Woldegebriel Assefa Woldegerima
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Woldegebriel Assefa Woldegerima has a total of 94 co-authors. Here are the top co-authors by number of shared publications:
- Jude Dzevela Kong
Dalla Lana School of Public Health, University of Toronto
9 shared publications Ali Asgary
School of Administrative Studies, Faculty of Liberal Arts and Professional Studies, York University, 4700 Keele Street, Toronto, M3J 1P3, Ontario, Canada
8 shared publications- Chigozie Louisa Jane Ugwu
Disease-Informed Modelling, Methods, and Systems (DIMMS) Laboratory, Department of Mathematics and Statistics, York University, 4700 Keele Street, Toronto, Ontario, M3J 1P3, Canada
8 shared publications Jianhong Wu
Laboratory for Industrial and Applied Mathematics (LIAM), Department of Mathematics and Statistics, York University
8 shared publications- Nicola Luigi Bragazzi
Department of Clinical Pharmacy, Saarland University, Saarbrücken, Germany
8 shared publications Jude Dzevela Kong
Africa-Canada Artificial Intelligence and Data Innovation Consortium (ACADIC), Canada
7 shared publications
Latest Publications
- Open Access
- 11 August 2026
- Published by Frontiers Media SA in Frontiers in Artificial Intelligence
- Vol. 9, 1886896
Predicting influenza in the post-COVID era: assessing LSTM, GRU, and transformer robustness to covariate shift
Forecasting influenza has become increasingly challenging due to post-COVID disruptions in seasonality and strain circulation. This work compares the performance of Long Short Term Memory Networks (LSTM), Gated Recurrent Unit (GRU), and transformer models in forecasting influenza spread using multivariate epidemiological and environmental data, with a focus on robustness under post-COVID non-stationarity. We compare LSTM, GRU, and transformer architectures within a multivariate deep learning framework using influenza and temperature data from Ontario (2014–2025), with data split into training, validation, and testing periods. Although recurrent models outperform transformers on limited, noisy data, all architectures exhibit marked performance collapse under post-COVID non-stationarity. The GRU and LSTM track pre-COVID seasonal peaks more closely, yet both substantially under-estimate the post-COVID resurgence, indicating that none of the models generalize across the regime shift. These findings position our study as a diagnostic of how architectural inductive biases break down under covariate shift. Furthermore, this manuscript assesses how the COVID-19 pandemic affected the accuracy and performance of machine learning algorithms and notes the integration of transfer learning and attention mechanisms to improve model performance.
- 11 August 2026
- Published by Frontiers Media SA in Frontiers in Artificial Intelligence
- Vol. 9, 1886896
Open AccessPredicting influenza in the post-COVID era: assessing LSTM, GRU, and transformer robustness to covariate shift
Forecasting influenza has become increasingly challenging due to post-COVID disruptions in seasonality and strain circulation. This work compares the performance of Long Short Term Memory Networks (LSTM), Gated Recurrent Unit (GRU), and transformer models in forecasting influenza spread using multivariate epidemiological and environmental data, with a focus on robustness under post-COVID non-stationarity. We compare LSTM, GRU, and transformer architectures within a multivariate deep learning framework using influenza and temperature data from Ontario (2014–2025), with data split into training, validation, and testing periods. Although recurrent models outperform transformers on limited, noisy data, all architectures exhibit marked performance collapse under post-COVID non-stationarity. The GRU and LSTM track pre-COVID seasonal peaks more closely, yet both substantially under-estimate the post-COVID resurgence, indicating that none of the models generalize across the regime shift. These findings position our study as a diagnostic of how architectural inductive biases break down under covariate shift. Furthermore, this manuscript assesses how the COVID-19 pandemic affected the accuracy and performance of machine learning algorithms and notes the integration of transfer learning and attention mechanisms to improve model performance.
