Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{210790,
        author = {Manjusha Rajendra Madas and Rudrani Sharad Patre and Sakshi Arun Rajole and Shreya Shriram Kadam},
        title = {A Multi-Modal AI Digital Twin for Personalized Cognitive Well-Being Prediction},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {6},
        pages = {2081-2088},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=210790},
        abstract = {Cognitive well-being is dynamic and multidimensional, involving stress, affect, workload, sleep-related behaviour and subjective experience. Existing computational approaches often rely on a single sensing modality or population-level models that do not adequately represent individual differences. This paper develops a multimodal artificial-intelligence (AI) digital-twin framework for personalized cognitive well-being monitoring. The framework is designed around five publicly documented secondary datasets: WESAD, StudentLife, SWELL-KW, DEAP and DREAMER. Together, these resources provide complementary physiological, behavioural, task-related and self-reported observations. Because the datasets were collected from different cohorts and under different protocols, they are treated as separate benchmark domains rather than merged into a fictitious common cohort. Dataset-specific encoders and task-appropriate labels feed a modality-aware fusion strategy with explicit availability masks. A global model is followed by subject-level calibration when sufficient individual history is available. The manuscript specifies preprocessing, subject-independent validation, baseline models, ablation experiments and reporting metrics. The dataset review establishes the feasibility and limitations of the proposed multi-domain design; however, no new model-performance values are claimed because the experiments have not yet been executed. The framework is intended as a research and monitoring aid, not a clinical diagnostic system.},
        keywords = {Human Digital Twin; Cognitive Well-Being; Multimodal Learning; Stress Recognition; Affective Computing; Personalization; Missing-Modality Learning; Wearable Sensing.},
        month = {November},
        }