Page 22 - Covid_NL_11_9_21

A network optimisation-based prediction model for COVID-19
outbreak developed by IIT Kharagpur
Dr Goutam Sen from IIT-Kharagpur is working on network optimisation-based prediction model for COVID-19 outbreak. The project is supported by the Science and Engineering Research Board (SERB), a statutory body under DST, Government of India, under the MATRICS scheme for studying mathematical modelling and computational aspects to tackle the COVID-19 pandemic.
Modelling of COVID-19 spread and other similar infectious diseases is a significantly challenging task due to the inherently stochastic contagion process. The Arogya Setu app developed by MeITY, Government of India, is a significant step forward to create a mechanism of location tracking of registered mobile numbers.
The epidemiological problem can be modelled as a constrained Steiner tree problem, which is NP-hard. So, an efficient heuristic algorithm is proposed to design to solve the underlying optimisation model and test its performance using a contagion simulation episode. A SIR- based agent simulation model has been developed to create benchmark dataset. The model’s inputs are carefully constructed from the features of COVID-19. For generating test dataset, a contagion episode is simulated following a stochastic contagion process. In this simulation, the transmission probabilities are estimated from the link level data (i.e. date and duration of contact, and some other demographic factors). The performance of the optimisation model and heuristic algorithm will be tested using these simulated datasets. Further, an optimisation model is developed to identify the most influential contacts in a network so that they can be targeted for testing and isolation. The validation of the model is in process.
In the network analysis and agent-based simulation, it has been observed that there are a handful number of people who are responsible for the explosive growth of the number of cases. These people are known as super spreaders and can be easily detected as hubs in the contact network. So, under limited resources, the model helps to target very specific people to testing and isolation, thus containing the spread of the disease.
Contact info:
gsen@iem.iitkgp.ac.in
DECOVID: Data-assimilation and error correction of viral infectious
disease models – study by IISc, Bengaluru
A team of researchers from Indian Institute of Science (IISc), Bengaluru, is working on a project that will study the new data assimilation and error correction theory for infectious disease models, numerical schemes and scalable computational systems to implement Bayesian data assimilation.
Several dynamical models are available for forecasting the spread of infectious diseases such as SIR, SEIR, SIS. These are differential equation based models that seek to model a complex phenomenon with several unknowns. The goal of the present project is to develop numerical schemes and algorithms for a Bayesian data assimilation methodology to rigorously correct forecast errors of differential equation-based viral infectious disease dynamical models, and to improve their prediction skill. The technology will be used to correct model errors due to uncertainty in any forward infectious disease model that is used in practice.
Contact info:
deepakns@iisc.ac.in
VOL. IV ISSUE 10
VIGYAN PRASAR 18
COVID-19 SCIENCE & TECHNOLOGY EFFORTS IN INDIA