Master’s in Edge AI Engineering for Health & Wearables
Online master's programme of 1,900 hours in edge AI engineering for health and wearables. 12 modules covering clinical data, lightweight models, quantisation and pruning, embedded frameworks, energy efficiency, federated learning, embedded security, MDR and AI Act regulation and edge MLOps, with an assigned technical tutor.
Master’s in Edge AI Engineering for Health & Wearables — studying from Australia
Health technology and biomechanics are growing areas in Australia, supported by a strong medical research sector and an established medical device industry. Wearable devices and occupational safety systems are increasingly used in mining, construction and healthcare. Medical devices are regulated nationally, requiring engineers to understand conformity, risk management and clinical evaluation. Employers include medical device manufacturers, hospitals, research institutes such as those affiliated with universities, and workplace safety consultancies.
Fees for Australia: A$11,000 (AUD), with instalments and employer sponsorship available.
Relevant bodies in Australia
Typical roles
Questions from Australia
Will a SEIUM qualification be recognised by Engineers Australia or other professional bodies for this area?
SEIUM qualifications are private qualifications and are not accredited by Engineers Australia or any other national body. Recognition is decided by individual employers and, where relevant, by the professional body assessing your application. Engineers Australia assesses skills and experience independently, so check directly with the body you hope to join before enrolling.
Which Australian employers might value this training in health, biomechanics or wearable safety work?
Employers in Australian medical device firms, rehabilitation services, workplace health and safety teams and wearable technology companies may consider advanced engineering training when hiring. Whether they value a private qualification depends on each employer's own policy and on your existing experience. Speaking with prospective employers about how they view private qualifications is a sensible step.
Can I study this alongside full-time work in Australia, and is employer funding possible?
The programmes are 100% online and taught in English, so you can study from any time zone, including Australian time zones, around your work hours. Fees are payable in instalments, and employer sponsorship is possible; ask your employer whether they fund training. Plan for the workload of 300 to 1,900 hours depending on the programme, and confirm the schedule with SEIUM.
Take the controls.
Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.
What you will be able to do
Design edge AI architectures for healthcare applications.
Prepare clinical wearable data with appropriate partitioning and labelling.
Train and compress lightweight models through quantisation and pruning.
Deploy models on microcontrollers and embedded accelerators.
Optimise the energy consumption of inference on portable devices.
Apply federated learning and privacy techniques to health data.
Meet the requirements of the MDR and the European AI Act.
Who it's for
- Embedded machine learning engineers.
- Firmware developers for medical devices.
- Clinical data scientists moving towards on-device inference.
- MLOps engineers working with device fleets.
- Specialists in healthcare AI regulation.
- IoMT systems architects.
- Engineering or science graduates looking to enter digital health.
What you'll work with
- Regulation (EU) 2017/745 (medical devices)
- Regulation (EU) 2024/1689 (artificial intelligence)
- IEC 62304 (medical device software life cycle)
- ISO 14971 (risk management)
- IEC 81001-5-1 (cybersecurity of health software)
- MDCG 2019-16 (cybersecurity guidance)
- Regulation (EU) 2016/679 (GDPR) (data protection)
- Good Machine Learning Practice (FDA, Health Canada, MHRA) (guiding principles for medical ML)
- TensorFlow Lite Micro
- Edge Impulse
- STM32Cube.AI
- PyTorch
- Flower
12 modules · 1900 h
1.1Edge, fog and cloud in healthcare1.2Latency, privacy and cost advantages1.3Memory and energy constraints1.4Hardware: embedded MCUs, NPUs and GPUs1.5Use cases in clinical wearables1.6Reference architectures1.7Life cycle of embedded models1.8The TinyML landscape1.9Comparison of processors for embedded inference1.10Case study: edge versus cloud decision
2.1Wearable signals and their nature2.2Clinical labelling and experts2.3Public PhysioNet databases2.4Class imbalance2.5Data quality and artefacts2.6Patient-wise partitions2.7Time-series data augmentation2.8Documentation with datasheets2.9Demographic bias in training data2.10Practical exercise: preparing an arrhythmia dataset
3.1Compact 1D convolutional networks3.2Depthwise separable networks3.3Lightweight recurrent models3.4Embedded decision trees and boosting3.5Neural architecture search3.6Clinical metrics: sensitivity and specificity3.7Probability calibration3.8Robust cross-validation3.9Anomaly detection with autoencoders3.10Laboratory: compact CNN for fall detection
4.1Post-training quantisation4.2Quantisation-aware training4.3Structured and unstructured pruning4.4Knowledge distillation4.5Weight compression and encoding4.6Latency and memory profiling4.7Accuracy versus resources trade-off4.8Numerical equivalence verification4.9Model formats and ONNX conversion4.10Practical exercise: INT8 quantisation with TensorFlow Lite
5.1TensorFlow Lite Micro5.2CMSIS-NN for Cortex-M5.3ONNX Runtime and TVM5.4NPU accelerators such as Ethos-U5.5NVIDIA Jetson for imaging5.6STM32 and nRF microcontrollers5.7Code generation with STM32Cube.AI5.8Benchmarking with MLPerf Tiny5.9Image inference on portable devices5.10Laboratory: deployment on a Cortex-M board
