CityUHK BMS5010: Artificial Intelligence in Health Science Research and Management — course overview with AskSia
BMS5010 Artificial Intelligence in Health Science Research and Management is a 3-credit-unit postgraduate course from the Department of Biomedical Sciences at City University of Hong Kong, offered in Semester A 2026/27 as one of the eight core courses (students choose six) of the MSc in Health Sciences and Biomedicine. A background in Python programming and data visualization is listed as preferred, not required.
Sia generates BMS5010 practice questions, walks through principles of ai; machine learning and model training step by step, and quizzes you on the material the report that weights most heavily.
Find what is wrong
A student trains a model to predict disease from omics data and reports 99% accuracy. Which line of the workflow is the bug?
X_train, X_test = split(X)
scaler.fit(X) # fit on all samples
model.fit(scaler.transform(X_train), y_train)
acc = model.score(scaler.transform(X_test), y_test)The syllabus lists model training, testing and validation as a core topic: the test set must stay unseen by every step of fitting.
Fit the scaler on X_train only, then transform both sets.
Choosing the metric matters too, but it is not the bug in this code.
The trap: Preprocessing counts as fitting. Anything that looks at the test samples before evaluation leaks. classic slip!
One report decides 40% of your grade. Group; reported GenAI ratio must be below 20%. This whole page is built around that.
Overview
What BMS5010 is, and where it sits
BMS5010 Artificial Intelligence in Health Science Research and Management is a 3-credit-unit postgraduate course from the Department of Biomedical Sciences at City University of Hong Kong, offered in Semester A 2026/27 as one of the eight core courses (students choose six) of the MSc in Health Sciences and Biomedicine. A background in Python programming and data visualization is listed as preferred, not required.
The syllabus takes a project-based approach that pairs theory with hands-on practice. Its topics are foundational AI concepts such as machine learning and deep learning, computer vision models, language models, graph models, AI for multi-omics data analysis, drug discovery, and disease diagnosis and prognosis. Lectures (two hours a week) introduce each health science problem and its AI applications; practical labs (one hour a week) work through curated datasets, optimising, evaluating and interpreting the models and comparing algorithms.
Assessment is 100% continuous: assignments 25%, a mid-term examination 25%, a final group project presentation 10% and the final group project report 40%. GenAI use is permitted with declaration on the assignments and project but not in the mid-term, and the report's reported GenAI ratio must stay under 20%.
Always treat your own course outline and the exam timetable as authoritative.
Difficulty & time commitment
Is BMS5010 hard, and how much time does it take?
BMS5010 is manageable if you keep a weekly rhythm and treat the back half as the main event. The pattern is consistent: it starts gently and steepens, and the heaviest assessment is the part that separates grades.
The difficulty curve and the assessment weighting point the same way: the back half is harder and worth more. Front-loading effort there is the highest-return decision in the course.
Is this course for you
Who tends to do well, and who tends to struggle
You will likely do well if
- You can already write a Python script that loads a dataset and plots it.
- You want to evaluate AI methods on biomedical data, not only read about them.
- You work well in a group with a shared codebase.
You may struggle if
- You expect a survey course; each lab asks you to run and compare models.
- You let the group carry the code; the assignment rubric says you must demonstrate it yourself.
- You paste GenAI output into the report without understanding it; the ratio is checked and the code is questioned.
- Keep a notebook that reproduces every lab end to end; it is your assignment evidence and your mid-term revision.
- Choose the group's health problem and dataset by week three and benchmark two methods early.
- Write the report section by section against the five rubric criteria.
Syllabus
The 10 topics, topic by topic
Topics from the public source. Their order is a content map, not an official weekly timetable or published exam weighting.
T1 · Principles of AI; machine learning and deep learning
Keyword syllabus; CILO 1The foundational concepts and their inspiration from health and biomedical science.
T2 · Model training, testing and validation
Keyword syllabus; CILO 3How a model is fitted, checked and kept honest.
T3 · Computer vision and convolutional neural networks
Keyword syllabusImage models for health science problems.
T4 · Language models
Keyword syllabusText models and their use in health research and management.
T5 · Graph models and graph neural networks
Keyword syllabusLearning on networks and molecular graphs.
T6 · AI for multi-omics data analysis
Course abstract topic 5Applying the methods to genomic and other omics data.
T7 · AI in drug discovery
Course abstract topic 6Models that screen and prioritise candidate compounds.
T8 · AI for disease diagnosis and prognosis
Course abstract topic 7Prediction of diagnosis and outcome from clinical and biomedical data.
