Sahyogi-Breaking Barriers With Signs Of Care

The implementation of the Sahyogi platform yielded strong and consistent performance across all major functional mod- ules. The real-time video consultation feature, powered by WebRTC, demonstrated exceptional stability and low latency during tests. Even under varying bandwidth conditions, the system maintained smooth video streaming and rapid frame transmission, allowing uninterrupted communication between patients and doctors. This reliability is essential in telehealth scenarios where visual cues and real-time interaction are critical for proper diagnosis. The speech-to-text component performed with high accu- racy, effectively converting the doctor’s spoken words into readable text. This proved particularly beneficial for hearing- impaired users, who were able to follow conversations without external assistance. The textual output remained clear even in the presence of mild background noise, indicating the robustness of the transcription module. The sign-language–to-text model also showed promising outcomes. Using an LSTM-based deep learning architecture, the system accurately recognized gestures within the trained Indian Sign Language (ISL) vocabulary. During evaluation, the model responded well to variations in hand speed, orientation, and lighting, maintaining reliable classification across diverse user inputs. These results confirm the feasibility of real-time gesture interpretation within live teleconsultations. Additionally, the integrated prescription module success- fully generated digital and graphical prescriptions after each completed session. Doctors were able to provide clearer in- structions using visual icons, which improved comprehension for elderly patients or individuals with low literacy levels. The backend storage system (MongoDB) efficiently handled data retrieval and updates, maintaining system responsiveness throughout all operations. Overall, the results demonstrate that Sahyogi functions as a complete, end-to-end telehealth plat- form capable of supporting inclusive and accessible medical communication.

[1]     Liang, X., Zhang, Y., Wang, L. (2023). Real-time sign language translation using deep learning for telemedicine applications. Journal of Healthcare Tech- nology, 15(2), 45-58. [2]     Kumar, P.,  Singh, S. (2022). Telehealth solu- tions for patients with speech and hearing impair- ments: A review of current technologies. Interna- tional Journal of Medical Informatics, 160, 13-20. doi: 10.1016/j.ijmedinf.2021.104294. [3]     Zhao, T., Lee, H. (2021). Multilingual speech transcrip- tion for healthcare communication: Bridging language barriers in telemedicine. Journal of Digital Health, 9(3), 112-120. [4]     Patel, A., Mehta, R. (2020). Enhancing healthcare ac- cessibility with AI-driven communication tools. Journal of Medical AI, 8(4), 65-75. [5]     Sharma, R., Kapoor, V. (2023). Deep learning ar- chitectures for Indian Sign Language recognition: A comprehensive survey. International Journal of Pattern Recognition and AI, 37(1), 1–18. [6]     Thomas, A., Bhat, S. (2022). Real-time gesture recogni- tion using LSTM networks for assistive communication systems. IEEE Transactions on Human-Machine Sys- tems, 52(4), 789–798. [7]     Chakraborty, M., Rajan, P. (2021). WebRTC-based telemedicine platforms: Challenges and opportunities for remote healthcare delivery. Journal of Telehealth and Virtual Care, 6(2), 101–110. [8]     Verma, D., Pillai, R. (2022). Speech-to-text systems in virtual healthcare: Improving accessibility for pa- tients with auditory impairments. IEEE Access, 10, 120345–120356. [9]     Singh, A., Rao, K. (2020). Machine-learning–enabled assistive technologies for patients with disabilities: A systematic review. ACM Computing Surveys, 53(5), 1– 28. [10]     Gupta, L., Abraham, A. (2023). Real-time sign gesture detection under variable lighting conditions for mobile healthcare applications. Journal of Ambient Intelligence and Humanized Computing, 14(3), 1767–1778.