The Sound of Your Footsteps can Predict for Dementia

Debayan Das (~Dexter1618)


1

Vote

Description:

Do you know how the sound of your footsteps can be used to predict the likelihood of a fatal fall? Do you know that they can also be used to predict whether you will develop dementia in the next 6 months or so?

This talk will be a gentle introduction to the world of clinical gait analysis and how your gait (a.k.a the way you walk) is a digital biomarker for predicting physical and cognitive health. We will take an strong multidisciplinary approach where we will combine know-how of clinical science, remote monitoring, assistive technology and state-of-art acoustic AI engineering to showcase how this is possible.

The topics that we will cover are as follows:

  1. Digital Biomarker Engineering : We will discuss the framework of how you can feature engineer health indictors ("biomarkers") from data collected by sensors in an IOT ecosystem like PIR, magnetic sensors and even microphones/audio recorders.
  2. Gait Analysis : We will do a crash course on Human Gait to set the foundation required to use gait data to predict for neurodegenerative conditions and cognitive decline.
  3. Acoustic Gait Analysis : We will spend majority of our time here to understand how acoustics engineering and machine learning can help in analysing human gait from just audio recordings to predict for cognitive decline discussed in #2 earlier.

To demonstrate a real life application, I will share how this concept was used to track the progression of dementia of an older adult over the past 2 years.

  1. I will showcase the tech stack to make this possible.
  2. I will showcase the challenges we faced along the way in undertaking niche research, architecture development, AI development and real-life feedback from the older adult, family members and care-givers.

Prerequisites:

There are no specific requirements for the audience other than the following:

  1. You care about using technology for social good and helping people in need, especially the elderly.
  2. You are passionate about multidisciplinary applications of AI which need you to learn know-how of other fields and practices to make it work
  3. You have a strong pull towards difficult problems with great impact potential when applied
  4. You care about "applied research" more than "academic research".
  5. You are either interested or is involved in digital healthcare sector.

Some existing understanding of Python & Machine Learning will help you grasp and put the concepts together faster.

Content URLs:

First draft of the slides for this talk are available here.

Let me know in comments below if the audience wants to cover anything specific or would like to see more code.

Speaker Info:

Debayan is the Interim Chief Data Officer at MiiCare, a Medtech company based out of London (with branches in the US, EU and India) specialising in AI-powered Virtual Wards and D2A processes for older adults in the UK. His responsibilities lie at the intersection of Clinical Data Science, Cloud Architecture Design & Implementation, Data Security and Applied Multidisciplinary Research focusing on Acoustic Machine Learning.

He is hands-on responsible for the AI and Data Infrastructures which empower older adults to live safely, happily and independently in the comfort of their own homes. After graduating from Amity University Kolkata's first batch with a B.Tech in CSE with the highest SGPA in 2019, Debayan spent the past 4 years collaborating with academic institutions and care service providers in the UK to develop and use AI for the social impact.

MiiCare UK is now launching in India to bring its state-of-the-art assistive technology to help the elderly and the vulnerable in the Indian communities to help them live their twilight years safely with dignity.

You can connect with him on LinkedIn.

PS: We are looking for individuals who are passionate in applied multidisciplinary research & engineering to grow our team in India. If you want to be part of MiiCare India, reach out to Debayan on LinkedIn.

Section: Data Science, AI & ML
Type: Talks
Target Audience: Intermediate
Last Updated: