● Open to research opportunities & collaborations

Hello, I'm

Tuan Anh Vu

Ph.D. Researcher in Survival Analysis & Artificial Intelligence

I develop statistical and AI models for time-to-event data—such as time to disease recurrence or equipment failure—even when follow-up is incomplete.

View my research What is survival analysis?

NameTuan Anh Vu

RolePh.D. Researcher

InstitutionPukyong National University

BasedBusan, Republic of Korea

Survival analysis in practice

What can we learn from the time before an event?

An “event” can be a disease returning, a machine failing, or a customer canceling a subscription. I develop models that estimate when these events may occur and how that timing varies across people or systems.

1Observe a starting pointFor example, treatment ends or a machine begins operating.

2Measure time to an eventWe record when the event occurs—or when observation ends without seeing it.

3Support a real decisionThe prediction can inform follow-up, maintenance, or timely intervention.

From demo data to an interpretable risk analysis

Follow the four steps below to explore the data, fit a model, and understand its output.

Step 1 of 4

Demo dataset

Breast cancer recurrence example

An educational cohort created to demonstrate the analysis workflow. The analytical goal is to predict recurrence-free survival.

…demo records

8input features

…demo events

Starting pointStudy/treatment baselineClinical variables are recorded here

→

Follow-up timetime, in daysUntil event or last contact

→

Event δ=1Recurrence-free endpoint reachedRecurrence or death

Censored δ=0Alive without recurrenceAt the last recorded follow-up

age

Age

Patient age at baseline

years

horTh

Hormonal therapy

Whether hormonal therapy was administered

yes / no

menostat

Menopausal status

Status recorded at baseline

pre / post

tsize

Tumor size

Measured primary tumor size

millimetres

tgrade

Tumor grade

Ordered pathological grade

I / II / III

pnodes

Positive nodes

Number of positive lymph nodes

count

progrec

Progesterone receptor

Tumor receptor measurement

fmol

estrec

Estrogen receptor

Tumor receptor measurement

fmol

Why ordinary regression is not enough

For δ=0, the true event time is unknown—not zero and not equal to the last follow-up. Survival methods retain this partial information instead of discarding or mislabelling it.

About

Statistical foundations.
Modern learning systems.

I am a Ph.D. student in the Department of Artificial Intelligence Convergence at Pukyong National University (PKNU), Republic of Korea. I received my B.Sc. in Computer Science from Dongguk University and my M.S. in Artificial Intelligence Convergence from PKNU.

My research lies at the intersection of survival analysis, statistical machine learning, and deep learning. I develop flexible neural models for right-censored time-to-event data, including hazard-based and accelerated failure time formulations.

I also study how survival models can be integrated with reinforcement learning to support sequential decisions. More broadly, my work combines the statistical foundations of survival analysis with the representation-learning and optimization capabilities of modern machine learning.

Research focus

I develop statistical and machine-learning methods for predicting time-to-event outcomes from complete and right-censored observations.

Survival Analysis & Time-to-Event Modeling

Estimating when events occur while correctly accounting for right-censored follow-up.

Deep Survival Analysis

Building neural hazard and accelerated failure time models for complex nonlinear patterns.

Statistical Machine Learning

Combining statistically valid objectives with modern representation learning.

Structured & High-Dimensional Data

Making robust predictions from tabular data and high-dimensional covariates.

Reinforcement Learning for Censored Data

Learning sequential treatment policies when time-to-event outcomes are censored.

Publications

2025Featured publication

CNN-Based Approaches for Various Types of Tabular Data

2025Peer-reviewed

A Multi-Modal CNN Approach for High-Dimensional Survival Data

2025Peer-reviewed

Understanding Frailty Effect in Frailty Models Using frailtyHL R Package

Selected projects

Current and completed work on building flexible models, learning from incomplete follow-up, and supporting decisions over time.

Ongoing

Deep M-Spline Models for Survival Analysis

Developing flexible neural survival models that capture nonlinear covariate effects while representing the baseline hazard or survival distribution with M-splines and I-splines.

PyTorch
M-Splines
I-Splines
PH & AFT
Maximum Likelihood

Ongoing

Reinforcement Learning for Censored Survival Outcomes

Adapting Q-learning and Bellman optimization to right-censored time-to-event outcomes for individualized sequential decision-making.

Survival Analysis
Q-Learning
AFT Models
Censoring Adjustment

Completed

CNN-Based Learning for Tabular Data

Investigated convolutional architectures and data representations for several types of tabular data; published in IEEE Access in 2025.

Python
Deep Learning
CNN
Tabular Data

Methods & tools

Programming

Python · R · SQL · LaTeX

Statistical Modeling

Cox PH · AFT Models · Frailty Models · Censored Data · Maximum Likelihood · Splines

Machine Learning & Deep Learning

PyTorch · scikit-learn · Neural Networks · CNNs · Deep Survival Analysis · Model Evaluation

Reinforcement Learning

Q-Learning · Deep Q-Learning · Bellman Equations · Sequential Decisions · Treatment Strategies

Scientific Computing

NumPy · pandas · SciPy · Simulation Studies · Reproducible Experiments

Research

Academic Writing · Literature Review · Experimental Design · Statistical Simulation · Research Visualization

Tools

Git · GitHub · Linux · Docker · VS Code · Jupyter Notebook

Journey

Research experience

Present

Ph.D. Researcher

Pukyong National University

Developing deep survival models, designing simulation studies, comparing statistical and machine-learning methods, and investigating reinforcement learning with censored outcomes.

Graduate Researcher

Pukyong National University

Research in machine learning, high-dimensional survival analysis, CNN-based modeling, and statistical survival methodology, contributing to peer-reviewed publications.

Education

Ongoing

Ph.D. in Artificial Intelligence Convergence

Pukyong National University (PKNU)

Advisor: Prof. Il Do Ha. Research in survival analysis, statistical machine learning, deep learning, and reinforcement learning for censored time-to-event data.

Completed

M.S. in Artificial Intelligence Convergence

Pukyong National University (PKNU)

Graduate research focused on machine learning and survival analysis under the supervision of Prof. Il Do Ha.

Completed

B.Sc. in Computer Science

Dongguk University

Built a foundation in computer science, programming, algorithms, data analysis, and machine learning.

Research conversations welcome

Let's talk about time-to-event data.

I welcome research collaborations in survival analysis, statistical machine learning, and methods for right-censored time-to-event outcomes.

Email me at guidetuanhp@gmail.com