in three phases. Initially, a comprehensive framework and state-of-the-art review are established regarding the application of Machine Learning (ML) in the context of WM. Subsequently, an experimental research method is implemented to evaluate various ML algorithms and modalities, searching for patterns within biological brain signals that identify performance and model adaptability. Finally, deductive and inductive techniques are applied to identify the key factors determining the efficiency of predictive models for WM performance, thereby contributing to the discovery of the implicated neural mechanisms.<br/><br/><b>Goal</b>: <br/>To analyze machine learning methodologies through the implementation of algorithms to characterize and predict Working Memory (WM) performance, exploring the underlying neural mechanisms of selective attention in cognitive tasks.<br/><br/><b>Research lines</b>: <br/>Telematics applied to medicine