Application of Deep Learning for Healthy Aging at Work

Author: Dr Petre Lameski STSM Period: 2019-04-20 – 2019-04-30

ECI: Yes

Hosting institution: New Bulgarian University. Sofia, Bulgaria

From ITC: No

 

Summary 

Aim and motivation:

Deep learning algorithms (DL) have been successfully used in different scientific areas and have proven to be very reliable when doing classification or predictive analysis on the data. DL has become a standard for many industrial applications such as image recognition, image detection, etc. The application of DL for healthy ageing is also becoming more and more popular in recent years however on the large scale, very few articles have been published that contain both «healthy ageing» and «deep learning» according to Google Scholar. «Healthy ageing at work» and «deep learning» on the other hand, yields 0 results.

 

This gives the motivation to research the existing approaches that include Deep learning and healthy ageing and try to find the most suitable to be applied for the purpose of healthy ageing at work. This brings us to the goal of this STSM: to research the available data and DL approaches and applications towards healthy ageing with a special accent to healthy ageing at work.

Proposed contribution to the scientific objectives of the Action:

The work done during this STSM is closely related to the RCO2 of the action (To design and create innovative ICT solutions that will be integrated into Smart Support Furniture and habitat environments)

The Deep learning algorithms are used for activity detection, recognition, health state estimation, ageing factors, etc, which are all important part of the decision support processes for the healthcare and the healthy ageing of the older population. The research performed during this STSM will improve the existing knowledge about DL approaches for their later application in ICT solutions for smart furniture and habitat especially for healthy ageing at work.

This STSM will increase collaboration and give an opportunity for joint paper publication between the host and the home institution personnel and other interested colleagues within and outside the SHELDON COST Action. For this, it is also closely related to RCO4 (To ensure dissemination, evaluation, and exploitation of the Action’s results together with establishing a strong network with the relevant industrial stakeholders), which goal will be further empowered by the joint publication.

Techniques:

The first part of the ST

SM will be used to investigate the literature and generate a short survey of related work regarding deep learning algorithms used for applications that consider healthy ageing in general. Also, available datasets will be considered. If no available datasets exist, a method for obtaining such datasets will be discussed and plans for further collaboration on this matter will be considered.

During the second part of the STSM, initial preparations such as code for DL algorithms will be done and initial experiments will be performed. The results will be presented and discussed in the hosts and preparations for publication will be made

The STSM will have a duration of 11 days with the following Plan:
1. Investigating literature for existing approaches for DL in healthy ageing and healthy ageing at work (2 days)
2. Obtaining available data (if available) or creating an experimental scenario for obtaining data about the ageing population at work (2 days)
3. Developing approaches, performing initial experiments and discussing existing approaches and obtained results compared to implemented experiments and results (5 days)
4. Preparation for publication of the results (2 days)

The results of this work will be published in conference, book chapter or journal and should be considered a deliverable of the action.

Leave a comment