The first speaker is Prof. Dimitris Bertsimas (MIT, Cambridge) who will present the ongoing work of his group of 20+ PhD and Master students and colleagues at MIT (http://www.covidanalytics.io/team) on a variety of aspects of COVID-19 as detailed in covidanalytics.io. The website contains many downloadable models and data.
Specifically, he will discuss
1) a new epidemiological model of the evolution of COVID-19, detailed in https://github.com/COVIDAnalytics/DELPHI
2) a new machine learning model (based on optimal classification trees) for predicting mortality using data from an Italian hospital from Cremona Italy detailed in http://www.covidanalytics.io/risk_calculator. Depending on progress he will discuss how to prioritize patients to use ICUs to save more lives.
3) a new optimization model for allocating ventilators among states in the US detailed in http://www.covidanalytics.io/ventilator_allocation
4) a data generation process from 130+ papers written in March 2020 regarding COVID-19 detailed in http://www.covidanalytics.io/dataset and
insights gleaned from it http://www.covidanalytics.io/interactive-graph
The emphasis of this work has been rapid prototyping and implementation. This talk also represents an invitation to researchers around the world to contribute with ideas and analyses of their own in a collaborative spirit with the objective to address this pandemic using analytics.
Wednesday, April 29, 2020
- Dimitris Bertsimas (MIT, Cambridge, MA, USA).