AHAlae Eddine El Hmimdi

Alae Eddine El Hmimdi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Alae Eddine El Hmimdi has a total of 4 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • The contribution of AI to neurological analysis of eye movements : application to learning disorders and dyslexia

    by Alae Eddine El Hmimdi

    La contribution de l'IA à l'analyse neurologique des mouvements oculaires : application aux troubles d'apprentissage et à la dyslexie: Cette thèse vise à apporter une contribution à la neuroscience en utilisant l'apprentissage profond pour l'analyse des données de mouvements oculaires. Pendant mon stage au laboratoire IRIS, nous avons déployé le logiciel AIDEALdans le cloud, initialement partie intégrante du traitement REMOBI & AIDEAL, facilitant ainsi l'accès et l'analyse anonyme des données, conformément au RGPD, afin d'améliorer les différentes méthodes et logiciels de traitement de signaux. Ensuite, en utilisant les enregistrements anonymisés d'AIDEAL, nous avons constitué une vaste banque de données sur les mouvements oculaires, utilisée pour construire des ensembles de données pour l'apprentissage automatique et profond pour diverses études. Parallèlement, nous avons appliqué des approches d'apprentissage automatique pour le dépistage de la dyslexie, en utilisant des données de 120 adolescents dyslexiques et non dyslexiques, avec des paramètres de mouvement binoculaire calculés par AIDEAL, à partir de plusieurs enregistrements de tâches visuelles, englobant des stimuli avec REMOBI (saccades, vergences), des tests de lecture et l'exploration libre de peintures. Nous avons ensuite exploré la prédiction de la vitesse de lecture à l'aide de données de tests visuels provenant de jeux de données de lecture, de saccades et de vergences. Dans une deuxième approche, nous avons analysé les paramètres oculomoteurs lors de l'exploration de peintures, en développant un nouveau descripteur appelé DGI, spécialisé pour les tableaux de peinture d'illusion optique, permettant une prédiction avec un taux d'exactitude jusqu'à 89,71 %. Nous avons ensuite étudié la stabilité de ce descripteur à travers sept tableaux de peinture ainsi que sa complémentarité avec les paramètres de mouvement oculaire calculés par AIDEAL. De plus, nous avons développé une approche de bout en bout pour la classification des mouvements oculaires en utilisant l'apprentissage profond, surpassant les techniques traditionnelles pour différencier les troubles d'apprentissage. Nous avons étendu notre approche pour inclure des informations contextuelles supplémentaires ainsi qu'une extension du problème initial binaire, afin d'identifier plusieurs groupes de pathologies, passant d'une classification binaire à une classification multi-annotations. Nous avons ensuite proposé deux nouvelles architectures (Hybride et Convolutionnelle) multi-segments, ainsi qu'une méthode physiologique d'augmentation de données (EMULATE), simulant les mouvements de tête avant et durant l'enregistrement des yeux, pour générer de nouvelles données. Nous avons également exploré des algorithmes d'auto-apprentissage pour le dépistage des pathologies, obtenant des résultats prometteurs dans la détection des troubles de lecture et d'autres pathologies. Enfin, passant de la recherche à l'application clinique, nous avons intégré les méthodes développées dans un assistant IA, facilitant leur utilisation pratique.

  • Open Access

    Distinguishing Dyslexia, Attention Deficit, and Learning Disorders: Insights from AI and Eye Movements

    by Alae Eddine El Hmimdi, Zoï Kapoula

    This study investigates whether eye movement abnormalities can differentiate between distinct clinical annotations of dyslexia, attention deficit, or school learning difficulties in children. Utilizing a selection of saccade and vergence eye movement data from a large clinical dataset recorded across 20 European centers using the REMOBI and AIDEAL technologies, this research study focuses on individuals annotated with only one of the three annotations. The selected dataset includes 355 individuals for saccade tests and 454 for vergence tasks. Eye movement analysis was performed with AIDEAL software. Key parameters, such as amplitude, latency, duration, and velocity, are extracted and processed to remove outliers and standardize values. Machine learning models, including logistic regression, random forest, support vector machines, and neural networks, are trained using a GroupKFold strategy to ensure patient data are present in either the training or test set. Results from the machine learning models revealed that children annotated solely with dyslexia could be successfully identified based on their saccade and vergence eye movements, while identification of the other two categories was less distinct. Statistical evaluation using the Kruskal–Wallis test highlighted significant group mean differences in several saccade parameters, such as a velocity and latency, particularly for dyslexics relative to the other two groups. These findings suggest that specific terminology, such as “dyslexia”, may capture unique eye movement patterns, underscoring the importance of eye movement analysis as a diagnostic tool for understanding the complexity of these conditions. This study emphasizes the potential of eye movement analysis in refining diagnostic precision and capturing the nuanced differences between dyslexia, attention deficits, and general learning difficulties.