T9 · Benchmarking and interpreting AI methods
CILO 3; project rubricComparing algorithms on curated datasets and reading the results critically.
T10 · Practical labs on curated datasets
Learning and teaching activity 2Optimising, evaluating and interpreting models; the practical guide for the assignments and the group project.
How it's assessed
Assessment structure
| Component | Weight | Format & timing |
|---|---|---|
| Assignments (run and demonstrate analysis code in practical labs) | 25% | Run and demonstrate the analysis code and results from the practical labs; GenAI may be used for literature searches, language and learning or debugging code, declared in the submission. Across the semester. Continuous; students are assessed on whether they understand the submitted code. |
| Mid-term examination | 25% | Mid-term examination on the principles and subject matter of the lectures. Mid-semester. Continuous assessment; GenAI not allowed. |
| Final group project presentation | 10% | Group presentation of the project in English with structured slides, answering questions and raising them on others' work. End of semester. Group. |
| Final group project report | 40% | Group report: state a health science problem and its datasets, provide runnable code for the chosen AI methods, benchmark them with interpretable tables or figures, and discuss pros and cons critically. End of semester. Group; reported GenAI ratio must be below 20%. |
- 100% continuous assessment: assignments 25%, mid-term examination 25%, final group project presentation 10%, final group project report 40%. GenAI use is permitted with declaration on the assignments and the project (students are assessed on understanding of submitted code), not permitted in the mid-term examination, and the group report's GenAI ratio as reported by Turnitin must be below 20%.
- A mid-term examination worth 25%, taken without GenAI, on the principles and subject matter of the lectures: the criterion is the ability to analyse, state and apply what was taught. There is no final examination; the semester ends with the group project presentation and report.
This is a coursework course. Coursework carries 75% of the grade and the final group project report is the single heaviest piece at 40%, so steady work across the semester decides your result more than any one sitting. Group; reported GenAI ratio must be below 20%.
Final exam timing: No final examination; the mid-term examination is part of the continuous assessment. Confirm the exact date and venue on your exam timetable.
How to actually pass it
A weekly rhythm, two checklists, and the traps to avoid
The course rewards consistency over cramming, and practice over re-reading. Here is the loop that works, then what to have nailed before each exam.
The weekly loop
The mistakes that cost marks
Formula & concept sheet
The vocabulary and formulas you must own
- Machine learning
- Algorithms that learn patterns from data rather than being explicitly programmed.
- Deep learning
- Machine learning with multi-layer neural networks.
- Convolutional neural network
- A neural network architecture for images and other grid data.
- Graph neural network
- A neural network that learns on nodes and edges, such as molecules or interaction networks.
- Language model
- A model trained on text that can represent or generate language.
- Validation set
- Data held out from training to tune and check a model.
- Multi-omics
- Combined genomic, transcriptomic, proteomic and other molecular data.
- Prognosis model
- A model predicting the likely course or outcome of a disease.
Where it fits
Prerequisites, related courses & why it matters
No prerequisites; precursor: a background in Python programming and data visualization is preferred. 3 credit units, P5; offered Semester A 2026/27; one of the MSHSB's eight core options.
Your BMS5010 study toolkit
Study the course with Sia, not just read about it
Use your permitted course notes with these study tools. Follow your course’s rules for AI and assessed work.
FAQ
Frequently asked questions
Is BMS5010 hard?
Moderate on the six-factor rubric, level with the HK anchor: the content is applied machine learning in Python, but the assessment is spread across labs, a mid-term and a group project with a published rubric.
What is the assessment breakdown?
Assignments 25%, mid-term examination 25%, final group project presentation 10%, final group project report 40%, per the CityU course syllabus; 100% continuous assessment.
Do I need to code?
Yes, in practice. A background in Python programming and data visualization is listed as preferred, the labs are code-based, and the group report must include runnable code.
Can I use GenAI?
On assignments and the project, yes, with a declaration of where and how, and you are assessed on whether you understand the submitted code; not in the mid-term; and the report's reported GenAI ratio must be under 20%.
Is it core?
It is one of the eight programme core courses of the MSc in Health Sciences and Biomedicine, of which students choose six; it is offered in Semester A.
Is there a textbook?
No compulsory reading. Additional readings are Géron's Hands-On Machine Learning, Deep Learning for the Life Sciences, Witten and Frank's Data Mining, and Bishop's Pattern Recognition and Machine Learning.
Study BMS5010 with Sia
Practise explaining the course concepts and check your reasoning. Confirm assessed-work requirements in your current outline.
Start studying with Sia
