VOICE Summit 2019 | VED Talks | Harjinder Sandhu | July 24
summary
Claims that conversational agents fail for clinical documentation and that passive listening systems are the viable path forward evidence a concrete trajectory. A speech recognition pipeline trained on millions of physician dictations boosts accuracy from about 20% to the high 90s. EMR workflows force clinicians to spend one hour with a patient and two hours on screens. Early products called AI scribes evolve into AI clinical assistants that prep charts, surface care gaps like colonoscopy or osteoporosis screening, and begin to work in pilot deployments.
AI-written summary · about