Harnessing agricultural models for disease identification and spread

Identify ecological niches supporting survival of emerging infectious agents and predict spread using open access mechanistic models and data

Dr Katherine Brown, a Senior Researcher based jointly at the University of Cambridge and the University of Texas at Austin.

Covid-19 laid bare our lack of a systematically developed infectious diseases alert system. Limited access to modelling tools for identifying and predicting the emergence and spread of pathogens in local environments hinders the ability of local and global health authorities to prepare and mitigate ongoing and future disease outbreaks. Openly available computational tools that harness existing databases and provide near instantaneous feedback through servers and webservices are enabling a wide variety of research and health support activities to be undertaken. However, using these resources requires expertise not easily available for modelling associated with infectious diseases. In addition, modelling platforms are primarily focused on human-to-human spread using an epidemiologist’s perspective that often underplays links between the survival and spread of disease and environmental conditions. Ecological niches for pathogens vary widely and exist in places where vulnerable humans inhabit. Given the wealth of data available about environmental conditions and reporting of diseases from local, national and global sources, there is enormous untapped potential to provide access to data and models in open access platforms. Access to these resources would enable researchers and public health experts to map and predict the spread of diseases to better inform health management strategies and decisions.

Our target audience is public health providers and researchers in resource-limited settings who need user-friendly, accessible computational tools for modelling and visualisation of ecological settings where pathogens can or could inhabit. A key goal of this project is to create a platform that could be deployed at a global level to enable improved access to data resources, computational simulations and data imaging (e.g., GIS mapping). A key outcome is to provide access to this platform to facilitate identification and prediction of disease spread as a means for informing policies in disease management, and for stimulating more basic and applied research at local levels. We understand that a limitation to using modelling and visualisation tools, particularly in resource-limited settings, is access to the expertise needed to use software – which in turn can be further limited by complicated user interfaces and appropriate computational facilities. Our aim here is to develop a platform that is accessible and sustainable, leveraging knowledge and experience for agroecosystem modelling in an existing platform using high performance computing. We will be engaging end users with vulnerable communities to develop and adapt this solution to create a platform that is both functional and addresses their most pressing needs.

This work aims to deliver output for the public good and includes:

1. Open-access models

2. A portal for accessing pre-calculated layered maps that combine environmental, human risk factors and disease data

3. An interactive platform for external users to explore and develop models to identify ecosystems that can support emerging pathogens and predictive tools to anticipate how pathogen might spread in response to normal and stressed environmental factors

4. Utilisation of the developed system by researchers and public health professionals to address issues associated with disease identification and spread, particularly in vulnerable communities

5. Peer-reviewed publications

Risks of contracting disease are often linked with a lack of recognition of the presence of pathogens in local environments and accompanying under-diagnosis. While health providers often recognise these gaps, it is often difficult to implement mitigation strategies and change policies without evidenced-based data and analyses. The need for developing openly accessible resources has been identified through interactions and consultations with stakeholders in the US and Africa.

The purpose of this solution is to provide an innovative open-accessibly resource to enable the generation of evidence-based data for pathogen identification and to provide high-level modelling and mapping tools for use by researchers and public health providers in resource-limited settings. The tool is “autocatalytic”: the more modelling and information in the system, the larger the application and testing domain and the more useful for the user.

Computational support via tools that streamline the development and deployment of intelligent science-based decision support systems (iDSS) are needed at a global level. In particular, the existing cyberecosystem that we have described (e.g., DataX, MINT and Cycles) provides information services to accelerate time to discovery and time to decision-relevant research products. Once models are developed and data are collected, getting that information and knowledge ‘off-the-shelf’ and into use requires aid. An iDSS with Artificial Intelligence, or AI-enabled interfaces that are end-user focused, will catalyse collaborations between expert modellers and non-technical users and facilitate communication of information to the communities and policymakers who need it.

In this context we are committed to providing capabilities for scaling the impact of our solution.

