Review on Application of Machine Learning in Detecting Suicidal Ideation for Social Media Users

In recent years, social media platforms have emerged as a novel domain for individuals to express their emotions,  including suicidal ideation, attempts, and behaviors. Consequently, these platforms have evolved into crucial data reposi-  tories and essential assessment criteria for detecting suicidal ideation. With the advent of artificial intelligence technology, the utilization of machine learning in detecting suicidal ideation among social media users has emerged as a scintillating subject. However, in China, the relative research is scarce and has not yet established a comprehensive system. To systematically review the research status and development context of suicide ideation detection, this paper presents a comprehensive summary of machine learning technology in empowering suicide ideation detection. Firstly, this paper provides an overview of the definition, process, commonly employed methods, and evaluation indicators for detecting suicidal ideation. Secondly, this paper provides a comprehensive overview of suicide ideation detection techniques, encompassing both traditional machine learning and deep learning approaches. The key methodologies, fundamental concepts, merits, and limitations of each method are thoroughly compared and analyzed. Furthermore, the urgent issues and innovative solutions in this field are summarized, with a particular focus on the application of large language models such as ChatGPT and multi-modal models. Finally, the limitations of machine learning in the application research of suicidal ideation detection on social media are comprehensively discussed, and future research directions are proposed, in order to further promote the formation of a new paradigm of data-driven, human-computer collaboration, interdisciplinary integration, and cross-cultural domain of suicide ideation detection.