CARA: healthcare resource demand forecasting and management

Landing Page
Menu
Homescreen
Input Parameters
Calculation Models_Austria
Calculation Models_International
Forecast of staffing demand
Forecasted patients
Expected Material Usage
Infrastructure required for ventilation of COVID-19 patients
Monitoring Beds/infrastructure for COVID-19
Monitoring demand of staffing COVID-19_General Ward
Monitoring_Staffing_ICU
Monitoring demand of staffing COVID-19_Overview
Monitoring for infrastructure/ventilation units
Projected course of staffing demand
Specific tasks for COVID patients clustered by severity_I
Specific tasks for COVID patients clustered by severity_II

Inspiration

Since no situation like this pandemic ever occured before, factors like high infectivity, exponential growth of case numbers and ununsually high demand of ventilation units combined with lack of expierience and empirical data led to a high amount of uncertainty regarding to amount and time of resources needed in acute care. As a team of interdisciplinary experts we wanted to tackle this challenge an offer a conveniant scalable solution.

What it does

We offer a cloud-based service that predicts staffing, infrastructure, and material demand in healthcare due to the Covid-19 crisis. It includes allocation and monitoring of resources in various treatment stages and categories of severity. Our predictions are based on national outbreak models by Oxford University and can be used at different organizational levels, from hospitals, healthcare providers, to states and nations.

How I built it

Hierarchical Bayesian inference framework for projecting positive case numbers in Austria, and Austrian federals states as well as a selection of countries, based on the assumption of logistic growth.
Normalization to expected patient arrivals and residency, based on user input, into different patient severity groups, requiring different treatment efforts and resources.
Translation of patient arrivals and residency with our team's key knowledge of Covid-19 related task-efforts and material demand, to projections of additional staffing, material and infrastructure requirements with uncertainties.
Embedding in a highly accessible and scalable infrastructure service with docker swarm hosted on Google Cloud Platform

Challenges I ran into

Linking different data sources and models

Accomplishments that I'm proud of

Since we started at the Hackthecrisis Hackathon in Austria some weeks ago we have reached users in 12 countries across 4 contintens as well as a collaboration with University Oxford.