Survival Analysis & Machine Learning Research Group

The Survival Analysis & Machine Learning Research Group focuses on methodology development and applied research in time-to-event modelling, statistical learning, and machine learning for censored and event-time data.

The group develops statistical methods for modelling complex survival processes, with particular emphasis on modern machine learning methods, deep survival architectures, interpretable predictive tools, and rigorous performance evaluation.

Our research integrates classical survival analysis with modern machine learning and deep learning methodologies, enabling improved modelling of heterogeneous, high-dimensional, and structured data arising in epidemiology, public health, actuarial science, and related quantitative fields.

The group supervises postgraduate students from mathematical statistics, actuarial science, data science, and related disciplines, and collaborates with researchers across South Africa and internationally.

Based in South Africa at the University of the Witwatersrand, the group maintains active collaborations across Africa, Europe, and beyond, especially in statistical machine learning, epidemiology, computational statistics, and health data science.

Under the guidance of Dr Justine B. Nasejje, students in the group work on both methodological and applied problems in modern survival modelling, preparing them to contribute meaningfully in academic, health, actuarial, and data science environments.

Research Themes

The group focuses on several interrelated themes in survival methodology and statistical learning.

Advanced Survival Methodology

• Time-to-event modelling and hazard-based inference
• Competing risks and multi-state survival processes
• Frailty, heterogeneity, and mixed-effects survival structures
• Survival analysis for longitudinal, clustered, and complex data

Machine Learning for Survival Data

• Random survival forests and ensemble-based survival methods
• Deep learning architectures for survival prediction
• Neural network methods for time-to-event modelling
• Survival prediction with high-dimensional and structured covariates

Interpretability and Model Evaluation

• Explainable machine learning for survival models
• Performance assessment using the C-index, Brier score, calibration, and related measures
• Benchmarking frameworks for classical and modern survival models
• Reproducible research workflows and software-oriented methodology

Applications in Health and Sustainability

• Maternal and child health in Sub-Saharan Africa
• Under-five mortality risk modelling
• Population health decision-support and risk stratification

Selected Articles & Research Outputs

Examples of recent and ongoing work from the group include:

• Machine Learning and Deep neural network architectures for survival probability prediction

– Whata,A., Nasejje,J.B., etal. (2025). Adapting and evaluating deep-pseudo neural network for survival data with time-varying covariates. Journal of Applied Statistics

– Nasejje, J.B., Whata, A., Chimedza, C. (2022). Statistical approaches to identifying significant differences in predictive performance between ML and statistical survival models. PLOS One.

• Machine learning approaches for modelling under-five mortality in Sub-Saharan Africa– Under-Five Mortality Risk Profiles in Sub-Saharan Africa: An Interactive R Shiny Application https://justinewits.shinyapps.io/ssa_risk_profiles/

– Nasejje, J.B., Mbuvha, R., Mwambi, H. (2022). Deep learning and random forest approach to socioeconomic drivers of under-five mortality in SSA. BMJ Open

• Interpretable survival modelling frameworks for epidemiological applications

– Kallah-Dagadu, G., Nasejje, J.B., et al. (2025). Breast cancer prediction based on gene expression data using interpretable machine learning techniques. Scientific Reports

• Benchmarking classical and machine learning survival models for population health data

– Bere, A., Maposa, I., Matsena-Zingoni, Z., Twabi, H.S., Batidzirai, J.M., Singini, G.C., Mchunu, N., Nasejje, J.B., et al. (2025). Modeling timing of sexual debut among women in Zimbabwe using a Geoadditive Discrete-Time survival approach. BMC Women’s Health

• A Springer book chapter in modern statistical and machine learning methodology

– Kgoale, T., Nasejje, J.B., Whata, A. (2024). Estimating Average and Individual Treatment Effects in the Presence of Time-Dependent Covariates. In Biostatistics Modeling and Public Health Applications, Springer, Cham

• Ongoing methodological manuscripts in survival analysis, deep learning, and predictive modelling

Collaborations

Our group collaborates with researchers and institutions including:

• Prof Marvin N. Wright — Leibniz Institute for Prevention Research and Epidemiology (BIPS), Germany

• Dr Wende Safari — Department of Health Services Research and Policy, Faculty of Public Health and Policy, London School of Hygiene & Tropical Medicine, United Kingdom.

• Gabriel Kallah-Dagadu — Senior Lecturer at Dept. Statistics & Actuarial Science, University of Ghana

• Georgios Aivaliotis — School of Mathematics, University of Leeds, United Kingdom.
• Rukia Nuermaimaiti — Teaching Fellow in Statistics, Imperial College London, United Kingdom.

• Prof Najmeh Nakhaeirad — University of Pretoria, South Africa
• Albert Whata — University of Pretoria, South Africa
• Researchers within the University of the Witwatersrand
• Collaborators across South Africa, Europe, and international statistical and machine learning networks

Examples of Postgraduate Research Topics

PhD projects

• Deep neural network architectures for time-to-event modelling
• Machine learning approaches for competing risks survival data
• Interpretable survival modelling using deep learning methods
• Statistical learning methods for heterogeneous survival data

Master’s projects

• Benchmarking machine learning survival models on population health data
• Random survival forests for child mortality risk prediction
• Deep learning methods for survival probability estimation
• Interpretation methods for survival machine learning models

Opportunities for Interested ×ãÇò¾º²ÊappÅÅÃûs

Honours, master’s, and PhD students joining our group will gain experience in both the mathematical foundations of survival analysis and the practical implementation of machine learning methods for time-to-event data.

×ãÇò¾º²ÊappÅÅÃûs are encouraged to work on projects that balance theory and application, often with opportunities to engage with health, epidemiological, and population-based datasets. The group values careful methodology, reproducible implementation, and research that contributes both statistically and substantively.

×ãÇò¾º²ÊappÅÅÃûs may also benefit from collaboration across disciplines and, where possible, engagement with international research partners.

What Can You Gain?

Our group is led by researchers with expertise in survival analysis, statistical learning, and machine learning. We welcome students interested in both the theoretical and applied sides of statistical research.

×ãÇò¾º²ÊappÅÅÃûs in the group can expect to:

• build strong foundations in modern survival analysis,
• gain experience with machine learning and deep learning for censored data,
• contribute to research with methodological and real-world relevance,
• engage in collaborative work within South Africa and internationally, and
• prepare for careers in academia, actuarial science, health analytics, and data science.

Further Details

For further reading, prospective students may consult the relevant University of the Witwatersrand pages for:

• funding opportunities and scholarships,
• the Wits postgraduate application process.

Prospective students interested in joining the group are encouraged to contact:

Dr Justine B. Nasejje

University of the Witwatersrand

justinenasejje@gmail.com

Prospective students should send:

• copies of their academic transcripts,
• a one-page motivation describing their interest in survival analysis or statistical machine learning, and
• any initial ideas towards a possible research topic and focus.