An automated tool that trains a multimodal classification model (EEG + fMRI) to predict phenotypic variables (sex, age, diagnosis) and assess whether fusing both modalities improves prediction compared to each modality alone.
This project is a complete and reproducible ML pipeline using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to: Predict tau PET SUVR as a continuous outcome, predict tau positivity as a binary outcome (classify participants as tau-positive or tau-negative), and identify which predictors drive model performance using SHAP explainability
This project applies machine learning to decode perceived emotions from fMRI data using ROI-based features. Data from the ds003548 OpenNeuro dataset are analyzed, with task labels extracted from events files. ROI time series are extracted using the MIST 64-ROI atlas, and mean signals during emotion blocks are classified using linear SVM. The goal is to distinguish between six conditions (happy, sad, angry, neutral, blank, scrambled), demonstrating key concepts and challenges in neuroimaging-based classification.
Firstly, this project investigates differences in frontoparietal brain connectivity between individuals diagnosed with ADHD and control participants during Spatial Working Memory Task, using fMRI-based connectivity data. In the second part of this project, To classify individuals as either having ADHD or being in control group based on functional connectivity data features machine learning models was tested by using k-fold cross validation.
This project trains machine learning classification models to make predictions of adhd diagnosis from brain fRMI connectivity measures which are obtained from a resting state ADHD dataset . The main goals of this project are to get more practice with machine learning tools and to learn how work with brain data more precisely fMRI data.
This project aims to evaluate the similarity between the representations of artificial neural networks (ANNs) and the visual system in the mouse brain using Representational Similarity Analysis (RSA).
While Parkinson’s disease (PD) is recognized by its motor symptoms, it is also characterized by non-motor symptoms (NMS), such as anxiety, depression, pain, etc. NMS often precede cognitive decline, making them potential predictors of such decline. This project aims to investigate the longitudinal association between non-motor symptoms and cognitive and neural decline in patients with PD. Early identification of individuals at higher risk of cognitive decline through their NMS presentation can facilitate timely interventions.
We used the ADHD-200 Sample dataset to implement various machine learning classification models, aimed at diagnosing ADHD through resting-state fMRI signals.
In this project, we aim to use machine learning on EEG data from participants’ language learning tasks on Duolingo. Specifically, we ask if EEG features can predict whether the participant has gotten a task right or wrong when they receive feedback. Using a k-nearest neighbours classifier, we achieve 98% accuracy in determining correct or incorrect answers based on EEG voltages from 8 electrodes.
Our project aims to develop different methods for the analysis of behavior in mice (in this case, exploration of an object) to determine which is the best approach to this kind of study. We were able to implement and compare three increasingly complex methods to determine exploration time: manual labeling; motion tracking and data analysis using a custom algorithm; training a Machine Learning classifier on our labeled data.
Copyright (c) 2026, BrainHack School; all rights reserved.
Template by Bootstrapious. Ported to Hugo by DevCows.