- Open Access
- 21 May 2026
- Published by Elsevier in SSRN Electronic Journal
A Semigroup-Theoretic Approach to the Basic Reproduction Number for an Age-Structured SIR Model
We develop a semigroup-theoretic framework to rigorously derive the basic reproduction number for an age-structured SIR epidemic model on the Banach space $L_1(\mathbb{R}_+)$. By formulating the infection dynamics as an abstract linear evolution equation, we construct the associated next-generation operator $K = FV^{-1} = \int_0^{\infty} FT(t) \ dt$, where $T(t)$ is the $C_0$-semigroup generated by the transition operator. We prove that $K$ is bounded, positive, and compact using the Fréchet–Kolmogorov compactness theorem, which enables the application of the Krein–Rutman theorem to characterize the basic reproduction number $\mathcal{R}_0$ as the spectral radius of $K$. In the case of separable transmission kernels, we prove that the next-generation operator reduces to a rank-one operator, collapsing the infinite-dimensional spectral problem to a scalar one and yielding explicit formulae for $\mathcal{R}_0$, while recovering the classical reproduction number from ordinary differential equation models as a special case. We also develop a numerical discretization of the next-generation operator, leading to a finite-dimensional matrix approximation whose dominant eigenvalue approximates the spectral radius. Simulations of the full age-structured system confirm the threshold behaviour predicted by the theory, illustrating convergence to the disease-free equilibrium when $\mathcal{R}_0 < 1$ and persistence when $\mathcal{R}_0 > 1$. This work provides a rigorous infinite-dimensional operator-theoretic framework for analyzing reproduction numbers in structured epidemic models and establishes a foundation for extensions to more general systems.- 21 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Semigroup-Theoretic Approach to the Basic Reproduction Number for an Age-Structured SIR Model
We develop a semigroup-theoretic framework to rigorously derive the basic reproduction number for an age-structured SIR epidemic model on the Banach space $L_1(\mathbb{R}_+)$. By formulating the infection dynamics as an abstract linear evolution equation, we construct the associated next-generation operator $K = FV^{-1} = \int_0^{\infty} FT(t) \ dt$, where $T(t)$ is the $C_0$-semigroup generated by the transition operator. We prove that $K$ is bounded, positive, and compact using the Fréchet–Kolmogorov compactness theorem, which enables the application of the Krein–Rutman theorem to characterize the basic reproduction number $\mathcal{R}_0$ as the spectral radius of $K$. In the case of separable transmission kernels, we prove that the next-generation operator reduces to a rank-one operator, collapsing the infinite-dimensional spectral problem to a scalar one and yielding explicit formulae for $\mathcal{R}_0$, while recovering the classical reproduction number from ordinary differential equation models as a special case. We also develop a numerical discretization of the next-generation operator, leading to a finite-dimensional matrix approximation whose dominant eigenvalue approximates the spectral radius. Simulations of the full age-structured system confirm the threshold behaviour predicted by the theory, illustrating convergence to the disease-free equilibrium when $\mathcal{R}_0 < 1$ and persistence when $\mathcal{R}_0 > 1$. This work provides a rigorous infinite-dimensional operator-theoretic framework for analyzing reproduction numbers in structured epidemic models and establishes a foundation for extensions to more general systems. - Open Access
- 21 May 2026
- Published by Elsevier in SSRN Electronic Journal
A Semigroup-Theoretic Approach to the Basic Reproduction Number for an Age-Structured SIR Model
We develop a semigroup-theoretic framework to rigorously derive the basic reproduction number for an age-structured SIR epidemic model on the Banach space $L_1(\mathbb{R}_+)$. By formulating the infection dynamics as an abstract linear evolution equation, we construct the associated next-generation operator $K = FV^{-1} = \int_0^{\infty} FT(t) \ dt$, where $T(t)$ is the $C_0$-semigroup generated by the transition operator. We prove that $K$ is bounded, positive, and compact using the Fréchet–Kolmogorov compactness theorem, which enables the application of the Krein–Rutman theorem to characterize the basic reproduction number $\mathcal{R}_0$ as the spectral radius of $K$. In the case of separable transmission kernels, we prove that the next-generation operator reduces to a rank-one operator, collapsing the infinite-dimensional spectral problem to a scalar one and yielding explicit formulae for $\mathcal{R}_0$, while recovering the classical reproduction number from ordinary differential equation models as a special case. We also develop a numerical discretization of the next-generation operator, leading to a finite-dimensional matrix approximation whose dominant eigenvalue approximates the spectral radius. Simulations of the full age-structured system confirm the threshold behaviour predicted by the theory, illustrating convergence to the disease-free equilibrium when $\mathcal{R}_0 < 1$ and persistence when $\mathcal{R}_0 > 1$. This work provides a rigorous infinite-dimensional operator-theoretic framework for analyzing reproduction numbers in structured epidemic models and establishes a foundation for extensions to more general systems.- 21 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Semigroup-Theoretic Approach to the Basic Reproduction Number for an Age-Structured SIR Model
