Fractal Unveils India’s First Multimodal Medical AI Assistant Vaidya

Fractal launches Vaidya.ai, India’s first multimodal medical AI assistant, designed to aid in diagnosis and treatment, and fine-tuned with 650k+ images and 200k+ text inputs across multiple languages.

Key Points

• Fractal has introduced Vaidya.ai, India’s first multimodal medical AI assistant, aimed at supporting users with diagnosis, treatment, and medical advice.
• Vaidya is fine-tuned with open-source models of 30-70 billion parameters, trained on over 650,000 images and 200,000 text inputs, and understands multiple Indian languages.
• The AI is not intended to replace doctors but to provide reliable healthcare information to users, complementing professional medical advice.
• Vaidya’s responses have been evaluated using PG NEET exam questions, achieving an accuracy level comparable to the top 500 doctors in the exam with 83% accuracy.
• Vaidya.ai is currently an AI research project by Fractal and could potentially evolve into a one-of-its-kind enterprise healthcare solution alongside their existing product, Qure.ai.

In a groundbreaking move, Fractal, an AI-driven firm, has unveiled Vaidya.ai, India’s first multimodal medical Large Language Model (LLM). Designed to assist users with diagnosis, treatment, and medical advice, Vaidya is positioned as a general-purpose medical assistant that bridges the gap between consumers and reliable healthcare information.

Co-founder and CEO Srikanth Velamakanni explained that Vaidya is fine-tuned using open-source base models that range between 30 and 70 billion parameters. The model has been trained on a vast dataset of over 650,000 images and 200,000 text inputs, making it capable of understanding multiple Indian languages such as Hindi, Kannada, Tamil, and Telugu.

Despite the advanced capabilities of Vaidya, Velamakanni emphasized that the AI is not intended to replace medical professionals. “It’s not even in the foreseeable future that we will have a situation where we don’t need doctors and AI may tell you what to do. AI is far from that. The problem we are solving is very different. It’s about aiding consumers in accessing healthcare information which is more reliable than any other sources of information,” he said.

Vaidya.ai complements Fractal’s existing healthcare AI product, Qure.ai, which is already a mature player in detecting conditions like tuberculosis, lung cancer, heart failure, and stroke. While Vaidya has the potential to become a unique enterprise solution in the healthcare industry, it is currently being developed as an AI research project rather than a commercial product.

The accuracy of Vaidya’s responses has been rigorously evaluated using PG NEET exam questions, the most competitive medical exam in India, with around 200,000 postgraduate doctors participating annually. Velamakanni noted that Vaidya currently ranks on par with the top 500 participants in this system, achieving an impressive 83% accuracy rate.

Suraj Amonkar, Chief AI Research and Platforms Officer at Fractal, highlighted the extensive development process behind Vaidya. “The development involved training models and frameworks using LLM and VLM architectures and validating the models on 20+ years of PG NEET/ PG Medical examination papers, among other metrics to measure how well the models performed,” he said. Amonkar also emphasized that while Vaidya offers valuable insights, it always encourages users to consult healthcare professionals for personalized medical advice.

Vaidya.ai represents a significant step forward in the integration of AI within the medical field, offering a blend of cutting-edge technology and accessible healthcare information. As Fractal continues to innovate, Vaidya.ai could pave the way for more advanced AI-driven solutions that support both medical professionals and the general public in making informed healthcare decisions.