AI in Medicine
Example of using AI for epileptic diagnosis.In our research, we focus on investigating the use of signal processing tools for biomedical diagnosis. In order to assist neurologist and biologist during their daily activity and save the time spent in the manual processing of the records, we need focus on building efficient machine learning models based on the extraction of pertinent and discriminative features for the prediction of biomedical signals.
Our proposed AI models focuses on the following points:
- Using Signal Processing (DSP) features
- Improving the raw data quality and reducing feature vector size
- Improving the classifier robustness or sensitivity to noise.
- In our different studies, we targeted the following aspects:
Abderrazak Chahid
Data Scientist
My research interests include feature extraction, computer vision, real-time implementation of smart decision making systems.
Publications
Improving Hand Gesture Recognition
New Feature extraction method based on the Quantization-based position Weight Matrix (QuPWM) method designed explicitly for multiclass classification on biomedical signals.
QuPWM: Feature extraction Method for MEG Epileptic Spike Detection
We developed a Feature extraction method, called QuPWM, for epileptic spikes detection in MEG signals. This method is based on combining the position weight matrix (PWM) method with digital quantization.
Quantum epileptic Spikes Detection
New Feature extraction method based on the Quantization-based Semi-Classical Signal Analysis designed explicitly for epileptic spikes using Magnetoencephalography (MEG) signals.
Poly(A) prediction signals in human genomic DNA
We developed New hybrid model for poly(A) signal prediction in human DNA is developed. It contains 8 deep neural networks and 4 logistic regression models. A novel feature generation method is used to extract relevant patterns in the DNA sequences.
