MOHAMED SRAITIH

Dr. MOHAMED SRAITIH

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 November 2021 in Journal of Clinical Medicine
Verified
Journal Article
Open access
PDF file
MSMOHAMED SRAITIH
MOHAMED SRAITIH
YJYounes Jabrane
Younes Jabrane
Amir Hajjam El HassaniAmir Hajjam El Hassani
Amir Hajjam El Hassani
The new advances in multiple types of devices and machine learning models provide opportunities for practical automatic computer-aided diagnosis (CAD) systems for ECG classification methods to be practicable in an actual clinical environment. This imposes the requirements for the ECG arrhythmia classification methods that are inter-patient. We aim in this paper to design and investigate an automatic classification system using a new comprehensive ECG database inter-patient paradigm separation to improve the minority arrhythmical classes detection without performing any features extraction. We investigated four supervised machine learning models: support vector machine (SVM), k-nearest neighbors (KNN), Random Forest (RF), and the ensemble of these three methods. We test the performance of these techniques in classifying: Normal beat (NOR), Left Bundle Branch Block Beat (LBBB), Right Bundle Branch Block Beat (RBBB), Premature Atrial Contraction (PAC), and Premature Ventricular Contraction (PVC), using inter-patient real ECG records from MIT-DB after segmentation and normalization of the data, and measuring four metrics: accuracy, precision, recall, and f1-score. The experimental results emphasized that with applying no complicated data pre-processing or feature engineering methods, the SVM classifier outperforms the other methods using our proposed inter-patient paradigm, in terms of all metrics used in experiments, achieving an accuracy of 0.83 and in terms of computational cost, which remains a very important factor in implementing classification models for ECG arrhythmia. This method is more realistic in a clinical environment, where varieties of ECG signals are collected from different patients.

23 August 2022 in Journal of Clinical Medicine
Verified
Journal Article
Open access
PDF file
MSMOHAMED SRAITIH
MOHAMED SRAITIH
YJYounes Jabrane
Younes Jabrane
Amir Hajjam El HassaniAmir Hajjam El Hassani
Amir Hajjam El Hassani
An automatic electrocardiogram (ECG) myocardial infarction detection system needs to satisfy several requirements to be efficient in real-world practice. These requirements, such as reliability, less complexity, and high performance in decision-making, remain very important in a realistic clinical environment. In this study, we investigated an automatic ECG myocardial infarction detection system and presented a new approach to evaluate its robustness and durability performance in classifying the myocardial infarction (with no feature extraction) under different noise types. We employed three well-known supervised machine learning models: support vector machine (SVM), k-nearest neighbors (KNN), and random forest (RF), and tested the performance and robustness of these techniques in classifying normal (NOR) and myocardial infarction (MI) using real ECG records from the PTB database after normalization and segmentation of the data, with a suggested inter-patient paradigm separation as well as noise from the MIT-BIH noise stress test database (NSTDB). Finally, we measured four metrics: accuracy, precision, recall, and F1-score. The simulation revealed that all of the models performed well, with values of over 0.50 at lower SNR levels, in terms of all the metrics investigated against different types of noise, indicating that they are encouraging and acceptable under extreme noise situations are are thus considered sustainable and robust models for specific forms of noise. All of the methods tested could be used as ECG myocardial infarction detection tools in real-world practice under challenging circumstances.

1 June 2021
Verified
Conference Paper
Closed access
MSMOHAMED SRAITIH
MOHAMED SRAITIH
YJYounes Jabrane
Younes Jabrane
AAAbdelghafour Atlas
Abdelghafour Atlas
The automated classifiers can assist cardiology experts in diagnosing heart-linked illnesses with the initial and well-predicted results of classified ECG signals and data of subjects. This work serves as a review of various research works on ECG data and signals classification using supervised machine learning approaches. We presented and reviewed, the performance and the strategies used to classify cardiac arrhythmias using five widely used classifiers, Support vector machine (SVM), Random forest (RF), Decision three (DT), Naïve Bayes (NB), and K-nearest neighbour (KNN). And we discussed a variety of limitations and revealed that there is still area for increase in the classification's performance, specially by minimizing the preprocessing and feature extraction step which can cause an important computational cost and different accuracy outcomes. Such a coordinated research review allows scientists to merge an unobstructed view on some aspects of ECG classification for the identification of the research issues unmet so far. We plan to consider the obstacles discussed and some limitations to further raise the classification performance of the automated classifiers for the applicability of such solutions in clinical diagnosis.

28 December 2024 in PsyArXiv
Verified
Preprint
Open access
MCMohamed-Amine Chadi
Mohamed-Amine Chadi
MSMOHAMED SRAITIH
MOHAMED SRAITIH
Sleep-related Breathing Disorders (SRBD) are one of the most common sleep disorders. Patients suffer from repetitions of complete stops (apneas) or partial reductions (hypopneas) in respiratory flow during sleep. The standard method for the diagnosis of sleep apnea is the attended cardiorespiratory polysomnography (PSG). However, this method is expensive, and the extensive recording equipment can have a significant impact on sleep quality, falsifying the results. This work aims to investigate and validate a simplified automated system that uses only a tracheal sound (TS) sensor. A signal processing-based algorithm was implemented to detect respiratory events and thus estimate the Apnea Hypopnea Index (AHI). For validation, 13 full-night PSG recordings were analyzed. The present method consists of using only TS for respiratory event detection. Considering PSG as a gold standard reference, the developed approach reached the following results: For apnea detection, positive predictive value (PPV) reached 82%, and the sensitivity was 90%. For hypopnea detection, PPV reached 40%, and the sensitivity was 58%. For AHI estimation, 93%accuracy was achieved. In conclusion, we worked on a new, simpler approach than PSG, yet very promising, to reliably detect apnea and estimate the AHI score. However, hypopnoea detection still needs more future improvements.