Imen Halima

Creation of a solution in the pharmaceutical field based on AI, allowing users (healthcare professionals, researchers, pharmacists) to easily query pharmaceutical documents (leaflets, studies, PDF reports, etc.) via a natural language chatbot.

Main features:

. Automatic document ingestion

. PDF document upload (drug leaflets, clinical trials, regulations).

. Intelligent data extraction with OCR + NLP.

. Structuring and indexing in a vector database.

. Automatic structuring in the database

. Semantic mapping to a predefined pharmaceutical database schema (drugs, active ingredients, dosages, side effects, interactions).

. Use of LLM models specialized in biomedical to extract relevant information.

. Results in the form of tables, lists, graphs.

. Visualization of results with Power BI: creation of interactive dashboards for analyzing extracted data (dosages, side effects, interactions, etc.)

. Export options (CSV, JSON, or interactive display).

• Development of cell segmentation, detection, and tracking algorithms from microscopic images to analyze cell deformability by merging supervised and unsupervised models (U-Net, GAN, VAE).

• Creation of a database of cells in different states generated via adapted acquisition scenarios to validate the machine learning model.

• Integration of a corrective system for printing commands to improve the quality of 3D bioprinting and reduce production defects for biologists, chemists, and rheologists.

• Use of unsupervised models and physical laws in latent space

Classification of human postures from color, thermal, and depth images based on Deep Learning methods

Image Processing and 3D Vision:

OpenCV, Pytorch, Keras, CUDA, (SSD, YOLOv3, Faster R-CNN). Technologies: Python, Scikit-Learn, Tensorflow, Keras, Linux. Gaussian law, optical flow, Bayesian models.