We develop a semigroup-theoretic framework to rigorously derive the basic reproduction number for an age-structured SIR epidemic model on the Banach space $L_1(\mathbb{R}_+)$. By formulating the infection dynamics as an abstract linear evolution equation, we construct the associated next-generation operator $K = FV^{-1} = \int_0^{\infty} FT(t) \ dt$, where $T(t)$ is the $C_0$-semigroup generated by the transition operator. We prove that $K$ is bounded, positive, and compact using the Fréchet–Kolmogorov compactness theorem, which enables the application of the Krein–Rutman theorem to characterize the basic reproduction number $\mathcal{R}_0$ as the spectral radius of $K$. In the case of separable transmission kernels, we prove that the next-generation operator reduces to a rank-one operator, collapsing the infinite-dimensional spectral problem to a scalar one and yielding explicit formulae for $\mathcal{R}_0$, while recovering the classical reproduction number from ordinary differential equation models as a special case. We also develop a numerical discretization of the next-generation operator, leading to a finite-dimensional matrix approximation whose dominant eigenvalue approximates the spectral radius. Simulations of the full age-structured system confirm the threshold behaviour predicted by the theory, illustrating convergence to the disease-free equilibrium when $\mathcal{R}_0 < 1$ and persistence when $\mathcal{R}_0 > 1$. This work provides a rigorous infinite-dimensional operator-theoretic framework for analyzing reproduction numbers in structured epidemic models and establishes a foundation for extensions to more general systems. - Open Access
- 2 February 2026
- Published by MDPI AG in Preprints.org
Well-Posedness of the Fisher–KPP Equation with Neumann, Dirichlet, and Robin Boundary Conditions on the Real Half Line
We consider the Fisher–KPP equation with Neumann boundary conditions on the real half line. We claim that the Fisher-KPP equation with Neumann boundary conditions is well-posed only for odd positive stationary solutions. We begin by proving that the Fisher-KPP equation with a Dirichlet boundary condition is stable, and with a Robin condition is stable only for odd positive stationary solutions. Then we inferred and proved that the Fisher-KPP equation with Neumann boundary conditions is stable only for odd positive stationary solutions. We solved the Fisher–KPP equation with Neumann boundary conditions to demonstrate the existence of the solution. In addition, we proved the uniqueness of the solution. Moreover, we proved the the solution of Fisher-Kpp equation with Dirichlet condition is stable. We also showed that the Fisher-KPP equation with Robin boundary conditions is stable only for odd positive stationary solutions. The uniqueness and existence proof of the Fisher-KPP equation with Robin condition are similar to the Neumann condition. Hence, we conclude that the Fisher-KPP equation on the real line is well-posed for the Dirichlet condition, and well-posed only for odd positive stationary solutions for both the Neumann condition and the Robin condition.
- 2 February 2026
- Published by MDPI AG in Preprints.org
Open AccessWell-Posedness of the Fisher–KPP Equation with Neumann, Dirichlet, and Robin Boundary Conditions on the Real Half Line
We consider the Fisher–KPP equation with Neumann boundary conditions on the real half line. We claim that the Fisher-KPP equation with Neumann boundary conditions is well-posed only for odd positive stationary solutions. We begin by proving that the Fisher-KPP equation with a Dirichlet boundary condition is stable, and with a Robin condition is stable only for odd positive stationary solutions. Then we inferred and proved that the Fisher-KPP equation with Neumann boundary conditions is stable only for odd positive stationary solutions. We solved the Fisher–KPP equation with Neumann boundary conditions to demonstrate the existence of the solution. In addition, we proved the uniqueness of the solution. Moreover, we proved the the solution of Fisher-Kpp equation with Dirichlet condition is stable. We also showed that the Fisher-KPP equation with Robin boundary conditions is stable only for odd positive stationary solutions. The uniqueness and existence proof of the Fisher-KPP equation with Robin condition are similar to the Neumann condition. Hence, we conclude that the Fisher-KPP equation on the real line is well-posed for the Dirichlet condition, and well-posed only for odd positive stationary solutions for both the Neumann condition and the Robin condition.
