# Cosmos Hackathon
----

There we go!!!

## Frontend (Mizso)

* App/web app for users to report their symptoms as well as health experts to get notifications of outbreaks.
* Symptoms reported by the user: fever, cough, Runny Nose, Fatigue
* Other symptoms: Shortness of breath, Headache, Difficulty in breathing, Diarrhea
* Location options: Bluetooth, Wi-Fi towers, evil GPS.

## Backend (Alok and Olga)

**Input/Output**
For each normal user we expect to receive a message from the web/app containing the following information: list of symptoms, measurements, residential postal code, location (?)

For expert use we would like to be able to send a data/image of the heatmap of "hotspots" for established cases.

**Machine Learning**

The magic of RNNs

**Infrastructure**
* *Done* Created Kubernetes cluster on GCP (europe-west-b1) for the endpoints. Explored use of Openshift however no external access given at the moment. Deployed a trial for ElasticSearch (Kibana and all) for the data backends. Endpoints now availabe.
* *In progress* Looking to push some synthetic data into ES.

**Basic API Backend**
* *In progress* Plug and play containers for the cosmos service: deployment for cosmos API and possibility to add ML containers as needed. Helm chart coming up.
* *In progress* Brushing up on API development to handle data to/from the app.
* *Wishful thinking* It would be interesting to go full on serverless here.

## Clinical/patient data (Rucha): in-progress
data of patients reporting Covid outcomes: already one observation study found
I will make a list and will try to classify it according to severity (to align with the idea of low, moderate, severe diagnosed by the app)

## Generate random sample from predictions: monte-carlo/bootstrap: in-progress
checking the possibilities to assign the prediction of symptoms to randomly generated sample: to prepare training data-set

## mentor preference (Rucha): done including legal
will look and send messages to mentors

## postal codes (Rucha): not available: possibly in Google API in-built
will look

##

## Questions to the 1st Mentor Talk (5:00 pm Zoom in Slack)

* **Privacy Concerns**: Are there limitations on technology use with respect to HIPAA-type compliance for medical/personal data? Jurisdiction requirements for the data held in different regions i.e. EU citisen data on US servers.
* How can we fix the issue of asking for permission with the blutooth access.
* Should we treat the same user who is loging in next day as a new user or we should create a profile for each user seperately.

**Followup thoughts**
* Look into motivation for encouraging user participation: gamefication or social participation.
* Applicability of the Grid model for establishing Trust between users and authorities.

## Marketing development questions
* What details to include in the pitch deck.
* How should we put it so that it looks different and not like every other app.
*