PHM5010 – Precision Biomarkers
Course Overview
Biomarkers, with potential to inform disease risk, diagnosis, prognosis, and treatment response are foundational to Precision Medicine because they enable evidence-based clinical decision-making and more targeted therapies. PHM5010 guides students through the end-to-end biomarker journey—from discovery to development, validation, and real-world clinical translation. Building on prior training in genomics, proteomics/metabolomics, biostatistics, and AI/ML in Semester 1 of the MScPHM programme, students will learn how big-data omics and multi-omics approaches can be used to identify and evaluate clinically meaningful biomarkers.
The course also highlights the realities of translation: what it takes to demonstrate analytical validity, clinical validity, and clinical utility, and why challenges such as reproducibility, bias, heterogeneity, and implementation barriers often prevent promising biomarkers from reaching practice. Learning is complemented by clinical and industry perspectives, including expert-led sessions and site visits that expose students to biomarker development in real healthcare and biotech settings.
A major component of PHM5010 is a portfolio-building individual assignment, where students develop a manuscript-style biomarker discovery study (data/AI-driven) or an AI-augmented biomarker review, supported by structured milestone presentations and mentorship.
Learning Outcomes
By the end of the course, the student will:
1) Differentiate the major classes, characteristics and intended uses of biomarkers
2) Formulate a clinically relevant biomarker question and design an appropriate discovery and development workflow that aligns the clinical problem, population, specimen type, assay platform and study design.
3) Apply and appraise AI- and ML-enabled approaches used in biomarker science, including AI assisted evidence review, feature selection, predictive modelling, hyperparameter tuning, validation and performance evaluation.
4) Evaluate biomarker evidence using the frameworks of analytical validity, clinical validity and clinical utility, and assess whether a candidate biomarker is sufficiently robust for translational or clinical use.
5) Interpret biomarker signatures in biological and clinical context by integrating in silico evidence, explainable AI principles, and considerations of reproducibility, generalizability and bias.
6) Compare major multi-omics and integrative modelling approaches for biomarker discovery, including supervised and unsupervised methods, latent-factor models, network-based integration and digital-twin frameworks.
7) Assess the scientific, operational and regulatory barriers to biomarker translation, including cohort heterogeneity, confounding, assay variability, overfitting, interpretability and implementation constraints.
8) Produce and communicate a portfolio-ready biomarker project in the form of a manuscript-style ML/DL AI-based discovery study, supported by structured milestone presentations and feedback.
Course Outline
PHM5010 covers both fundamentals and current frontiers in precision biomarker science, including:
1) Fundamentals of Precision Biomarkers
a. Characteristics of an ideal clinical molecular biomarker
b. Classification of biomarkers (risk, diagnostic, prognostic, predictive, monitoring, etc.)
c. Principles of clinical interpretation and decision impact
2) Biomarker Discovery (Big Data + Omics Focus)
a. Omics-driven biomarker discovery workflows
b. Public datasets and discovery study design
c. Statistical and AI/ML approaches for feature selection and validation
3) Biomarker Development and Translation
a. From candidate biomarker to clinical-grade tool
b. Analytical validity, clinical validity, and clinical utility
c. Translation of omics-based biomarkers into clinical practice
4) Multi-Omics and Emerging Directions
a. Multi-omics biomarker discovery: rationale, key steps, and pitfalls
b. Digital twin concepts for multi-omics biomarker discovery
c. Current state of omics-based biomarkers in clinical use
5) Challenges and Real-World Adoption
a. Reproducibility and generalizability
b. Cohort heterogeneity and population bias
c. Regulatory, operational, and workflow barriers in implementation
6) Translational Exposure
a. Experiential learning to connect classroom concepts to real translational settings (e.g clinical lab, biomarker startup and multinational pharma company.
Course Requirements
To succeed in PHM5010, students are expected to have a strong foundation in the programme’s Semester 1 core modules, specifically:
· PHM5001 Precision Human Genomics
· PHM5002 Precision Proteomics and Metabolomics
· PHM5003 Big Data Statistical Analyses for Precision Medicine
· PHM5005 AI and Machine Learning for Precision Medicine
Students should also concurrently read PHM5004 (High Performance Computing for Precision Medicine) to support computational workflows relevant to omics and large-scale biomarker analysis.