tinyML Talks on August 24, 2021 “Edge Machine Learning for Mobile Health Technologies” by Amir Aminifar

We held our next tinyML Talks webcast. Amir Aminifar from Lund University presented Edge Machine Learning for Mobile Health Technologies on August 24, 2021.

August 24 forum

IMPORTANT: Please register here

Machine learning will be an essential component of the next-generation Internet of Things (IoT) systems, including mobile health and wearable technologies. The adoption of machine learning in such systems creates several new opportunities, e.g., real-time and early detection of health abnormalities. However, enabling machine learning in mobile health and wearable technologies also involves several challenges. In particular, such systems are extremely limited in terms of resources (processing power, communication bandwidth, memory storage, and battery lifetime) due to the requirements w.r.t. portability, wearability, and social stigma. In this talk, we discuss the main challenges facing the TinyML community and introduce a new generation of edge machine-learning techniques for such resource-constrained mobile health and wearable technologies.

Amir Aminifar is currently a WASP Assistant Professor in the Department of Electrical and Information Technology at Lund University, Sweden. He received his Ph.D. degree from the Swedish National Computer Science Graduate School, Linköping University, Sweden, in 2016. During 2016-2020, he held a Scientist position in the Institute of Electrical Engineering at the Swiss Federal Institute of Technology (EPFL), Switzerland. Amir Aminifar has been involved in several national/international projects, including the Medical Informatics Platform (MIP) of the European Human Brain Project (HBP), the ML-edge Swiss National Science Foundation (SNSF) project, the e-Glass Swiss Federal Institute of Technology (EPFL) project, and the Wallenberg AI, Autonomous Systems and Software Program (WASP). He has a history of successful collaboration with industrial companies and medical partners, including General Motors, Texas Instruments, SmartCardia, and the Lausanne University Hospital. His research interests are centered around tiny/edge machine learning on Internet of Things (IoT), mobile health (m-Health), and wearable technologies.


Watch on YouTube:
Amir Aminifar

Download presentation slides:
Amir Aminifar

Feel free to ask your questions on this thread and keep the conversation going!

1 Like

What is the advantage of e-glasses vs. e.g. a kind of “musical-headset” as carrier for the electronics?
From the technical point of view, they could be very similar. But such devices should be designed to be used in most situation. It is not very common for example to attend a meeting with headsets, while it is perfectly OK to attend a meeting with glasses, e.g. e-Glass.

You used the e-glass as wearables. Would other forms, such as hats or headbands, provide different kind of sensor data, for constructing models?
Yes, but hats and headbands are very often stigmatizing, if you wear them indoors, during meetings, etc.

How do you deal with the noisy signals on device?
It depends on the biosignal, but we consider the standard techniques for each biosignal.

Just curious to see, if the goal is just to detect the condition or to prompt for some action along with it.
Here, the goal is to detect seizures and maybe notify family members, etc.

How do the medical community - doctors, etc. react to these devices?
We have received a very positive response from the medical community in general.