Focus Areas

Modeling and Simulation

To develop optimal planning and intervention strategies for COVID-19, we are making mathematical models at two levels: regional levels and national level. These models take into account Nepal’s demography, road and flight networks, mobility, healthcare infrastructure, infection status etc. For regional level, adapting a well-known SEIR model, our initial work (available as pre-print, and is under review) has already proved useful in raising awareness among many people that just lockdown will not help much if it is not accompanied by aggressive active case finding and isolation.

For the national level we are working on inter-municipal disease spread analysis. Our graph-based model represents each local government region (N=753 across Nepal) as nodes and the edges represent their connectivity (human mobility). Based on Nepal’s road networks, average daily mobility between these nodes, the availability of health facilities in each of these nodes, and the risk score estimated from data in WP1, we are modeling various partial lockdown scenarios and its impact on the disease spread. This will help policy makers gain insights on partial lockdown scenarios and take decisions based on how it impacts the disease spread. For this, we are also engaging with a telecom provider to obtain anonymised and aggregated mobility data from which we can develop an Origin-Destination mobility matrix that provides a measure of human flow density between any two nodes.

Pandey, Kiran Raj, Anup Subedee, Bishesh Khanal, and Bhagawan Koirala. "COVID-19 Control Strategies and Intervention Effects in Resource Limited Settings: A Modeling Study." medRxiv (2020).

National Covid Data HUB

We are creating a software infrastructure to help centralize the fragmented and distributed COVID-19 data collection efforts of the Government. We emphasise on a unified national database that hosts the public data with the right public agency with appropriate oversight. We have an architecture that allows for, among other things, complete anonymization of this sensitive data, compliance with HL7 standards, and transparent data edit/deletion/usage reviews and audits for an accountable multi-agency and multi-user collaboration.

Its dashboard is designed with an in-browser database working in tandem with the server side multi-model databases. It is thus capable of continuing to work despite poor network conditions and can sync with the central database once network connection is available again. We will also be releasing iOS and Android apps soon.

The local government bodies are collecting data such as symptoms and locations of foreign returnees, the health ministry is collecting RDT and PCR test results, while CCMC and Police are also independently doing contact tracing and keeping track of foreign returnees. Our platform can enable all of these bodies to use an integrated system to enter and visualize data so that the efforts of each of these agencies provide synergy to develop an integrated analytics and advanced contact tracing and surveillance.


Using sample anonymised data obtained from a telecom provider of Nepal, we are exploring advanced approaches for contact tracing. Relatively low smartphone penetration and even less adoption of new apps mean app based contact tracing cannot cover a large majority of Nepali population. Current approach of the government to manually identify contacts by individual phone calls is not scalable if the infected numbers rise, and is also prone to miss many contacts as it is easy for humans to forget or intentionally conceal information about contacts. The importance of accurate contact tracing is even more important in our setting where there is extremely limited testing capability. These resources should be used in a highly optimized and efficient manner.

We propose to combine the data obtained from the above two focus areas, and use advanced deep learning based approaches on the aggregated and anonymised data to obtain anonymised ids of people who are most likely to be infected or in direct contact with existing infected individuals. These anonymised id can then be accessed by the people who are doing contact tracing to convert into actual phone numbers and follow them up, or do pooled testing. We also have a sufficient and necessary foundation for monitoring and geofencing in-place sheltering and quarantines, and for helping plan optimal transports of essential supplies and personnel.