6.1Device energy budget6.2Event-driven inference6.3Duty cycles and low-power modes6.4Measuring real power consumption6.5Energy harvesting6.6Always-on coprocessors6.7Sampling frequency trade-offs6.8Battery life and user experience6.9SRAM and flash memory in microcontrollers6.10Practical exercise: power profile of a model
7.1Atrial fibrillation on a microcontroller7.2Epilepsy monitoring7.3Hypoglycaemia prediction7.4Fall detection in older people7.5Sleep and apnoea analysis7.6Medical imaging on portable ultrasound devices7.7Rehabilitation with inertial sensors7.8Clinical alarms and alarm fatigue7.9Early detection of respiratory exacerbations7.10Case study: offline fall alarm
8.1Fundamentals of federated learning8.2Aggregation and data heterogeneity8.3Differential privacy in gradients8.4Homomorphic encryption and secure computation8.5Membership inference attacks8.6GDPR and on-device processing8.7On-device personalisation8.8Governance of distributed models8.9On-device continual learning8.10Laboratory: federated simulation with Flower
9.1Secure boot and signed firmware9.2Secure OTA updates9.3Protecting the model against extraction9.4Adversarial attacks on signals9.5IEC 81001-5-1 in health software9.6MDCG 2019-16 cybersecurity guidance9.7Vulnerability management9.8SBOM and third-party components9.9Radio firmware and process isolation9.10Workshop: threat model of a wearable
10.1Regulation (EU) 2017/745 and software10.2Regulation (EU) 2024/1689 on AI10.3High-risk systems10.4IEC 62304 and software classes10.5ISO 14971 applied to AI10.6Model changes and change control10.7GMLP principles from the FDA, Health Canada and MHRA10.8Technical documentation of the algorithm10.9Transparency and information for clinical users10.10Case study: predetermined change control plan
11.1Data and model versioning11.2Continuous integration for firmware11.3Hardware-in-the-loop testing11.4Field monitoring11.5Data drift detection11.6Feedback from clinical users11.7Device fleet management11.8Safe withdrawal and rollback11.9Audit logs and model traceability11.10Practical exercise: MLOps pipeline with OTA deployment
12.1Clinical problem and edge constraints12.2Data selection and labelling12.3Baseline model design12.4Optimisation and embedded deployment12.5Performance and power measurement12.6Clinical evaluation of metrics12.7Security and privacy analysis12.8Regulatory fit of the algorithm12.9Thesis report12.10Master's Thesis: defence before a panel
How you'll learn
Training is 100% online on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. There are no fixed timetables: you progress at your own pace within each intake, so the programme is compatible with full-time work.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is TensorFlow Lite Micro, Edge Impulse, STM32Cube.AI, PyTorch and Flower, in free, academic or evaluation versions. Testing is studied through real procedures: model deployment on microcontrollers, power consumption measurements on development boards and federated learning simulations with public clinical data. Assessment is continuous by module, plus the final Master's Thesis.
Standards covered: Regulation (EU) 2017/745, Regulation (EU) 2024/1689, IEC 62304, ISO 14971, IEC 81001-5-1, MDCG 2019-16, Regulation (EU) 2016/679 (GDPR), Good Machine Learning Practice (FDA, Health Canada, MHRA).
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Where this takes you
Requirements
- University degree in engineering, science or architecture, or equivalent demonstrable professional experience.
- Internet connection and a computer capable of running the programme software.
- English sufficient to read technical standards.
Ready to start?
Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.
Frequently asked questions
The programme totals 1,900 hours over 12–14 months. You set your own pace within each intake, and your tutor helps you plan the workload around your job.
You work with TensorFlow Lite Micro, Edge Impulse, STM32Cube.AI, PyTorch and Flower, using free, academic or evaluation versions. Your tutor indicates the free alternative wherever one exists.
A SEIUM Institute diploma, awarded on passing the continuous assessment and the Master's Thesis. It is SEIUM's own qualification, not an official state degree; its recognition depends on the programme completed and the regulations of each country.
Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the higher programme. Ask about the recognition policy when you enrol.
A university degree in engineering, science or architecture (or equivalent professional experience), a computer able to run the programme software and English sufficient to read technical standards.
Start dates are set by SEIUM's faculty for each intake. Our admissions team will confirm the next available date when you apply.
Related programmes
Master’s in Thermal Comfort Engineering
Online master's programme of 1,900 hours in Thermal Comfort Engineering, with 12 modules covering human thermophysiology, comfort indices, thermal environment measurement, heat and cold stress, comfort…
Master’s in Biosensors and Biosignal Processing Engineering
Online master's programme of 1,900 hours in biosensor engineering and biosignal processing. 12 modules covering electrochemical and optical biosensors, biopotential acquisition, ECG, EEG, EMG and PPG…
Master’s in Medical Hardware Design and Safe Electronics Engineering
Online master's programme of 1,900 hours in medical hardware design and safe electronics engineering. 12 modules covering IEC 60601-1 electrical safety, EMC, analogue design for biomedical…
More in Health, Biomechanics, Wearables & Operational Safety
All Health, Biomechanics, Wearables & Operational Safety programmes → · How to apply · Fees & financing · Scholarships