Hospital response management in a crisis - Samsung Solve For Tomorrow
Online-Phase

Hospital response management in a crisis

Simulation of an epidemy propagation and of the hospitals' response to it. The distribution of people in different diseases stages and the load on the hospitals can be measured over time.
7 Kommentare

Die Herausforderung

The idea is to help predict regionwise outbreaks of a disease and follow the impact of social and logistical measures. Overloaded hospitals predicted by the simulation can be studied in order to adapt the response to the crisis.

Die Zielgruppe

The healthcare system and authorities deciding on the epidemy management.

Das Team

Being two students in life science engineering and feeling involved against the current crisis, we would like to combine our immunologic and programming knowledge to help manage the response. We wouldn't mind the help of an epidemiologist and someone involved in healthcare ressources management.

Die Lösung

Until now the simulation runs with 500 points (groups of people) following the SIR (susceptible-immune-removed) model for contagion. Four different infectious stages are modelled and a parameter of susceptibility towards the infection has been implemented. Two different kind of hospitals exist: overloaded and not. Different regions are created and infected people will be treated in hospitals of the regions but some people might travel between the regions.

Fragen an die Community

We might need some data to have a better fit for the probabilities of the disease's spread. A closer contact with people in charge of pandemy management would also be of a great help.
4+
Iléane LefevreAmira NeumannBoris KunzTâm Nguyên

0 thoughts on “Hospital response management in a crisis

  • Holger Hoffmann-Riem

    Hi Amira, how would you ensure that this project stays “alive”? I think one big challenge is that nobody knows when the next pandemic will pop up. If this is in 20 years, then most people will have forgotten about covid 19, and willingness to pay for a model on an ongoing basis might be low. At the same time operating systems will continue to change, so code written in 2020 is unlikely to be of much use in 2040. Do you have any idea how to deal with this? Best wishes, Holger

    0
    • Tâm Nguyên

      I think the main interest of this project is that even if the code may have to be completely rewritten, it provides an interesting extension to the usual SIR model, and provided we work enough on it, it could become the basis for an extended standard model which could then be applied to any kind of epidemic in any kind of infrastructure network. We would need to formalise and standardize our inputs and outputs to ensure it is still usable in the coming years. But provided we work in opensource it could very well developp into some participative and derivative work that is continuously updated. Who knows? But we’re eager to see!

      0
        • Tâm Nguyên

          I have had a look at the model linked above and I’m surprise at the depth of their model as well as the quite similar structure we ended up adopting; in a way I’m quite happy we both created the same structure and similar flow between stages.

          But the model you linked is already an extension of SIR, whereas in the simple barebone model you have only three population flowing (Susceptible, Infected, Recovered).

          The different approach we took is first very similar on the patient and disease spread basis to the model you linked, we have close to the same flow with a few somewhat alternatively titled categories. For the resolution they directly solve the differential equations for each population, which we didn’t we used a stochastic process to move points randomly to simulate the spread, while the probabilities are run to transition between stages. While this may be computationnaly slower and provide a stochastic approximation of the answer what this method allow is for greater fine tuning of parameter, say people meetup at central hub, then we can model that while the differential equation cannot without adding some pretty heavy calculation to it. We basically reproduce and extended a fair bit the model detailled in this video: https://www.youtube.com/watch?v=gxAaO2rsdIs

          But most importantly, we do not interests ourselves only in the number of hospital bed but how they’re available in a network of hospital. How many beds are available per region, what kind of flow of ressources is allowed between hospital in different regions. This added depth is quite useful to simulate the regional impact and overloading of certain region while the country at large may not be overloaded. Provided we run the simulation several time we are able to predict which hospitals would be most likely to get overloaded first and how it propagates to other region. And I think it could therefore be applied to reinforce our infrastructural network and inform measure to restrict the flow toward certain region.

          0
          • Holger Hoffmann-Riem

            I think the flow of patients could be your USP. Who would be “paying” for your results? Do you see this as a product run by unpaid volunteer (which means you would be “paying” by working for free), or do you want to set up a commercial service that hospitals pay for?

            0
  • Holger Hoffmann-Riem

    I think the flow of patients could be your USP. Who would be “paying” for your results? Do you see this as a product run by unpaid volunteer (which means you would be “paying” by working for free), or do you want to set up a commercial service that hospitals pay for?

    0
    • Amira Neumann

      Until now we have designed our project as open source to contribute to the global response against the Covid-19 pandemy. The project is not yet in it’s final stage where it could directely help governmental workers by simulating different senarios of the spread of the virus and help to take decisions. That is why we can’t already decide if this would be possibly become a commercial service.
      The most important, independent of the decision to make it an open source project or a commercial service, would be to improve it by the help of an epidemiologist and of hospital staff members.

      0

Hast du eine Anregung oder eine Frage?

Skip to content