- Open Access
- 14 January 2026
- Published by Elsevier in One Health
- Vol. 22, 101324
Assessing the impact of vaccination and behavioural change on Mpox transmission in high-risk groups in the Democratic Republic of Congo using an age-structured mathematical model
- 14 January 2026
- Published by Elsevier in One Health
- Vol. 22, 101324
Open AccessAssessing the impact of vaccination and behavioural change on Mpox transmission in high-risk groups in the Democratic Republic of Congo using an age-structured mathematical model
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Mathematical Biosciences and Engineering
- Vol. 23 (3) , 547-566
Modular arithmetic as an alternative to model seasonal time-varying transmission rates: Influenza as a case study
Seasonal infectious diseases like influenza pose a recurrent challenge to public health. While compartmental models, such as the susceptible-infectious-recovered-susceptible (SIRS) framework, are standard tools, representing the time-varying transmission rate, $ \beta(t) $, in an interpretable yet effective manner remains a key challenge. Existing established alternative methods for modeling seasonality use sinusoidal forcing functions, flexible splines, etc. In this paper, we propose and apply a modular approach where $ \beta(t) $ is defined using distinct, epidemiologically intuitive seasonal rates, with smooth transitions between them. To begin, we develop this framework as a theoretical tool, demonstrating its capacity to generate realistic, recurring seasonal outbreaks under plausible parameter assumptions. We then calibrate and assess this model against real-world, monthly laboratory-confirmed influenza surveillance data from Ontario, Canada, for the pre-pandemic period of 2014–2019. A systematic optimization using a coarse grid search followed by stochastic refinement calibrates the model to the observed data. The calibrated model, featuring a mean immunity duration of approximately 235 days, achieves a strong fit with the historical case data (Pearson correlation $ r = 0.80 $). Our results demonstrate that this modular arithmetic-based framework is a practical and effective tool for modeling real-world influenza dynamics, successfully bridging the gap between theory and empirical surveillance.- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Mathematical Biosciences and Engineering
- Vol. 23 (3) , 547-566
Open AccessModular arithmetic as an alternative to model seasonal time-varying transmission rates: Influenza as a case study
Seasonal infectious diseases like influenza pose a recurrent challenge to public health. While compartmental models, such as the susceptible-infectious-recovered-susceptible (SIRS) framework, are standard tools, representing the time-varying transmission rate, $ \beta(t) $, in an interpretable yet effective manner remains a key challenge. Existing established alternative methods for modeling seasonality use sinusoidal forcing functions, flexible splines, etc. In this paper, we propose and apply a modular approach where $ \beta(t) $ is defined using distinct, epidemiologically intuitive seasonal rates, with smooth transitions between them. To begin, we develop this framework as a theoretical tool, demonstrating its capacity to generate realistic, recurring seasonal outbreaks under plausible parameter assumptions. We then calibrate and assess this model against real-world, monthly laboratory-confirmed influenza surveillance data from Ontario, Canada, for the pre-pandemic period of 2014–2019. A systematic optimization using a coarse grid search followed by stochastic refinement calibrates the model to the observed data. The calibrated model, featuring a mean immunity duration of approximately 235 days, achieves a strong fit with the historical case data (Pearson correlation $ r = 0.80 $). Our results demonstrate that this modular arithmetic-based framework is a practical and effective tool for modeling real-world influenza dynamics, successfully bridging the gap between theory and empirical surveillance. - Open Access
Place, gender, and uneven progress in pediatric and adolescent HIV across sub-Saharan Africa: a regional meta-analytic assessment (2000–2024)
Background: Marked declines in pediatric HIV have been achieved across sub-Saharan Africa, yet progress among adolescents remains uneven and strongly patterned by place. Drawing on place-based perspectives in health geography, we conceptualize Eastern and Southern Africa (ESA) and West and Central Africa (WCA) not merely as epidemiological regions, but as dis-tinct assemblages of health systems, gender norms, and historical investment trajectories shaping HIV risk and care. Using UNICEF-harmonized indica-tors from 2000 to 2024, we examined how place structures progress toward the 2030 Sustainable Development Goal (SDG 3.3) target. Methods: We pooled country-level estimates using inverse-variance random-effects meta-analysis (restricted maximum likelihood for τ 