Harrison Zhu
A5089110295- COVID-19 epidemiological studies
- COVID-19 Pandemic Impacts
- Data-Driven Disease Surveillance
- Gaussian Processes and Bayesian Inference
- COVID-19 and Mental Health
- Vaccine Coverage and Hesitancy
- Influenza Virus Research Studies
- Global Health Care Issues
- Generative Adversarial Networks and Image Synthesis
- Probabilistic and Robust Engineering Design
- Retinoids in leukemia and cellular processes
- Air Quality Monitoring and Forecasting
- COVID-19 Clinical Research Studies
- SARS-CoV-2 and COVID-19 Research
- COVID-19 and healthcare impacts
- Acne and Rosacea Treatments and Effects
- Scientific Measurement and Uncertainty Evaluation
- Climate Change and Health Impacts
- Food Security and Health in Diverse Populations
- Advanced Neural Network Applications
- Chronic Disease Management Strategies
- Advanced Multi-Objective Optimization Algorithms
- Health disparities and outcomes
- Bayesian Methods and Mixture Models
- COVID-19 diagnosis using AI
Baylor College of Medicine
2024
Imperial College London
2020-2022
University of Oxford
2020
Novartis (Switzerland)
2020
Emodo (United States)
2020
Age-specific contact How can the resurgent epidemics of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) during 2020 be explained? Are they a result students going back to school? To address this question, Monod et al. created matrix for infection based on data collected in Europe and China extended it United States. Early pandemic, before interventions were widely implemented, contacts concentrated among individuals similar age highest school-aged children, between children...
Abstract As of 1st June 2020, the US Centres for Disease Control and Prevention reported 104,232 confirmed or probable COVID-19-related deaths in US. This was more than twice number next most severely impacted country. We jointly model epidemic at state-level, using publicly available death data within a Bayesian hierarchical semi-mechanistic framework. For each state, we estimate individuals that have been infected, are currently infectious time-varying reproduction (the average secondary...
Summary Brazil is an epicentre for COVID-19 in Latin America. In this report we describe the Brazilian epidemic using three epidemiological measures: number of infections, deaths and reproduction number. Our modelling framework requires sufficient death data to estimate trends, therefore limit our analysis 16 states that have experienced a total more than fifty deaths. The distribution among highly heterogeneous, with 5 states—Sao Paulo, Rio de Janeiro, Ceara, Pernambuco Amazonas—accounting...
Following the emergence of a novel coronavirus (SARS-CoV-2) and its spread outside China, Europe has experienced large epidemics. In response, many European countries have implemented unprecedented non-pharmaceutical interventions including case isolation, closure schools universities, banning mass gatherings and/or public events, most recently, wide-scale social distancing local national lockdowns. this technical update, we extend semi-mechanistic Bayesian hierarchical model that infers...
Abstract The UK and Sweden have among the worst per-capita COVID-19 mortality in Europe. stands out for its greater reliance on voluntary, rather than mandatory, control measures. We explore how timing effectiveness of measures UK, Denmark shaped each country, using a counterfactual assessment: what would impact been, had country adopted others’ policies? Using Bayesian semi-mechanistic model without prior assumptions mechanism or interventions, we estimate time-varying reproduction number...
Italy was the first European country to experience sustained local transmission of COVID-19. As 1st May 2020, Italian health authorities reported 28,238 deaths nationally. To control epidemic, government implemented a suite non-pharmaceutical interventions (NPIs), including school and university closures, social distancing full lockdown involving banning public gatherings non essential movement. In this report, we model effect NPIs on using data average mobility. We estimate that...
Abstract As of 1st June 2020, the US Centers for Disease Control and Prevention reported 104,232 confirmed or probable COVID-19-related deaths in US. This was more than twice number next most severely impacted country. We jointly modelled epidemic at state-level, using publicly available death data within a Bayesian hierarchical semi-mechanistic framework. For each state, we estimate individuals that have been infected, are currently infectious time-varying reproduction (the average...
