Published Projects

Cognitive Neuroscience

Real-time Emotion Recognition

2022-2023

(Neurohumanities Lab)
- Brain-Computer Interface based on the Database for Emotion Analysis using Physiological Signals (DEAP) in 8-channel EEG via PCA and RF [paper]; [repo]

^{Blanco-Ríos, Candela-Leal et al. (2024) Front. Hum. Neurosci.}

Mental Fatigue Prediction

2021

(Advanced Learner Assistance System [ALAS])
- 8-channel EEG [paper], real-time cloud-based implementation [proceeding], 4-channel EEG [proceeding], real-time implementation [poster]; [repo], [repo]

Interest in STEM Prediction

2021

(Talent and Passion Detection Through Biometrics)
- Subject-analysis [proceeding], Machine Learning model [proceeding], in-depth Power Spectral Density and Functional Connectivity analysis [paper]; [repo]

Computational Neuroscience

FeTA Challenge @ MICCAI

2024

- Trained a 7-label fetal MRI U-Net model zoo with spatial, intensity and resolution augmentation for enhanced multi-site generalization; [poster]

Human Biomechanics

Biomechanical Force Prediction

2021-2022

(Biomechanics for the Digital Twin)
- RNN proof-of-concept model design using external dataset [paper], RNN model fitting by gathering and using own data [proceeding]; [repo]

^{Candela-Leal et al. (2022) Appl. Sci.}

Non-Published Projects

Cognitive Neuroscience

Cognitive Load Dynamics in Chess

2023

- Calculated Task Completion Time (TCT) based on EEG biomarker theta C4, validated with Cognitive Load Theory (CLT) stratifying by chess level; [slides]

Brain on Acting

2022

- Unveil brain-to-brain communication patterns between actors during an acted scene via brain connectivity analysis and bisprectrum calculation; [poster]

Computational Neuroscience

High-res Fetal Subplate Segmentation

2024

- Implemented Bivariate Gaussian Smoothing (BGS) for fast construction of a high-resolution training dataset, further fitted into a pre-trained U-Net; [slides]

Non-linear qMRI for CHD Classification

2024

- Designed Recursive RF importance (RRFi) for feature selection (20,453), leading to a parsimonious 5-feature kNN model (F1-score = 0.88); [slides]

Human-Machine Interaction

Digital Twin of the Workspace

2022

- Design of a throughput monitoring system via a Human Action Recognition (HAR) CNN + RNN DL model, using a 3D camera and a 360° LiDAR pointcloud sensor, in addition to CV tracking using CCTV footage; [poster] [technical report]

University/Personal Projects

Closed-loop BCI for Attention

2024

- Trained a 3-feature MLR model with 0.72 R² that predicted attention by collecting 4-channel EEG CPT-II data, implemented in a real-time analog haptic neurofeedback and validated using a 12-min video; [poster] [technical report]

Knee Flexion Angle Estmation

2022-2023

- Leveraged by a L515 depth camera and 2D OpenPose keypoints, a real-time algorithm estimated knee flexion angle during a drop vertical jump, to identify risk factors for anterior cruciate ligament injury; [poster] [technical report]

Predictive Text Application

2020-2021

- Capstone project for JHU's Data Science specialization (288 h), fitted a Katz backoff model to predict next word via n-grams; [repo] [milestone report]

^{Application demo written on R and built using shiny; hosted at RPubs}