Explainable and calibrated clinical ML

Prediction models for clinical decision support whose probabilities can be trusted and whose decisions can be explained to clinicians.

Example: general anaesthesia triage in paediatric dentistry.

Medical image and 3D segmentation

Deep learning for semantic segmentation, from U-Net on medical images to superpoint graphs on large-scale 3D point clouds.

Example: B.Sc. thesis on point cloud segmentation.

Time-series and anomaly detection

Signal processing and unsupervised models for sensor and bio-signal data, including predictive maintenance in heavy industry.

Example: Isolation Forest models at NICICO.

A calibrated prediction model for general anaesthesia triage in paediatric dentistry: development, explainability, and preliminary external evaluation

Seyed-Ali Sadegh-Zadeh, Mahshid Bagheri, Zahra Amini, Shayan Saadat, Mohammad Amin Barati, Delaram Jarchi

A calibrated ML prediction model on clinical tabular data for general anaesthesia triage, with explainable AI (XAI) techniques to ensure clinical transparency.

Introduction to Artificial Intelligence in Healthcare

Zahra Amini (lead author), et al. Shahid Beheshti University of Medical Sciences Press.

An interdisciplinary textbook bridging machine learning, data analytics and healthcare, for medical students and professionals.

Large-scale semantic point cloud segmentation with superpoint graphs

Zahra Amini. Department of Software Engineering, Sirjan University of Technology. Supervisor: Dr. Amir Salarpour. Grade: 20/20.

A 3D point cloud processing framework for semantic segmentation.

Project title (explainable ML)

One sentence on the problem, one on the method, one on the result.

Project title (medical image segmentation)

One sentence on the problem, one on the method, one on the result.

Project title (time-series)

One sentence on the problem, one on the method, one on the result.