Becoming the Benchmark for Healthcare AI

🚀 Advancing Clinical Speech Recognition: Setting a New Bar for Enterprise-grade Ambient AI in Healthcare. 

Following the latest round of improvements for Abridge AI models, here are some key comparisons against the next-best, state-of-the-art commercial speech recognition systems:

• 83% relative reduction in transcription errors on new medications
• 24% relative reduction in word error rate on clinical conversations
• 15% relative improvement in transcription accuracy for accented English 

One year ago, we published Pioneering the Science of AI Evaluation, a groundbreaking research report that showcased the deep science behind Abridge’s standard-setting automatic speech recognition and note generation models. 

Today, we’re raising the bar once again with substantial improvements that widen the gap between our system and the leading industry standard across key metrics like medical term recall, word error rate, and accent robustness.

Read our updated whitepaper for more details about these exciting updates.