  • Open Access

    Can Saccade and Vergence Properties Discriminate Stroke Survivors from Individuals with Other Pathologies? A Machine Learning Approach

    by Alae Eddine El Hmimdi, Zoï Kapoula

    Recent studies applying machine learning (ML) to saccade and vergence eye movements have demonstrated the ability to distinguish individuals with dyslexia, learning disorders, or attention disorders from healthy individuals or those with other pathologies. Stroke patients are known to exhibit visual deficits and eye movement disorders. This study focused on saccade and vergence measurements using REMOBI technology V3 and the Pupil Core eye tracker. Eye movement data were automatically analyzed with the AIDEAL V3 (Artificial Intelligence Eye Movement Analysis) cloud software developed by Orasis-Ear. This software computes multiple parameters for each type of eye movement, including the latency, accuracy, velocity, duration, and disconjugacy. Three ML models (logistic regression, support vector machine, random forest) were applied to the saccade and vergence eye movement features provided by AIDEAL to identify stroke patients from other groups: a population of children with learning disorders and a population with a broader spectrum of dysfunctions or pathologies (including children and adults). The different classifiers achieved macro F1 scores of up to 75.9% in identifying stroke patients based on the saccade and vergence parameters. An additional ML analysis using age-matched groups of stroke patients and adults or seniors reduced the influence of large age differences. This analysis resulted in even higher F1 scores across all three ML models, as the comparison group predominantly included healthy individuals, including some with presbycusis. In conclusion, ML applied to saccade and vergence eye movement parameters, as measured by the REMOBI and AIDEAL technology, is a sensitive method for the detection of stroke-related sequelae. This approach could be further developed as a clinical tool to evaluate recovery, compensation, and the evolution of neurological deficits in stroke patients.


  • Open Access

    ORASIS-MAE Harnesses the Potential of Self-Learning from Partially Annotated Clinical Eye Movement Records

    by Alae Eddine El Hmimdi, Themis Palpanas, Zoï Kapoula

    Self-supervised learning (SSL) has gained significant attention in the past decade for its capacity to utilize non-annotated datasets to learn meaningful data representations. In the medical domain, the challenge of constructing large annotated datasets presents a significant limitation, rendering SSL an ideal approach to address this constraint. In this study, we introduce a novel pretext task tailored to stimulus-driven eye movement data, along with a denoising task to improve the robustness against simulated eye tracking failures. Our proposed task aims to capture both the characteristics of the pilot (brain) and the motor (eye) by learning to reconstruct the eye movement position signal using up to 12.5% of the unmasked eye movement signal patches, along with the entire REMOBI target signal. Thus, the encoder learns a high-dimensional representation using a multivariate time series of length 8192 points, corresponding to approximately 40 s. We evaluate the learned representation on screening eight distinct groups of pathologies, including dyslexia, reading disorder, and attention deficit disorder, across four datasets of varying complexity and size. Furthermore, we explore various head architecture designs along with different transfer learning methods, demonstrating promising results with improvements of up to approximately 15%, leading to an overall macro F1 score of 61% and 61.5% on the Saccade and the Vergence datasets, respectively. Notably, our method achieves macro F1 scores of 64.7%, 66.1%, and 61.1% for screening dyslexia, reading disorder, and attention deficit disorder, respectively, on clinical data. These findings underscore the potential of self-learning algorithms in pathology screening, particularly in domains involving complex data such as stimulus-driven eye movement analysis.


  • Open Access

    Physiological Data Augmentation for Eye Movement Gaze in Deep Learning

    by Alae Eddine El Hmimdi, Zoï Kapoula

    In this study, the challenges posed by limited annotated medical data in the field of eye movement AI analysis are addressed through the introduction of a novel physiologically based gaze data augmentation library. Unlike traditional augmentation methods, which may introduce artifacts and alter pathological features in medical datasets, the proposed library emulates natural head movements during gaze data collection. This approach enhances sample diversity without compromising authenticity. The library evaluation was conducted on both CNN and hybrid architectures using distinct datasets, demonstrating its effectiveness in regularizing the training process and improving generalization. What is particularly noteworthy is the achievement of a macro F1 score of up to 79% when trained using the proposed augmentation (EMULATE) with the three HTCE variants. This pioneering approach leverages domain-specific knowledge to contribute to the robustness and authenticity of deep learning models in the medical domain.


