Symptom-Based Classification of Common Syndromes Using Machine Learning: A Review

Symptom-based disease classification has emerged as a critical application of machine learning (ML) and artificial intelligence (AI) in modern healthcare. With the growing availability of electronic health records, patient-reported outcomes, and conversational health systems, automated interpretation of symptoms offers an efficient approach for early diagnosis and clinical decision support. This review systematically analyzes recent advancements in symptom-based classification of common syndromes using traditional machine learning, deep learning, natural language processing (NLP), and large language models (LLMs). The study highlights methods that process structured symptom vectors as well as unstructured free-text symptom descriptions obtained from chatbots and clinical narratives. A comprehensive comparison of twenty recent studies published between 2024 and 2025 is presented, focusing on employed methodologies, key advantages, and reported limitations. Furthermore, the review synthesizes major research findings, identifies open challenges such as data sparsity, explainability, and clinical reliability, and discusses future research directions. This paper aims to serve as a consolidated reference for researchers and practitioners working on intelligent symptom-based disease prediction systems and to guide future development of robust, interpretable, and clinically deployable models.