2 with Hartung–Knapp adjustment). HIV incidence and AIDS-related mortality were analysed on the log scale, and mother-to-child transmission (MTCT) on the logit scale. Outcomes were pooled by region, age group (0–14; 15–19), sex, and year. We quantified percentage change from 2010 to 2024, the achieved annualized rate of change (ARC), and the required ARC from 2024 to 2030 to achieve a 90% reduction from 2010 levels. Robustness was assessed using prespecified sensitivity analyses. Findings: Between 2010 and 2024, child HIV incidence declined by 72% in ESA and 66% in WCA, while adolescent incidence declined by 54% and 62%, respectively. In 2024, MTCT remained above elimination thresholds in both regions (9.9% in ESA; 17.6% in WCA). Adolescent girls experienced substantially higher HIV incidence than boys, with pooled female-to-male incidence rate ratios of 4.13 (95% CI: 3.32–5.13) in ESA and 4.92 (4.04–5.99) in WCA. Despite progress, achieved declines among adolescents (ARC: −5.4%/year in ESA; −6.6%/year in WCA) fall well short of the acceleration required to meet 2030 targets. Conclusion: HIV progress among children and adolescents in sub-Saharan Africa is deeply place-dependent. While ESA reflects the benefits of earlier and sustained health system investments, persistent structural and gendered vulnerabilities continue to constrain adolescent outcomes, particularly in WCA. Achieving SDG 3.3 will require place-responsive strategies that integrate gender-sensitive adolescent prevention, differentiated care, and strengthened PMTCT within the specific social, political, and health system contexts shaping risk and access. Highlights: HIV declines among children and adolescents show strong place-based patterning. Child HIV incidence declined faster in ESA than in WCA, 2010–2024. Adolescent girls had four- to five-fold higher HIV incidence than boys. MTCT remains above elimination thresholds, especially in West and Central Africa. Meeting 2030 targets requires accelerated, place-responsive strategies.
Open AccessPlace, gender, and uneven progress in pediatric and adolescent HIV across sub-Saharan Africa: a regional meta-analytic assessment (2000–2024)
Background: Marked declines in pediatric HIV have been achieved across sub-Saharan Africa, yet progress among adolescents remains uneven and strongly patterned by place. Drawing on place-based perspectives in health geography, we conceptualize Eastern and Southern Africa (ESA) and West and Central Africa (WCA) not merely as epidemiological regions, but as dis-tinct assemblages of health systems, gender norms, and historical investment trajectories shaping HIV risk and care. Using UNICEF-harmonized indica-tors from 2000 to 2024, we examined how place structures progress toward the 2030 Sustainable Development Goal (SDG 3.3) target. Methods: We pooled country-level estimates using inverse-variance random-effects meta-analysis (restricted maximum likelihood for τ 2 with Hartung–Knapp adjustment). HIV incidence and AIDS-related mortality were analysed on the log scale, and mother-to-child transmission (MTCT) on the logit scale. Outcomes were pooled by region, age group (0–14; 15–19), sex, and year. We quantified percentage change from 2010 to 2024, the achieved annualized rate of change (ARC), and the required ARC from 2024 to 2030 to achieve a 90% reduction from 2010 levels. Robustness was assessed using prespecified sensitivity analyses. Findings: Between 2010 and 2024, child HIV incidence declined by 72% in ESA and 66% in WCA, while adolescent incidence declined by 54% and 62%, respectively. In 2024, MTCT remained above elimination thresholds in both regions (9.9% in ESA; 17.6% in WCA). Adolescent girls experienced substantially higher HIV incidence than boys, with pooled female-to-male incidence rate ratios of 4.13 (95% CI: 3.32–5.13) in ESA and 4.92 (4.04–5.99) in WCA. Despite progress, achieved declines among adolescents (ARC: −5.4%/year in ESA; −6.6%/year in WCA) fall well short of the acceleration required to meet 2030 targets. Conclusion: HIV progress among children and adolescents in sub-Saharan Africa is deeply place-dependent. While ESA reflects the benefits of earlier and sustained health system investments, persistent structural and gendered vulnerabilities continue to constrain adolescent outcomes, particularly in WCA. Achieving SDG 3.3 will require place-responsive strategies that integrate gender-sensitive adolescent prevention, differentiated care, and strengthened PMTCT within the specific social, political, and health system contexts shaping risk and access. Highlights: HIV declines among children and adolescents show strong place-based patterning. Child HIV incidence declined faster in ESA than in WCA, 2010–2024. Adolescent girls had four- to five-fold higher HIV incidence than boys. MTCT remains above elimination thresholds, especially in West and Central Africa. Meeting 2030 targets requires accelerated, place-responsive strategies.
