The challenge, proposed by Zeiss, involves creating a patient-doctor-workflow incorporating Amazon Alexa and Amazon Web Services. The basic idea is that a patient can take pictures of their eyes and send them off to a medical professional for review.
Our project mainly consists of four components.
DocConnect Alexa Skill
The DocConnect Skill guides patients through the registration process and is used to initiate the photo upload on Telegram. It was created in the Alexa Developer Console and uses an AWS Lambda Funciton as its endpoint.
DocConnect Telegram Bot
The Telegram bot is what allows users to send photos to doctors. It's configured to receive updates via a WebHook which invokes a Lambda function over AWS API Gateway.
PatientConnect Alexa Skill
The PatientConnect Skill allows doctors to create patient records in the database and view patient images.
picCheck Lambda Function
The picCheck Lambda function processes incoming images, which allows a rudimentary assessment of quality. This is accomplished by detecting the presence of two eyes, making sure they cover a large enough portion of the image and calculating the average brightness in the area. In a real application, these metrics could be used to filter out images of insufficient quality before they are even stored.
Challenges we ran into
- The Echo Show camera can't be accessed by Alexa Skills
- The Amazon Drive API is invite only
What we learned
- Setting up Alexa-Skills and connecting them to Amazon-Lambda functions to create responses
- Using visual exponents in Alexa-Skills (especially with EchoShow)
- Work with Amazon-Lambda: Create and communicate between lambda functions (invokes)
- Digital image processing with the open source software "Open CV" using python
- Eye recogintion using haar cascades
- Interacting with the AWS Services DynamoDB and S3 buckets
- Creating telegram bots for image upload and interaction with the DocConnect system
- Working in a team
Log in or sign up for Devpost to join the conversation.