HTN project for extracting and summarizing symptoms from doctor-patient interactions
Nobody enjoys waiting in the ER. Especially if it's a life-threatening condition.
When patients arrive to the ER, they have to first get screened by the triage nurse and then are examined by a licensed medical provider to determine the level and urgency of care they need. However in high stake scenarios, patients are in panic mode and struggle the communicate their story to medical responders, resulting in being put in long wait time queues.
With CondensER, we bridge the gaps of communication between first medical responders and patients in the ER, so that doctors can get them the help they need faster and reduce patient wait times.
Using voiceflow, conversations are recorded between the nurse and the patient, in which important medical conditions and symptoms are extracted using AI from the audio and summarized concisely for the doctor. By doing so, the doctor can diagnose their symptoms and find them the treatment they need.
CondensER was built using Cohere API and Assembly AI API
Our model has a 70% efficiency in extracting text from medical information.