Portrait of Karim Loulou is unavailable

Karim Loulou

karim.loulou@mila.quebec
Professional Master's - Université de Montréal

Publications

Complexity-driven feature selection for enhancing tuberculosis detection
Sana Ben Mahjouba
Johannes C. Ayena
Youssef Ouakrim
Karim Loulou
Simon Grandjean Lapierre
Mihaja Raberahona
Neila Mezghani
Most existing machine learning approaches for tuberculosis (TB) screening typically utilize large, high-dimensional acoustic feature sets wi… (see more)thout examining their intrinsic discriminative power. To address this gap, we introduce a complexity-based feature selection approach that evaluates temporal and spectral descriptors using Fisher score (F1), class-distribution overlap (F2), and Shannon entropy (F4). Applied to the CODA-TB dataset (9772 audio recordings from 1105 participants), the proposed method identified 7 highly informative features from the original 26 features, primarily consisting of mel-frequency cepstral coefficients (MFCC) derivatives and spectral-shape measures. The resulting model achieved performance comparable to full-feature baselines while reducing feature dimensionality by 73% and computational cost by up to 14×. Comparative evaluation against four established feature selection techniques, supported by ablation and statistical analyses, confirmed the efficiency and robustness of the complexity-driven strategy, with no statistically significant loss in performance. These findings highlight the potential of lightweight, interpretable, and computationally efficient models for TB cough-based screening in resource-constrained environments.
2026-09-01
Franklin Open (published)
doi.org
Predicting chronic pain using wearable devices: a scoping review of sensor capabilities, data security, and standards compliance
Johannes C. Ayena
Amina Bouayed
Myriam Ben Arous
Youssef Ouakrim
Karim Loulou
Darine Ameyed
Isabelle Savard
Leila El Kamel
Neila Mezghani
Background: Wearable devices offer innovative solutions for chronic pain (CP) management by enabling real-time monitoring and personalized p… (see more)ain control. Although they are increasingly used to monitor pain-related parameters, their potential for predicting CP progression remains underutilized. Current studies focus mainly on correlations between data and pain levels, but rarely use this information for accurate prediction. Objective: This study aims to review recent advancements in wearable technology for CP management, emphasizing the integration of multimodal data, sensor quality, compliance with data security standards, and the effectiveness of predictive models in identifying CP episodes. Methods: A systematic search across six major databases identified studies evaluating wearable devices designed to collect pain-related parameters and predict CP. Data extraction focused on device types, sensor quality, compliance with health standards, and the predictive algorithms employed. Results: Wearable devices show promise in correlating physiological markers with CP, but few studies integrate predictive models. Random Forest and multilevel models have demonstrated consistent performance, while advanced models like Convolutional Neural Network-Long Short-Term Memory have faced challenges with data quality and computational demands. Despite compliance with regulations like General Data Protection Regulation and ISO standards, data security and privacy concerns persist. Additionally, the integration of multimodal data, including physiological, psychological, and demographic factors, remains underexplored, presenting an opportunity to improve prediction accuracy. Conclusions: Future research should prioritize developing robust predictive models, standardizing data protocols, and addressing security and privacy concerns to maximize wearable devices' potential in CP management. Enhancing real-time capabilities and fostering interdisciplinary collaborations will improve clinical applicability, enabling personalized and preventive pain management.
2025-05-21
Frontiers in Digital Health (published)
doi.org