Machine learning · Biomedical signal processing · Python
My research focuses on estimating arterial stiffness and studying age-related vascular changes from photoplethysmography (PPG). I hold an M.Sc. in Mechatronics from the University of Tehran. This portfolio brings together the related research code and smaller projects in classification, clustering, and transfer learning.
Email · Google Scholar · ORCID
| Project | Focus and current evidence |
|---|---|
| Feature-based cf-PWV estimation — IEEE Access | PPG/SDPPG features and XGBoost regression. Five-fold evaluation on 4,333 eligible radial PWDB subjects: mean RMSE 0.1809 m/s and R² 0.9925. |
| PPG age-category benchmark | MLP, CNN1D, and CNN2D evaluated with five shared folds at two sites, using 4,374 PWDB subjects per site. VGG16 has separate artificial-data software checks. |
| CNN–BiLSTM–Attention for cf-PWV | Compare waveform and spectrogram representations with temporal attention. Both paths have training and checkpoint checks; public-data preparation is verified and full PWDB evaluation remains pending. |
| ResNet-18 spectrogram regression | PyTorch regression with separate training, validation, and test subjects. The CPU demo and checkpoint checks pass; full PWDB evaluation remains pending. |
PWDB contains simulated virtual adults. The recorded public-data evaluations use the current refactored implementations; full reproduction of every published result and clinical validation have not been established. Each research repository links its paper, protocol, and validation evidence.
- MobileNetV2 on CIFAR-10: frozen-backbone transfer learning, held-out image classification, and checkpoint reuse in an executed notebook.
- Stroke classification pipeline: preprocessing inside cross-validation, imbalanced-class metrics, and model selection on a synthetic tabular example.
- NumPy implementations of linear SVM, multinomial Naive Bayes, and K-means, plus a PyTorch tabular MLP.
Signal preprocessing · PPG/SDPPG features · Spectrograms · Classification and regression · Subject-level evaluation · Saved-model inference
Tools: Python · NumPy · pandas · SciPy · scikit-learn · TensorFlow/Keras · PyTorch · Matplotlib
For collaboration or project enquiries: kianaabrisham@gmail.com.