Global health seeks to understand and accommodate the complex systems of our planet-wide society as they relate to health(Salm et al., 2021). Traditionally, global health institutions (e.g. WHO) have, in a top-down manner, negotiated with the highest levels of government across our world’s nations toward aligned health policy (Salm et al., 2021). Unfortunately, top-down approaches have produced mixed results. While they succeed shaping in economic matters, they consistently fail to achieve social progress (Hoffman & Røttingen, 2015; Hoffman et al., 2015).

Recently, such institutions have taken steps towards a more holistic, multi-level approach under the banner of the One Health (OH) approach (Osterhaus et al., 2020). The OH approach strives to mobilize multiple sectors, disciplines, and communities at varying levels of society to work together to foster well-being and tackle threats to health and ecosystems (Adisasmito et al., 2022), with the inclusion of digital health(Benis et al., 2021; Ho, 2022).

However, there remain challenges regarding the implementation of OH. Key challenges include policy and funding, education and training, as well as multi-actor, multi-domain and multi-level collaborations (dos S. Ribeiro et al., 2019). This is despite the increasing accessibility to knowledge and digital research tools through the internet.

To tackle those key challenges in Global Health (and more specifically in Global Health and Machine Learning), we propose a bottom-up or grassroots community based means of participatory research to bring together researchers from varying parts of society. Participatory research, unlike conventional research, emphasizes the value of research partners in the knowledge-production pro-cess where the research process itself is defined collaboratively and iteratively (English et al., 2018).

In this work, we review some existing Grassroots Participatory Communities (GPCs) and propose a Grassroots Participatory Community Framework summarising the participatory approaches by Machine Learning communities. We intend this framework to enable any small group of individuals, with scarce resources, to build and sustain an online community within the space of 2 or so years. Under this framework, we provide a recommended roadmap to create a machine learning for global health community in Africa (ML4GHA) GPC, as a means to alleviate some of the problems highlighted in implementing OH.

With the increasing number of omics data, there is a great need to incorporate these datasets together to create a better and more robust understanding of the underlying biological processes. We transform this problem into a noisy multiview independent component analysis (ICA) task by assuming that each observed dataset (view) is a linear mixture of independent latent biological processes. Furthermore, we assume that each view contains a mixture of shared and individual sources. To computationally estimate the sources, we optimize a constrained form of the joint log-likelihood of the observed data among all views. Finally, we apply the proposed model in a challenging real-life application, where the estimated shared sources from two large transcriptome datasets (observed data) provided by two different labs (two different views) lead to a more plausible representation of the underlying graph structure than existing baselines.

The recent outbreak of the novel coronavirus known as the COVID-19 pandemic has harmed the lives of millions of people across the globe and has imposed a significant threat to global healthcare due to its severe transmission capacity. It’s of utmost importance to be able to accurately forecast the COVID-19 pandemic and to provide the necessary precautionary measures to protect the health of individuals and prevent the spread of this deadly widespread virus. In this paper, we propose to forecast the upcoming newly infected patients that are likely to be affected by COVID-19 in prior using a novel deep learning framework, Spatio-Temporal Attention Based Graph Convolution Networks (STAGCN) to effectively make use of spatial and temporal relationships. Instead of using traditional time-series forecasting techniques at a single city using raw data, we model the problem using graphs and aim at taking into account the dependency that an infection in one city has on its neighbors. Our experiments show that STAGCN effectively captures this dependency and consistently outperforms the other conventional methods.

Dynamic distribution shifts caused by evolving diseases and demographic changes require domain-incremental adaptation of clinical deep learning models. However, this process of adaptation is often accompanied by catastrophic forgetting, and even the most sophisticated methods are not good enough for clinical applications. This paper studies incremental learning from the perspective of mode connections, that is, the low-loss paths connecting the minimisers of neural architectures (modes or trained weights) in the parameter space. The paper argues for learning the low-loss paths originating from an existing mode and exploring the learned paths to find an acceptable mode for the new domain. The learned paths, and hence the new domain mode, are a function of the existing mode. As a result, unlike traditional incremental learning, the proposed approach is able to exploit information from a deployed model without changing its weights. Pre-COVID and COVID-19 data collected in Oxford University hospitals is used as a case study to demonstrate the need for domain-incremental learning and the advantages of the proposed approach.

Binary phylogenetic trees inferred from biological data are central to understanding the shared evolutionary history of organisms. Inferring the placement of latent nodes in a tree by any optimality criterion (e.g., maximum likelihood) is an NP-hard problem, propelling the development of myriad heuristic approaches. Yet, these heuristics often lack a systematic means of uniformly sampling random trees or effectively exploring a tree space that grows factorially, which are crucial to optimisation problems such as machine learning. Accordingly, we present Phylo2Vec, a new parsimonious representation of a phylogenetic tree. Phylo2Vec maps any binary tree with n leaves to an integer vector of length n. We prove that Phylo2Vec is both well-defined and bijective to the space of phylogenetic trees. The advantages of Phylo2Vec are twofold: i) easy uniform sampling of binary trees and ii) systematic ability to traverse tree space in very large or small jumps. As a proof of concept, we use Phylo2Vec for maximum likelihood inference on five real-world datasets and show that a simple hill climbing-based optimisation efficiently traverses the vastness of tree space from a random to an optimal tree.

Disease spread represents an increasing challenge in refugee and internally displaced person (IDP) settlements. The movement and interaction of people within camps is influenced by their layout, which therefore has the potential to significantly affect disease spread. This work aims at creating a methodology to explore the potential effects of different camp layouts as mitigating factors in the spread of diseases within settlements. We showcase proof-of-concept experiments by leveraging the JUNE agent-based epidemic model, discuss the kind of operational insights this methodology can facilitate, and provide a framework for future investigations.