To effect transformational impact we intend to leverage expertise associated within our team in two main ways:

1. Creation of a sustainable open-access platform for modelling and visualisation tools at the Texas Advanced Computing Center (TACC) that will be publicly accessible worldwide.

2. Engagement and development of these tools to address needs of stake-holders through building relationships facilitated by the Cambridge Global Challenges team with Global South-based research organisations, public health institutions and NGO’s.

Our project is currently in the proof-of-concept stage. As part of our project development, we have identified some key measurable indicators of progress and impact of our solution:

1. Establish procedures that enable reproducible approaches and the generation of methods/replicable results using “FAIR” data products and documented processes.

Progress: Production of peer-reviewed publications; and deposition of data and procedures in open access repositories

Impact: Citations of publications, data and/or protocols

2. Develop interfaces that promote open access to modelling, map generation and data input capabilities.

Progress: Creation of interactive atlases for map visualisation; and adaptation of Cycles modelling software for pathogen prediction

Impact: Utilisation of these open accessibly resources by stake-holders including those working with data associated with individuals in vulnerable communities

3. Undertake “blind” analyses using historical data to identify parameters/variables that can be used to establish relationships between human risk factors and environmental characteristics to predict pathogen prevalence and/or emergence.

Progress: Creation of maps, data sources and modelling protocols in the DataX platform

Impact: Documented use of these “outputs” in research activities and/or as evidence in the development of public health strategies for pathogen identification, monitoring and/or spread, particularly in areas within or that border the Global South.

1. Financial

Sufficient resources are required to employ staff to carry out the required work to implement this solution. Currently, we have been laying the groundwork for adaptation of computational modelling tools for disease prediction. This work has contributed to our proof-of-concept status but needs staff continuity and specific expertise to realise the solution within the three-year time-frame.

2. Data access

We anticipate that participation in the Trinity Challenge will facilitate greater ease of access to health reporting data that are critical to development of credible maps and the modelling proposed in our solution. We are interested in exploiting the reach of Trinity Challenge partners and resources to access as much data related to disease reporting as possible.

3. Engagement

Our budget includes a request for support to partner organisations in the Global South who will co-develop our solution with our team. We will be working with Cambridge Global Challenges to identify groups in areas of sub-Saharan Africa where Cycles modelling has been implemented already. To enable a high-level of engagement with these stake-holders who can work with us on pilot development for future deployment, our budget includes financial support to ensure that we have engagement in appropriate target environments.

The University of Texas at Austin; Texas Advanced Computing Center; Planet Texas 2050 Grand Challenge; The University of Cambridge; Cambridge Global Challenges; Penn State University.

We are applying to The Trinity Challenge because the approach we taking to implement the ideas in our solution are not only cross-disciplinary, bringing together a new approach to modelling using agroecosystem simulation alongside disease/pathogen data - but can support an international team that includes stakeholders working with vulnerable communities. 

The barriers that The Trinity Challenge can help us overcome are:

1. Financial: Support will enable us to sufficient support dedicated staff to develop and implement our solution within the 3 year timeline. 

2. Data Access: We anticipate that we will be able to more effectively use open-source datasets that can enhance our modelling capabilities with The Trinity Challenge's data catalogue and associated expertise that we hope to access through its network of members and partners. 

3. Engagement with Stakeholders: Our budget includes resources that enable us to support stakeholders in LMIC's who will be engaged in developing our solution to meet their needs and is accessible within their working conditions. 

Through Dr. Brown's joint appointment, we are already able to partner with the University of Cambridge and her existing network of contacts that include Cambridge Global Challenges and the Low-to-Middle-Income Country Covid-19 group. In addition, Dr. Pierce is already working with Microsoft, using the Cloud Computing Services of Azure and we anticipate that utilising these resources for aspects of our solution will offer even greater flexibility, particularly to external users outside of UT Austin, where high performance computing facilities may not be necessary for execution of all activities proposed here.