Skip to content
View pascalbartschi's full-sized avatar

Highlights

  • Pro

Block or report pascalbartschi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
pascalbartschi/README.md

Hi, I'm Pascal ๐Ÿ‘‹

๐Ÿš€ Computational Biologist | AI for Science | Data Enthusiast

๐Ÿ”ฌ About Me

  • ๐ŸŽ“ MSc in Computational Biology from ETH Zurich, specializing in machine learning for scientific applications
  • ๐Ÿง  Interested in deep learning, probabilistic modeling, and neural operators applied to real-world data
  • ๐Ÿš‘ Passionate about leveraging ML to tackle challenges in healthcare and the life sciences

๐Ÿ“„ Master Thesis

Inference-Time Guidance in Pocket-Conditioned Molecular Diffusion Models: Limits in Preventing Steric Clashes (2026) Implemented a structure-based molecular diffusion model using geometric classifier guidance derived from protein side-chain interactions to steer ligand generation within binding pockets.

diffusion_animation_16to9.mp4

๐Ÿ›  Recent Semester Projects

๐Ÿง  AI for Health & Biology

  • ICU Time-Series Mortality Prediction (2025) Built deep learning models and LLM prompts to predict ICU patient mortality from multivariate 48-hour time series data.

  • Interpretable Medical Prediction Models (X-Rays | Tabular) (2025) Developed interpretable ML models for pneumonia detection from chest X-rays and heart failure prediction, combining CNN-based explainability (Integrated Gradients, Grad-CAM) with transparent tabular methods (Logistic Lasso, SHAP, NAMs).

  • Bayesian Optimization for Molecule Design (2024) Applied constrained Bayesian optimization to tune molecular structures balancing bioavailability and synthesizability.

  • cfDNA Machine Learning Analysis (2022) Analyzed cell-free DNA profiles with classical ML to detect disease-relevant patterns from multi-cohort patient datasets.

๐ŸŒŠ Physics-Informed ML

๐ŸŽฎ Reinforcement & Probabilistic Learning

๐Ÿ“Š GitHub Stats

Pinned Loading

  1. tipping-point-symmetry tipping-point-symmetry Public

    Experiments performed with symmetric ecosystem model displaying hysteresis, as well as data storage and analysis.

    HTML

  2. bnn-satellite-img-classification bnn-satellite-img-classification Public

    Using a bayesian neural network trained with the SWAG method classifiy the topology of satellite images with a probilistic approach.

    Jupyter Notebook

  3. bo-molecule-synthetization bo-molecule-synthetization Public

    Using baysian optimization to maximize bioavailability of a drug candidate while respecting a constraint function.

    Python

  4. fno-wave-equation fno-wave-equation Public

    Modelling dynamics governing the wave equation using a Fourier Neural Operator trained with One-to-One and All2All training.

    Jupyter Notebook 2

  5. foundation-neural-operator-allen-cahn foundation-neural-operator-allen-cahn Public

    Foundation model using Fourier Neural Operators (FNO) to approximate solutions of the Allen-Cahn equation, exploring generalization across varying phase field dynamics.

    Jupyter Notebook

  6. ICU-TimeSeries-Mortality-Prediction ICU-TimeSeries-Mortality-Prediction Public

    Leveraging AI to predict mortality of intensive care patients in a binary classification problem from time series ICU data.

    Jupyter Notebook