Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Fouad Trad has a total of 26 co-authors. Here are the top co-authors by number of shared publications:

  • Ali Chehab

    American University of Beirut

    13 shared publications
  • Ali Chehab

    Electrical and Computer Engineering Department, American University of Beirut, Beirut, Lebanon

    3 shared publications
  • Ghada El-Hajj Fuleihan

    Calcium Metabolism and Osteoporosis Program, WHO Collaborating Center for Metabolic Bone Disorders, Division of Endocrinology and Metabolism, Department of Internal Medicine, American University of Beirut Medical Center, Beirut, Lebanon

    2 shared publications
  • Ryan Yammine

    Calcium Metabolism and Osteoporosis Program, WHO Collaborating Center for Metabolic Bone Disorders, Division of Endocrinology and Metabolism, Department of Internal Medicine, American University of Beirut Medical Center, Beirut, Lebanon

    2 shared publications
  • Ali Hussein

    Electrical and Computer Engineering, American University of Beirut, Beirut 1107-2020, Lebanon

    1 shared publication
  • Ali Mustafa

    American University of Beirut

    1 shared publication

Latest Publications

  • Open Access

    Analyzing medical students’ performance through explanatory and predictive lenses: a retrospective longitudinal study (Preprint)

    by Sandrella Bou Malhab, Fouad Trad, Patrick Khalifeh, Ndia Asmar, Sola Bahous

    BACKGROUND Competency-based medical education (CBME) increasingly relies on data-driven assessment to support early identification of at-risk students, yet few studies have applied machine learning (ML) alongside traditional inferential statistics to the same longitudinal medical education dataset to compare their predictive and explanatory value. OBJECTIVE This study aimed to examine medical students' academic performance using both explanatory (inferential) and predictive (ML) approaches, to identify academic determinants of competency attainment, clarify their sequence across the curriculum, and to develop a data-driven model supporting academic monitoring and timely intervention. METHODS This retrospective study included medical students enrolled at the Lebanese American University Gilbert and Rose-Marie Chaghoury School of Medicine (LAU-SOM) from 2013 to 2025. Predictors included demographic characteristics, premedical admission metrics (grade point average, Medical College Admission Test subscores, interview score), and longitudinal knowledge and clinical-skills outcomes across four medical school years, along with competency gap-matrix classifications. Sequential multivariable linear, logistic, and multinomial regression models tested statistical relationships among variables. In parallel, nine tree-based ML algorithms were trained and compared using admission-only versus sequential predictor sets, with model selection via cross-validated R2 (regression) or weighted F1 score (classification), and interpretation via SHapley Additive exPlanations (SHAP) values. RESULTS The cohort was 50.7% male. Inferential models showed that early performance (Med 1) was predicted mainly by premedical GPA and MCAT sub-scores, while performance in later years was increasingly predicted by prior in-program academic outcomes, particularly Med 2 IFOM and OSCE scores. ML models mirrored this pattern: admission-only models showed moderate accuracy for Med 1–2 outcomes (test R2 up to 0.603) but weak performance for Med 3–4 outcomes (test R2 as low as 0.064). Incorporating sequential data improved prediction, raising test R2 to 0.328–0.831 across Med 2–4 outcomes, with the largest gain for Med 3 knowledge performance (R2 from 0.152 to 0.672). Sequential predictors similarly improved binary gap-status classification, increasing weighted F1 score from 0.632 to 0.773 and accuracy from 0.629 to 0.776. CONCLUSIONS Integrating longitudinal, sequentially updated academic performance data markedly improves both statistical explanation and machine-learning prediction of student outcomes and competency gaps. This dual explanatory-predictive framework offers a practical, data-driven early-warning system that can be embedded into digital learning-management platforms to support real-time academic monitoring and timely intervention within CBME

