Building computational tools at the intersection of physics, AI, and kidney disease research.
I am a physicist and software engineer based in Lagos. I build physics-inspired machine learning models to help solve clinical medical problems.
Currently researching: Chronic Kidney Disease (CKD) progression using EHR data.
- 🔬 Research focus: Scientific Machine Learning (SciML), Survival Analysis, Clinical Data Modeling, and Physics-Informed Neural Networks.
- 🧑🏫 Teaching: Neural Networks, Reinforcement Learning, and applied ML to postgraduate students.
- 🎓 Looking ahead: Exploring formal clinical research pathways in Computational Biology and Health Data Science.
- ☁️ Certified: AWS Solutions Architect.
- MIMIC-Renal-Dynamics (Repository private pending publication)
- An open-source exploration of how Chronic Kidney Disease progresses over time. I use the MIMIC-IV dataset and state-space modeling to track renal health degradation. (Includes preprocessing pipelines, survival analysis, and preprint manuscript).
- pinn-glomerular-filtration
- A 2D Physics-Informed Neural Network (PINN) that embeds convection-diffusion equations directly into the loss function to model kidney filtration under hypertensive conditions.
- Achieved <2.5% relative L2 error against the analytical solution.
- 📄 Preprint: Zenodo
- hybrid-vqe-drug-discovery
- An experimental pipeline combining classical multi-layer perceptrons with a 4-qubit quantum circuit (PennyLane) for molecular feature generation.
- 📄 Preprint: Zenodo
- Tox21-GNN-Property-Prediction
- A Message Passing Neural Network (MPNN) built with PyTorch Geometric for multi-label toxicity prediction across 12 Tox21 assays.
- AI & SciML: PyTorch, PyTorch Geometric, Scikit-Learn, Scikit-Survival, PINNs.
- Domain: EHR Data Analysis, RDKit (Cheminformatics), R (Biostatistics).
- Infrastructure: AWS, Docker, Git, Linux.
- 💼 LinkedIn: linkedin.com/in/abiodun-sojobi
- 📧 Email: asojobi@unilag.edu.ng
- 🏫 Faculty Profile: Rome Business School
"A physics-trained approach to medical discovery."