  • Open Access

    Deep Learning-Based Detection of Learning Disorders on a Large Scale Dataset of Eye Movement Records

    by Alae Eddine El Hmimdi, Zoï Kapoula, Vivien Sainte Fare Garnot

    Early detection of dyslexia and learning disorders is vital for avoiding a learning disability, as well as supporting dyslexic students by tailoring academic programs to their needs. Several studies have investigated using supervised algorithms to screen dyslexia vs. control subjects; however, the data size and the conditions of data acquisition were their most significant limitation. In the current study, we leverage a large dataset, containing 4243 time series of eye movement records from children across Europe. These datasets were derived from various tests such as saccade, vergence, and reading tasks. Furthermore, our methods were evaluated with realistic test data, including real-life biases such as noise, eye tracking misalignment, and similar pathologies among non-scholar difficulty classes. In addition, we present a novel convolutional neural network architecture, adapted to our time series classification problem, that is intended to generalize on a small annotated dataset and to handle a high-resolution signal (1024 point). Our architecture achieved a precision of 80.20% and a recall of 75.1%, when trained on the vergence dataset, and a precision of 77.2% and a recall of 77.5% when trained on the saccade dataset. Finally, we performed a comparison using our ML approach, a second architecture developed for a similar problem, and two other methods that we investigated that use deep learning algorithms to predict dyslexia.


  • Open Access

    Predicting Dyslexia in Adolescents from Eye Movements during Free Painting Viewing

    by Alae Eddine El Hmimdi, Lindsey M Ward, Themis Palpanas, Vivien Sainte Fare Garnot, Zoï Kapoula

    It is known that dyslexics present eye movement abnormalities. Previously, we have shown that eye movement abnormalities during reading or during saccade and vergence testing can predict dyslexia successfully. The current study further examines this issue focusing on eye movements during free exploration of paintings; the dataset was provided by a study in our laboratory carried by Ward and Kapoula. Machine learning (ML) classifiers were applied to eye movement features extracted by the software AIDEAL: a velocity threshold analysis reporting amplitude speed and disconjugacy of horizontal saccades. In addition, a new feature was introduced that concerns only the very short periods during which the eyes were moving, one to the left the other to the right; such periods occurred mostly during fixations between saccades; we calculated a global index of the frequency of such disconjugacy segments, of their duration and their amplitude. Such continuous evaluation of disconjugacy throughout the time series of eye movements differs from the disconjugacy feature that describes inequality of the saccade amplitude between the two eyes. The results show that both AIDEAL features, and the Disconjugacy Global Index (DGI) enable successful categorization of dyslexics from non-dyslexics, at least when applying this analysis to the specific paintings used in the present study. We suggest that this high power of predictability arises from both the content of the paintings selected and the physiologic relevance of eye movement features extracted by the AIDEAL and the DGI.


  • Open Access

    Predicting Dyslexia and Reading Speed in Adolescents from Eye Movements in Reading and Non-Reading Tasks: A Machine Learning Approach

    by Alae Eddine El Hmimdi, Lindsey M Ward, Themis Palpanas, Zoï Kapoula

    There is evidence that abnormalities in eye movements exist during reading in dyslexic individuals. A few recent studies applied Machine Learning (ML) classifiers to such eye movement data to predict dyslexia. A general problem with these studies is that eye movement data sets are limited to reading saccades and fixations that are confounded by reading difficulty, e.g., it is unclear whether abnormalities are the consequence or the cause of reading difficulty. Recently, Ward and Kapoula used LED targets (with the REMOBI & AIDEAL method) to demonstrate abnormalities of large saccades and vergence eye movements in depth demonstrating intrinsic eye movement problems independent from reading in dyslexia. In another study, binocular eye movements were studied while reading two texts: one using the “Alouette” text, which has no meaning and requires word decoding, the other using a meaningful text. It was found the Alouette text exacerbates eye movement abnormalities in dyslexics. In this paper, we more precisely quantify the quality of such eye movement descriptors for dyslexia detection. We use the descriptors produced in the four different setups as input to multiple classifiers and compare their generalization performances. Our results demonstrate that eye movement data from the Alouette test predicts dyslexia with an accuracy of 81.25%; similarly, we were able to predict dyslexia with an accuracy of 81.25% when using data from saccades to LED targets on the Remobi device and 77.3% when using vergence movements to LED targets. Noticeably, eye movement data from the meaningful text produced the lowest accuracy (70.2%). In a subsequent analysis, ML algorithms were applied to predict reading speed based on eye movement descriptors extracted from the meaningful reading, then from Remobi saccade and vergence tests. Remobi vergence eye movement descriptors can predict reading speed even better than eye movement descriptors from the meaningful reading test.