InjazAI

Researchers developed the DeepLens Diagnosis Agent, a five-stage agentic workflow that achieves 60.14% diagnostic accuracy at 35-45% lower cost than frontier models like Claude Sonnet 4.5 and Gemini 3.1 Pro, making it suitable for high-stakes environments in MENA.

1 min read

DeepLens Diagnosis Agent: Agentic Workflow Achieves 60% Accuracy at Lower Cost Than Frontier LLMs

DeepLens Diagnosis Agent: Revolutionizing Medical Diagnosis at Low Cost

FAQ

What is the DeepLens Diagnosis Agent?

It is a five-stage agentic system using a small medical model (JSL Medical Small 7B v2) with RAG to improve diagnostic accuracy at low cost.

How does DeepLens compare to frontier models like Claude Sonnet 4.5?

DeepLens outperforms Claude Sonnet 4.5 by 9.70 percentage points in accuracy while being 35-45% cheaper.

Can DeepLens be used in MENA healthcare?

Yes, due to its low cost, high accuracy, and transparency, it is suitable for hospitals and clinics in the region.

What are the advantages of the agentic workflow over single-shot prompting?

The agentic workflow improves accuracy by 36 percentage points and provides auditable intermediate outputs, facilitating error localization.

Source: arXiv cs.AI

AI-assisted content, human-reviewed.