  • Open Access

    CLASP: Cost-Optimized LLM-based Agentic System for Phishing Detection

    by Fouad Trad, Ali Chehab

    Phishing websites remain a significant cybersecurity threat, necessitating accurate and cost-effective detection mechanisms. In this paper, we present CLASP, a novel system that effectively identifies phishing websites by leveraging multiple intelligent agents, built using large language models (LLMs), to analyze different aspects of a web resource. The system processes URLs or QR codes, employing specialized LLM-based agents that evaluate the URL structure, webpage screenshot, and HTML content to predict potential phishing threats. To optimize performance while minimizing operational costs, we experimented with multiple combination strategies for agent-based analysis, ultimately designing a strategic combination that ensures the per-website evaluation expense remains minimal without compromising detection accuracy. We tested various LLMs, including Gemini 1.5 Flash and GPT-4o mini, to build these agents and found that Gemini 1.5 Flash achieved the best performance with an F1 score of 83.01% on a newly curated dataset. Also, the system maintained an average processing time of 2.78 seconds per website and an API cost of around $3.18 per 1,000 websites. Moreover, CLASP surpasses leading previous solutions, achieving over 40% higher recall and a 20% improvement in F1 score for phishing detection on the collected dataset. To support further research, we have made our dataset publicly available, supporting the development of more advanced phishing detection systems.

  • Open Access

    Evaluating the Efficacy of Prompt-Engineered Large Multimodal Models versus Fine-Tuned Vision Transformers in Image-Based Security Applications

    by Fouad Trad, Ali Chehab

    The success of Large Language Models (LLMs) has spurred the rise of Large Multimodal Models (LMMs), which integrate multiple modalities, such as text and images, to address complex data analysis tasks. As these black-box models gain popularity due to their ease of use and adaptability, there is growing interest in understanding their potential to replace or complement task-specific models in domain-specific applications. This article evaluates the applicability and effectiveness of prompt-engineered LMMs, specifically LLaVA, BakLLaVA, Moondream, Gemini 1.5 Flash, and GPT-4o, compared to fine-tuned Vision Transformer (ViT) models in addressing cybersecurity challenges of varying complexity. Our study examines three distinct tasks: (1) detecting visual triggers indicative of potential backdoors in two scenarios (digit recognition and traffic sign classification), (2) identifying phishing attempts from Web site screenshots, and (3) classifying malware based on visual representations. The results reveal that prompt-engineered LMMs perform competitively on tasks with visually evident or moderately complex features, such as trigger detection and phishing classification, with GPT-4o and Gemini 1.5 Flash demonstrating superior performance among LMMs. However, for the highly specialized task of malware classification, LMMs exhibit notable limitations when relying solely on prompting. Fine-tuning GPT-4o significantly improves its performance, yet it still lags behind fine-tuned ViT models, which consistently achieve higher accuracy across all tasks. While ViTs deliver superior precision and robustness, they require substantial resources for training, fine-tuning, and maintenance. In contrast, LMMs provide flexibility and ease of deployment, making them an appealing alternative for scenarios where resource constraints or rapid implementation are critical. This study highlights the tradeoffs between these approaches, emphasizing that while ViTs are indispensable for high precision, specialized applications, LMMs offer a scalable and versatile solution for less complex or resource-limited tasks.


  • Open Access

    Stacking Large Language Models is All You Need: A Case Study on Phishing Url Detection

    by Hawraa Nasser, Fouad Trad, Ali Chehab

    Prompt-engineered Large Language Models (LLMs) have gained widespread adoption across various applications due to their ability to perform complex tasks without requiring additional training. Despite their impressive performance, there is considerable scope for improvement, particularly in addressing the limitations of individual models. One promising avenue is the use of ensemble learning strategies, which combine the strengths of multiple models to enhance overall performance. In this study, we investigate the effectiveness of stacking ensemble techniques for chat-based LLMs in text classification tasks, with a focus on phishing URL detection. Notably, we introduce and evaluate three stacking methods: (1) prompt-based stacking, which uses multiple prompts to generate diverse responses from a single LLM; (2) model-based stacking, which combines responses from multiple LLMs using a unified prompt; (3) hybrid stacking, which integrates the first two approaches by employing multiple prompts across different LLMs to generate responses. For each of these methods, we explore meta-learners of varying complexities, ranging from Logistic Regression to BERT. Additionally, we investigate the impact of including the input text as a feature for the meta-learner. Our results demonstrate that stacking ensembles consistently outperform individual models, achieving superior performance with minimal training and computational overhead. These findings highlight the potential of stacking ensembles in mitigating the limitations of existing methods and significantly enhancing the efficiency and accuracy of chat-based LLMs for text classification tasks.


