With the COVID-19 situation, there is an elevated sense of anxiety around health issues within the community. As a result, hospitals, medical centres and GPs are receiving higher number of enquiry calls.
I realised that developing an automated interactive voice solution would help hospitals and doctors to avoid responding manually to enquiry calls. This solution would answer most of the COVID-19 queries and redirect to hospitals or GPs only for other critical health issues. During this thought process, I recalled an IVR(Interactive Voice Response) based solution developed for a Singapore based hospital. I developed this solution as an automated interactive interface between a patient and the medical staff, with a goal of reducing the time spent by nurses in receiving calls.
Based on this concept, I developed an Amazon Connect solution which can be handy during this unprecedented time. The outcome expected is reduction in the number of calls coming through to hospitals and GPs; reduces the stress on hospitals to manually answer calls. The solution is designed to be low-cost service underpinned by health adversary algorithms for COVID-19, capable of automated voice interactions with patients/callers and routing calls to hospitals only in-case of emergencies. The algorithm and the workflows can be easily modified as and when the requirements change.
This solution, developed using Amazon Connect, a contact centre service by AWS is easily scalable and capable of handling large volume of phone calls.
I have used:
- Amazon Lex – For conversational interfaces; provides advanced deep learning functionality for automatic speech recognition and natural language processing
- Lambda – To process the business logic
- SNS – To send text notifications to callers
The contact centre flow is not limited to medical sector and can be used across other genres of industry including telecommunication, call centres, taxation department etc. One of the challenges with this initial flow was gathering requirements on unprecedented territory. The other challenge was to feed Lex with all possible scenarios of an expected answer.
Solution flow diagram: