Yufan Zheng

Yufan Zheng

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 March 2021 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
CZChou-Jun Zhan
Chou-Jun Zhan
YZYufan Zheng
Yufan Zheng
HZHaijun Zhang
Haijun Zhang
Quansi WenQuansi Wen
Quansi Wen
The rapid geographic spread of COVID-19, to which various factors may have contributed, has caused a global health crisis. Recently, the analysis and forecast of the COVID-19 pandemic have attracted worldwide attention. In this work, a large COVID-19 data set consisting of COVID-19 pandemic, COVID-19 testing capacity, economic level, demographic information, and geographic location data in 184 countries and 1241 areas from December 18, 2019, to September 30, 2020, were developed from public reports released by national health authorities and bureau of statistics. We proposed a machine learning model for COVID-19 prediction based on the broad learning system (BLS). Here, we leveraged random forest (RF) to screen out the key features. Then, we combine the bagging strategy and BLS to develop a random-forest-bagging BLS (RF-Bagging-BLS) approach to forecast the trend of the COVID-19 pandemic. In addition, we compared the forecasting results with linear regression (LR) model, $K$ -nearest neighbors (KNN), decision tree (DT), adaptive boosting (Ada), RF, gradient boosting DT (GBDT), support vector regression (SVR), extra trees (ETs) regressor, CatBoost (CAT), LightGBM (LGB), XGBoost (XGB), and BLS.The RF-Bagging BLS model showed better forecasting performance in terms of relative mean-square error (RMSE), coefficient of determination ( $R^{2}$ ), adjusted coefficient of determination ( $R_{\mathrm{ adj}}^{2}$ ), median absolute error (MAD), and mean absolute percentage error (MAPE) than other models. Hence, the proposed model demonstrates superior predictive power over other benchmark models.

6 February 2023 in Information Sciences
Verified
Journal Article
Closed access
CZChou-Jun Zhan
Chou-Jun Zhan
YZYufan Zheng
Yufan Zheng
LSLujiao Shao
Lujiao Shao
GCGuanrong Chen
Guanrong Chen
HZHaijun Zhang
Haijun Zhang

1 September 2025 in Infectious Disease Modelling
Verified
Journal Article
Open access
YZYufan Zheng
Yufan Zheng
KYKeqi Yue
Keqi Yue
EWEric W.M. Wong
Eric W.M. Wong
HYHsiang-Yu Yuan
Hsiang-Yu Yuan
While Aedes mosquitoes, the Dengue vectors, are expected to expand due to climate change, the impact of human mobility on them is largely unclear. Changes in human mobility, such as staying at home during the pandemic, likely affect mosquito abundance. We aimed to assess the influence of human mobility on the abundance and extensiveness of Aedes albopictus, taking account of the nonlinear lagged effects of weather, during the COVID-19 pandemic in Hong Kong. Google human mobility indices (including residential, parks, and workplaces) and weather conditions (total rainfall and mean temperature) along with Aedes albopictus abundance and extensiveness, monitored using Gravidtrap were collected between April 2020 and August 2022. Distributed lag non-linear models with mixed-effects models were used to explore their influence in three areas of Hong Kong. Time spent at home (i.e., residential mobility) was negatively associated with mosquito abundance. The model projected that if residential mobility in 2022 was returned to the pre-pandemic level, the mosquito abundance would increase by an average of 80.49 % compared to actual observation. The relative risk (RR) of mosquito abundance was associated with low rainfall (500 mm) within 3 months was also associated with a peak RR of 1.41. Warm conditions (21–30 °C, compared with 20 °C) were associated with a higher RR of 1.47 after half a month. Human mobility is a critical factor along with weather conditions in mosquito prediction, and a stay-at-home policy may be an effective intervention to control Aedes albopictus.

19 April 2024 in medRxiv
Verified
Preprint
Open access
PDF file
YZYufan Zheng
Yufan Zheng
KYKeqi Yue
Keqi Yue
EWEric Wong
Eric Wong
HYHsiang-Yu Yuan
Hsiang-Yu Yuan
Background: While Aedes mosquitoes, the Dengue vectors, were expected to expand their spread due to international travel and climate change, the effects of human mobility and low rainfall conditions on them are largely unknown. We aimed to assess these influences during the COVID-19 pandemic in Hong Kong, characterized by varying levels of human mobility. Methods: Google’s human mobility indices (including residential, parks and workplaces) and weather conditions (total rainfall and mean temperature) together with Aedes albopictus abundance and extensiveness monitored using Gravidtrap were obtained between April 2020 and August 2022. Distributed lag non-linear models with mixed-effects models were used to explore their influence in three areas in Hong Kong. Findings: The relative risk (RR) of mosquito abundance was associated with low rainfall (500 mm) within 3 months was also associated with a peak of RR at 1.41. Warm conditions (21-30°C; compared with 20°C) were associated with a higher RR of 1.47 after half a month. Residential mobility was negatively associated with mosquito abundance. The model projected that if residential mobility in the year 2022 was reduced to the level before the COVID-19 pandemic, the mosquito abundance would increase by an average of 80.49% compared to the actual observation. Significance: Both the human mobility and the lag effect of meteorological factors can be critical for the prediction of vector dynamics, and stay-at-home policy may be useful for its control in certain regions. AUTHOR SUMMARY: Previous studies have demonstrated that both meteorological factors and human mobility were linked to the risk of Dengue transmission, with rainfall potentially exerting delayed effects. Moreover, dry conditions have been found to increase Dengue risk in recent years. However, the impact of these factors on vector (mosquito) activity remains unclear. This study assessed the effect of human mobility and rainfall on the Dengue mosquito. The Gravitrap indices were used to characterize local mosquito ( Aedes Albopictus ) abundance and extensiveness conditions. We used established Gravitrap indices to characterize mosquito abundance and extensiveness in Hong Kong. We found that i) the decrease in residential mobility might increase mosquito abundance and extensiveness; and ii) low rainfall (<50 mm) was associated with a higher risk of mosquito abundance after 4.5 months. Additionally, heavy rainfall was associated with increased mosquito activity risk.The future mosquito activity risk is expected to increase because of the relaxation of social distancing measures after the COVID-19 pandemic along with climate change. The results suggest that non-linear delayed effects of meteorological factors together with human mobility change can be used for the Dengue mosquito forecast. Social distancing may be a way to reduce the risk of Aedes albopictus .

1 September 2026 in EcoEvoRxiv
Verified
Preprint
Open access
PDF file
YZYufan Zheng
Yufan Zheng
ESElle Saber
Elle Saber
BGBernd Gruber
Bernd Gruber
HNHelen Nahrung
Helen Nahrung
ACAngus Carnegie
Angus Carnegie
RDRichard Duncan
Richard Duncan
Human-mediated biological invasions are accelerating worldwide, but detecting early incursions and tracking their spread remains difficult. Opportunistic citizen science platforms, such as iNaturalist and eBird, generate large volumes of presence-only records, offering new opportunities to detect and monitor invasive species. However, these data come with uneven sampling effort, strong spatial and taxon biases, and lack explicit non-detections, complicating efforts to infer species absences. To address these challenges, we develop a dynamic occupancy-detection model (DODM) that estimates colonisation, local extinction, and occupancy dynamics for invasive species using presence-only data. The model infers detection probability as a function of background sampling effort from related taxa, enabling separation of non-detection from true absence. This detection process, combined with a temporal correlation structure and spatial dependency, is used to infer spatiotemporal changes in species occupancy using a hierarchical Bayesian model that captures uncertainty in how invasions unfold over space and time. Using iNaturalist observations, we apply this model to infer detection probability, occupancy trajectories, dynamic colonisation and extinction probabilities for eight recently introduced insect species in eleven locations worldwide. Our results show that a well-structured DODM can unlock dynamic information from presence-only data, offering a flexible and generalizable framework for monitoring and risk mapping biological invasions.