We were entrusted with the implementation of an ambitious and responsible project - to automate the work of emergency medical call centers by introducing an audio bot capable of conducting a dialogue with a patient and making a decision on calling an ambulance team.
The first stage of the study was the analysis of speech recognition and generation models. Based on a huge amount of open data and machine learning approaches, we formed an effective model that gave better results on our lexical base than standard solutions from AI solution providers.
Novoskor AI audio bot was fully integrated into the call center's IP telephony system, our specialists designed a simple and user-friendly flow of communication.
Our machine learning team used Python.
For integration into the VoIP system, we also made scripts on PHP.
Our solution was actually implemented in an ambulance call center and has shown its effectiveness in significantly reducing the workload on operators without losing the quality of patient care.