Telemedicine is revolutionizing healthcare delivery, especially in remote areas and situations where face-to-face consultations are infeasible. However, as the adoption of digital health solutions accelerates, it brings forth challenges related to the authenticity of remote sessions. A significant threat in recent years has been spoofing attacks, such as photo impersonations or video replays, which may threaten the authenticity of remote interactions. It is even more difficult for healthcare professionals to gauge a patient’s emotional well-being in the absence of physical signals. This can lead them to miss the subtle signs of distress or anxiety. To address these multifaceted issues, this paper introduces a novel framework for real-time liveliness detection and mood analysis. By merging the capabilities of the YOLO (you only look once) v8 model enhanced with spatial attention mechanism (SAM) and frequency analysis (FA) using fast Fourier transform (FFT) integrated with a deep convolutional neural network (CNN). Preliminary findings, showcased through real-time emotion tracking and comprehensive emotion distribution charts, highlight the unique ability of our system to accurately measure emotions in real time. This not only offers profound insights into a patient’s mental state but also paves the way for proactive healthcare interventions in telemedicine.