Shakiba Baharvand

Shakiba Baharvand

Researcher in Computational Neuroscience and ML for Health

Welcome.

I am a Computer Engineering (Software) graduate from the University of Tehran. My research lies at the intersection of Computational Neuroscience, NeuroAI, and Machine Learning for Health. I am passionate about developing AI and software solutions that decode biological complexity to enable personalized healthcare.

Current Research

I am currently a Research Intern at the Royan Institute, where my work focuses on two distinct areas:

  • NeuroAI & Coding: Analyzing High-Density Multi-Electrode Array (HD-MEA) recordings using statistical techniques to evaluate how biological neural networks encode spatial activity patterns.
  • Health Informatics: Designing a data analysis platform that utilizes smartphone and smartwatch data to monitor Parkinson’s disease progression and generate clinical reports.

Selected Achievements

  • Best Undergraduate Project (2025): I was awarded for my thesis, “Development of a System for Parkinson’s Disease Diagnosis Through Voice Analysis Using AI”. I designed MPD-Net, a hybrid CNN-LSTM-Attention architecture, which achieved an AUC score of 1.00 on the IPVS dataset and 0.87 on UAMS.
  • Publication: I co-authored “The Eu-Ret mouse is a novel model of hyperdiploid B-cell acute lymphoblastic leukemia”, published in Leukemia (May 2024).
  • Neuromatch Academy (2024): I completed intensive training in Computational Neuroscience, modeling behavioral predictions using keypoint-based features on the CalMS21 dataset.

MPD-Net: Parkinson’s Disease Diagnosis

  • Overview: Designed a hybrid deep learning architecture to detect Parkinson’s disease from voice recordings.
  • Methodology: Utilized a CNN-LSTM-Attention model to capture both spatial features (spectrograms) and temporal dependencies in speech. Implemented Grad-CAM and SHAP for model interpretability.
  • Results: Achieved AUC scores of 1.00 (IPVS), 0.87 (UAMS), and 0.65 (mPower).
  • Code: GitHub Repository

Transcriptomic Analysis of B-cell Leukemia

  • Overview: Contributed computational analysis for the characterization of the Eu-Ret mouse model.
  • Methodology: Performed analysis of bulk RNA sequencing data using R (tidyverse, DESeq2, clusterProfiler). Conducted differential expression and pathway enrichment analyses to identify dysregulated genes in leukemic populations.
  • Outcome: Published in Leukemia (2024).

Behavioral Prediction Model (Neuromatch Academy)

  • Overview: Built a supervised behavioral prediction model using the CalMS21 dataset.
  • Methodology: Employed keypoint-based features to model and predict animal behavior within a computational neuroscience framework.

Background

Previously, I served as a Research Intern at the BC Children’s Hospital Research Institute (BCCHR) in Vancouver. There, I performed computational analysis of bulk RNA sequencing data using R and conducted differential expression analyses on leukemic B-cell populations. I have also led workshops on R programming, RNA-Seq analysis, and Computational Neuroscience for students.

Technical Toolkit

  • Languages: Python, R, C++, Java, Kotlin.
  • Machine Learning: PyTorch, TensorFlow, Scikit-learn, Deep Learning (CNNs, LSTMs, Attention), Model Interpretability (SHAP, Grad-CAM).
  • Bioinformatics: Neural Data Analysis, Bulk RNA-seq, Differential Expression Analysis.
  • Tools: Git, Linux, Docker, Power BI, Qt Framework.

Life Outside the Academia!

I enjoy classic poems and webtoons. I also like to draw comics. I have a unique appreciation (yes… appreciation!) for lizards 🦎 and find joy in coding - not just as a tool for science, but as a creative medium. And coding without a cup of coffee ☕? I don’t think that’s possible! 🙂