Knowing COVID-19 epidemiological distributions, such as the time from patient admission to death, is directly relevant effective primary and secondary care planning, moreover, mathematical modelling of pandemic generally. We determine distributions for patients hospitalized with using a large dataset (N = 21 000 − 157 000) Brazilian Sistema de Informação Vigilância Epidemiológica da Gripe database. A joint Bayesian subnational model partial pooling used simultaneously describe 26 states one...
1 Abstract Brazil is currently reporting the second highest number of COVID-19 deaths in world. Here we characterise initial dynamics across country and assess impact non-pharmaceutical interventions (NPIs) that were implemented using a semi-mechanistic Bayesian hierarchical modelling approach. Our results highlight significant these NPIs had states, reducing an average R t > 3 to 1.5 by 9-May-2020, but failed reduce < 1, congruent with worsening epidemic has experienced since. We...
Abstract We propose a new framework to model the COVID-19 epidemic of United Kingdom at level local authorities. The fits within general for semi-mechanistic Bayesian models epidemic, with some important innovations: we proportion infections that result in reported deaths and cases as random variables. This is contrast standard frameworks latent infection deterministic function time varying reproduction number, R t . tailored designed be updated daily based on publicly available data....
Abstract We propose a new framework to model the COVID-19 epidemic of United Kingdom at local authority level. The fits within general for semi-mechanistic Bayesian models based on renewal equations, with some important innovations, including random walk modelling reproduction number, incorporating information from different sources, surveys estimate time-varying proportion infections that lead reported cases or deaths, and underlying as latent variables. is designed be updated daily using...
Summary Following initial declines, in mid 2020, a resurgence transmission of novel coronavirus disease (COVID-19) has occurred the United States and parts Europe. Despite wide implementation non-pharmaceutical interventions, it is still not known how they are impacted by changing contact patterns, age other demographics. As COVID-19 control becomes more localised, understanding demographics driving these impacts loosening interventions such as school reopening crucial. Considering dynamics...
Sequential VAEs have been successfully considered for many high-dimensional time series modelling problems, with variant models relying on discrete-time mechanisms such as recurrent neural networks (RNNs). On the other hand, continuous-time methods recently gained attraction, especially in context of irregularly-sampled series, where they can better handle data than methods. One class are Gaussian process variational autoencoders (GPVAEs), VAE prior is set a (GP). However, major limitation...
Abstract Stochastic processes provide a mathematically elegant way to model complex data. In theory, they flexible priors over function classes that can encode wide range of interesting assumptions. However, in practice efficient inference by optimisation or marginalisation is difficult, problem further exacerbated with big data and high dimensional input spaces. We propose novel variational autoencoder (VAE) called the prior encoding ( $$\pi $$ <mml:math...
Knowing COVID-19 epidemiological distributions, such as the time from patient admission to death, is directly relevant effective primary and secondary care planning, moreover, mathematical modelling of pandemic generally. We determine distributions for patients hospitalised with using a large dataset ( N = 21,000 – 157,000) Brazilian Sistema de Informação Vigilancia Epidemiológica da Gripe database. A joint Bayesian subnational model partial pooling used simultaneously describe 26 states one...
The widespread availability of satellite images has allowed researchers to model complex systems such as disease dynamics. However, many have missing values due measurement defects, which render them unusable without data imputation. For example, the scanline corrector for LANDSAT 7 broke down in 2003, resulting a loss around 20\% its data. Inpainting involves predicting what is based on known pixels and an old problem image processing, classically PDEs or interpolation methods, but recent...
Stein variational gradient descent (SVGD) is a deterministic particle inference algorithm that provides an efficient alternative to Markov chain Monte Carlo. However, SVGD has been found suffer from variance underestimation when the dimensionality of target distribution high. Recent developments have advocated projecting both score function and data onto real lines sidestep this issue, although can severely overestimate epistemic (model) uncertainty. In work, we propose Grassmann (GSVGD) as...
For many survey-based spatial modelling problems, responses are observed as spatially aggregated over survey regions due to limited resources. Covariates, from weather models and satellite imageries, can be at different resolutions, making the pre-processing of covariates a key challenge for any task. We propose Gaussian process regression model flexibly handle multiresolution by employing an additive kernel that efficiently aggregate features across resolutions. Compared existing approaches...