  • Open Access

    Parsimonious and explainable machine learning for predicting mortality in patients post hip fracture surgery

    by Fouad Trad, Bassel Isber, Ryan Yammine, Khaled Hatoum, Dana Obeid, Mohammad Chahine, Rachid Haidar, Ghada El-Hajj Fuleihan, Ali Chehab

    Hip fractures among the elderly population continue to present significant risks and high mortality rates despite advancements in surgical procedures. In this study, we developed machine learning (ML) algorithms to estimate 30-day mortality risk post-hip fracture surgery in the elderly using data from the National Surgical Quality Improvement Program (NSQIP 2012–2017, n = 62,492 patients). Our approach involves two models: one estimating the patients’ 30-day mortality risk based on pre-operative conditions, and another considering both pre-operative and post-operative factors. We performed comprehensive data cleaning and preprocessing, then applied tenfold cross-validation with randomized search to the training set to identify optimal hyperparameters for various machine learning models. We used logistic regression, Naive Bayes, random forest, AdaBoost, XGBoost, CatBoost, Gradient Boosting, and LightGBM. The models’ performances were evaluated on the test set using the Area Under the Receiver Operating Characteristic Curve (AUC). The best pre-operative model was AdaBoost, achieving an AUC of 0.792 with 29 features (predictors), and the best post-operative model was CatBoost, achieving an AUC of 0.885 with 45 features. After modeling, we derived feature importance for each of the two models and decreased the number of features to reach a parsimonious highly performing model. The pre-operative model achieves an AUC of 0.725 with the eight most important features and the post-operative model achieves an AUC of 0.8529 with the six most important features. To ensure the models’ decision-making is compatible with clinical decisions and common practices, we applied explainability techniques such as SHAP to reveal the patterns learned by the models. These patterns were found to be clinically plausible. In summary, our approach involving data preprocessing, model tuning, feature selection, and explainability achieved state-of-the-art performance in predicting 30-day mortality rates following hip fractures surgery using a limited set of features, making it highly applicable in clinical settings.


  • Open Access

    Streamlining systematic reviews with large language models using prompt engineering and retrieval augmented generation

    by Fouad Trad, Ryan Yammine, Jana Charafeddine, Marlene Chakhtoura, Maya Rahme, Ghada El-Hajj Fuleihan, Ali Chehab

    Systematic reviews (SRs) are essential to formulate evidence-based guidelines but require time-consuming and costly literature screening. Large Language Models (LLMs) can be a powerful tool to expedite SRs. We conducted a comparative study to evaluate the performance of a commercial tool, Rayyan, and an in-house LLM-based system in automating the screening of a completed SR on Vitamin D and falls. The SR retrieved 14,439 articles, and Rayyan was trained with 2,000 manually screened articles to categorize the rest as most likely to exclude/include, likely to exclude/include and undecided. We analyzed Rayyan’s title/abstract screening performance using different inclusion thresholds. For the LLM, we used prompt engineering for title/abstract screening and Retrieval-Augmented Generation (RAG) for full-text screening. We evaluated performance using article exclusion rate (AER), false negative rate (FNR), specificity, positive predictive value (PPV), and negative predictive value (NPV). Additionally, we compared the time required to complete screening steps of the SR using both approaches against the manual screening method. Using Rayyan, including considered as undecided or likely to include for title/abstract screening resulted in an AER of 72.1% and an FNR of 5%. The total estimated screening time, including manual review of articles flagged by Rayyan, was 54.7 hours. Lowering the Rayyan threshold to ‘likely to exclude’ reduced the FNR to 0% and the AER to 50.7%, but increased the screening time to 81.3 h. Using the LLM system, after title/abstract and full-text screening, 78 articles remained for manual review, including all 20 identified by traditional methods. The LLM achieved an AER of 99.5%, specificity of 99.6%, PPV of 25.6%, and NPV of 100%, with a total screening time of 25.5 h, including manual review of the 78 articles, reducing the manual screening time by 95.5%. The LLM-based system significantly enhances SR efficiency, compared to manual methods and Rayyan while maintaining low FNR.