Published Paper Details:

This paper presents a bibliometric analysis of the research literature underpinning a proposed AI-driven, deep-learning-based multimodal emotion-recognition framework for stress and depression detection. Rather than drawing on a full commercial citation database such as Scopus or Web of Science, this study performs a structured bibliometric profiling of the curated corpus of sixty-six (66) publications compiled through targeted literature search for the associated PhD research synopsis, spanning the period 1998-2022. The analysis characterizes the corpus along five dimensions: (i) temporal distribution and growth trend of publications; (ii) distribution by publication type (journal, conference, workshop, dataset repository); (iii) leading publication venues; (iv) distribution by modality and topical focus (facial, speech, text, EEG, physiological, multimodal fusion, benchmark dataset, and survey literature); and (v) keyword frequency derived from publication titles, used as a proxy for thematic emphasis. The results show a pronounced growth in publication volume between 2017 and 2019, accounting for 42.4% of the reviewed corpus, coinciding with the broader adoption of deep-learning architectures in affective computing. Facial-expression research and general multimodal-fusion research jointly account for 48.5% of the corpus, while EEG-inclusive multimodal studies, entropy-based fusion approaches, and metaheuristic-optimization-guided fusion strategies