In this blog, we discuss how University of Technology Sydney (UTS) took on the challenge of at-home rehabilitation for stroke patients using biomedical engineering and robotics, and AWS solutions such as AWS Internet of Things (IoT).

Stroke is the third most common cause of death in Australia and a leading cause of disability according to the Australian Brain Foundation. A stroke occurs when blood clots or broken blood vessels cut off the blood supply to the brain.

Rehabilitation helps someone who has had a stroke relearn skills that are lost when part of the brain is damaged. The goal of rehabilitation is to improve or restore speech, cognitive, motor, or sensory skills.

Most patient rehabilitation programs are delivered as inpatient, outpatient, or at skilled nursing facilities. These delivery methods are expensive, can mean absences from home, or require patients to travel long distances.

At-home rehabilitation solves some of the obstacles of facility-based programs but there are other challenges with these programs such as the need for expensive, specialised equipment, lack of real time monitoring, and immediate therapist feedback.

Students at the School of Biomedical Engineering at UTS decided to use robotics and cloud technology to overcome the challenges of at-home rehabilitation. Their goals were to deliver a low-cost solution which provides therapeutic movement training, with real time remote monitoring and feedback capability.

Improving at-home rehabilitation with Franky, a robotic exoskeleton arm

The UTS solution is “Franky,” a robotic exoskeleton arm made of 171 carbon-reinforced pieces printed at the UTS – ProtoSpace. Franky is combined with a host of sensors connected to a touchscreen interface with wireless communication capabilities.

Franky is worn as an exoskeleton that provides physical assistance to guide patients through therapeutic movements. Patients can use Franky without rehabilitation experts needing to be physically present. The exoskeleton’s sensors can read the patient’s movements by measuring the brain’s electrical signals. Franky then relays this data to rehabilitation experts who can provide real-time feedback.

Franky can detect if they are not performing exercises correctly and will demonstrate the correct way to perform these movements.

“We’re using an industrial level communication interface that can receive around 100 different inputs and process them with machine learning,” says Kairui Guo, Chief Technology Officer of the program

Seeing Franky in action

To control Franky, the biomedical engineering team uses a Raspberry Pi, as it provides high performance edge computing at low cost. AWS IoT Core connects the Raspberry Pi to the cloud and collects near-real time data as the patient exercises, sending the information to Amazon Simple Storage Services (S3). The data loads using Amazon Kinesis Data Firehose, a simple and reliable way to load streaming data into S3.

If you want to see more of Franky refer to this video  –

A fast retrieval database is critical for the rehabilitation experts to access the data that patients generate during rehabilitation. Amazon DynamoDB  provides fast, flexible NoSQL database services at single-digit millisecond performance at any scale. Rehabilitation experts can monitor and communicate with a patient from anywhere in the world using a website hosted on S3 and Amazon CloudFront for low latency content delivery.

The following diagram provides a high-level overview of the AWS services used:


AWS IoT Core allows for remote management message processing, connection recovery, and shadow state. The end result is a simple and stable architecture that allows the biomedical engineering team to deliver an at-home, low-cost technology solution that improves stroke patient recovery time.

“With AWS IoT Core, the team did not have to spend time and engineering resources on building an IoT solution. They could focus on improving patient outcomes,” said Kairui Guo, Chief Technology Officer for the program.

In early trials, Franky is showing the benefits of using AWS IoT by saving patients travelling time, improving data collection, and avoiding physical contact with patients during the pandemic.

Next steps

The next phase of the journey will be for clinical trials of Franky to evaluate the efficacy and adoptability of the technology in an at-home environment. Trial data will be analysed to measure the improvement in recovery time and patient adoption of Franky.

In 2022 the biomedical engineering team intends to incorporate AWS IoT Greengrass and AWS IoT Greengrass ML Inference. AWS IoT Greengrass will allow the team to use machine learning models built, trained, and optimised in the cloud and then run inference locally on devices. Data gathered from the inference running on AWS IoT Greengrass can be sent back to Amazon SageMaker where it can be tagged and used to continuously improve the quality of machine learning models.

Learn more about building connected device solutions using AWS IoT.

About the Author

Nils De Vries is a Solutions Architect with Amazon Web Services. He works with higher education customers in Australia, helping them adopt cloud technology to build scalable and secure solutions using AWS. In his spare time, he spends outdoors riding his mountain bike, and has a goal to automate everything in his house.

Leave a Reply