Key takeaways
- Eka Care partners with NVIDIA to create India's first offline-capable medical scribe using AI technology.
- Develops a multilingual scribe model catering to India's diverse languages and accents for better clinical documentation.
- Focuses on secure real-time documentation, minimizing reliance on cloud connectivity to enhance data privacy.
- Utilizes NVIDIA's AI software to refine Indian medical conversation datasets, improving accuracy and performance.
- Evaluates advanced speech recognition models to integrate with language processing for streamlined clinical workflows.
Bengaluru-based health AI platform Eka Care has announced a collaboration with NVIDIA to develop what it calls India’s first offline-capable, unified medical scribe model. It aims to enable secure, real-time clinical documentation even in low-connectivity settings.
The initiative will focus on building an end-to-end, multilingual medical scribe tailored to India’s diverse linguistic and accent landscape. A variant of the model is being designed to run entirely offline on a doctor’s device, reducing dependence on cloud connectivity and addressing concerns around data privacy and latency.
As part of the collaboration, Eka Care will leverage NVIDIA’s AI software stack and compute infrastructure to refine large datasets of Indian medical conversations. The company is using NVIDIA NeMo Curator to clean and structure clinical data and integrating the NVIDIA Nemotron-CC dataset during pre-training to prevent “catastrophic forgetting”, where AI models lose previously learned knowledge while adapting to new tasks.
Eka Care is also evaluating NVIDIA Nemotron Speech models, including Parakeet automated speech recognition, with the goal of merging speech recognition and large language model capabilities into a single architecture. The company said this approach aims to reduce computational overhead while maintaining clinical accuracy.
“Building a medical scribe for India requires navigating complex code-mixing and diverse accents. By collaborating with NVIDIA, we aren’t just improving our accuracy; we are building a sovereignty-first model that can live on a doctor's mobile, independent of cloud connectivity. This is the future of secure, private patient documentation,” said Vikalp Sahni, Founder and CEO of Eka Care.
Tobias Halloran, Director of EMEAI Startups and Venture Capital at NVIDIA, said India’s AI startup ecosystem is gaining momentum, supported by access to accelerated computing, scalable AI infrastructure, and programmes such as NVIDIA Inception and the NVIDIA VC Alliance.