- Open Access
- 3 December 2025
- Published by Springer Nature in BMC Infectious Diseases
- Vol. 26 (1)
Pre-exposure vaccination in the high-risk population is crucial in controlling mpox resurgence in Canada
As mpox spread continues across several endemic and non-endemic countries around the world, vaccination has become an integral part of the global response
- 3 December 2025
- Published by Springer Nature in BMC Infectious Diseases
- Vol. 26 (1)
Open AccessPre-exposure vaccination in the high-risk population is crucial in controlling mpox resurgence in Canada
As mpox spread continues across several endemic and non-endemic countries around the world, vaccination has become an integral part of the global response
- Open Access
- 1 December 2025
- Published by KEAI Publishing in Infectious Disease Modelling
- Vol. 10 (4) , 1575-1591
Bayesian hierarchical modeling of Mpox in the African region (2022–2024): Addressing zero-inflation and spatial autocorrelation
Mpox remains a signi_cant public health challenge in endemic regions of Africa. Understanding its spatial distribution and identifying key drivers in high-risk countries is critical for guiding e_ective interventions. This study applies a Zero-Inated Poisson (ZIP) model with spatial autocorrelation to estimate the adjusted relative risk (RR) of Mpox incidence across 24 African countries, strati_ed by Human Development Index (HDI) levels. The model accounts for overdispersion and excess zeros by incorporating spatial random e_ects and socio-environmental covariates, and was validated through model diagnostics and sensitivity analysis, demonstrating robustness of results. Spatial analysis revealed substantial heterogeneity in Mpox incidence, with elevated risk in the Democratic Republic of Congo (DRC), Nigeria, and Central African Republic (CAR) persisting after covariate adjustment (p < 0:001). Higher HDI levels were inversely associated with Mpox risk, with HDI quintile Q4 (very high HDI) showing a signi _cant reduction (aRR = 0.431; 95 % CrI: 0.099{0.724). Protective factors in low-risk areas included increased life expectancy at birth (aRR = 0.768; 95 % CrI: 0.688{0.892), higher educational attainment (aRR = 0.774; 95 % CrI: 0.680{0.921), nonlinear increases in gross national income (GNI) per capita, and a greater density of skilled health workers (aRR = 0.788; 95 % CrI: 0.701{0.934). Conversely, higher urban density was associated with increased Mpox risk, underscoring the inuence of population clustering on transmission dynamics. Notably, statistically signi_cant elevated-risk areas persisted in endemic countries of Western and Central Africa after covariate adjustment (p < 0:001). In contrast, previously undetected risk emerged in parts of Southern and Eastern Africa post-adjustment, revealing latent patterns obscured in the crude analysis (p < 0:001). Exceedance probability maps identi_ed countries with P(RR > 1) > 0.9 as priority areas for intensi_ed surveillance and targeted intervention. These patterns were not fully explained by the included covariates, suggesting the inuence of unmeasured factors such as environmental and climate variability, zoonotic reservoirs, or human{animal interactions. Further research is needed to deepen understanding of Mpox epidemiology and support locally tailored interventions.
- 1 December 2025
- Published by KEAI Publishing in Infectious Disease Modelling
- Vol. 10 (4) , 1575-1591
Open AccessBayesian hierarchical modeling of Mpox in the African region (2022–2024): Addressing zero-inflation and spatial autocorrelation
Mpox remains a signi_cant public health challenge in endemic regions of Africa. Understanding its spatial distribution and identifying key drivers in high-risk countries is critical for guiding e_ective interventions. This study applies a Zero-Inated Poisson (ZIP) model with spatial autocorrelation to estimate the adjusted relative risk (RR) of Mpox incidence across 24 African countries, strati_ed by Human Development Index (HDI) levels. The model accounts for overdispersion and excess zeros by incorporating spatial random e_ects and socio-environmental covariates, and was validated through model diagnostics and sensitivity analysis, demonstrating robustness of results. Spatial analysis revealed substantial heterogeneity in Mpox incidence, with elevated risk in the Democratic Republic of Congo (DRC), Nigeria, and Central African Republic (CAR) persisting after covariate adjustment (p < 0:001). Higher HDI levels were inversely associated with Mpox risk, with HDI quintile Q4 (very high HDI) showing a signi _cant reduction (aRR = 0.431; 95 % CrI: 0.099{0.724). Protective factors in low-risk areas included increased life expectancy at birth (aRR = 0.768; 95 % CrI: 0.688{0.892), higher educational attainment (aRR = 0.774; 95 % CrI: 0.680{0.921), nonlinear increases in gross national income (GNI) per capita, and a greater density of skilled health workers (aRR = 0.788; 95 % CrI: 0.701{0.934). Conversely, higher urban density was associated with increased Mpox risk, underscoring the inuence of population clustering on transmission dynamics. Notably, statistically signi_cant elevated-risk areas persisted in endemic countries of Western and Central Africa after covariate adjustment (p < 0:001). In contrast, previously undetected risk emerged in parts of Southern and Eastern Africa post-adjustment, revealing latent patterns obscured in the crude analysis (p < 0:001). Exceedance probability maps identi_ed countries with P(RR > 1) > 0.9 as priority areas for intensi_ed surveillance and targeted intervention. These patterns were not fully explained by the included covariates, suggesting the inuence of unmeasured factors such as environmental and climate variability, zoonotic reservoirs, or human{animal interactions. Further research is needed to deepen understanding of Mpox epidemiology and support locally tailored interventions.
- Open Access
- 18 August 2025
- Published by MDPI AG in Preprints.org
Predictive Intelligence for Mpox Virus Control in the Democratic Republic of Congo: Opportunities, Equity, and Pathways Forward
Machine learning (ML) and predictive intelligence are increasingly recognized as transformative tools for epidemic preparedness, yet their translation into operational use in Mpox-endemic and resource-limited settings such as the Democratic Republic of Congo (DRC) remains limited. This narrative review examines emerging evidence on opportunities, challenges, and equity considerations for integrating predictive approaches into Mpox control. We highlight principles of spatiotemporal learning, hybrid and ensemble modeling, and geospatial risk mapping under conditions of sparse and heterogeneous data. Rather than benchmarking algorithms, the review emphasizes context-sensitive priorities, including covariate selection, interpretability, and alignment with One Health frameworks. Pathways forward include transparent evaluation, participatory model development, and investment in local analytic capacity. By situating predictive intelligence within DRC’s health system realities, we argue that these approaches hold promise for strengthening early warning, improving targeted interventions, and advancing equitable epidemic preparedness.
- 18 August 2025
- Published by MDPI AG in Preprints.org
Open AccessPredictive Intelligence for Mpox Virus Control in the Democratic Republic of Congo: Opportunities, Equity, and Pathways Forward
Machine learning (ML) and predictive intelligence are increasingly recognized as transformative tools for epidemic preparedness, yet their translation into operational use in Mpox-endemic and resource-limited settings such as the Democratic Republic of Congo (DRC) remains limited. This narrative review examines emerging evidence on opportunities, challenges, and equity considerations for integrating predictive approaches into Mpox control. We highlight principles of spatiotemporal learning, hybrid and ensemble modeling, and geospatial risk mapping under conditions of sparse and heterogeneous data. Rather than benchmarking algorithms, the review emphasizes context-sensitive priorities, including covariate selection, interpretability, and alignment with One Health frameworks. Pathways forward include transparent evaluation, participatory model development, and investment in local analytic capacity. By situating predictive intelligence within DRC’s health system realities, we argue that these approaches hold promise for strengthening early warning, improving targeted interventions, and advancing equitable epidemic preparedness.