Keynote talk at IEEE Conference on Intelligent Methods, Systems, and Applications (IMSA), Cairo, Egypt, July 2023

1. User-Centric AI Analytics for Chronic Health
Conditions Management
Professor Aladdin Ayesh
School of Natural and Computing Sciences
University of Aberdeen
Aberdeen, UK
aladdin.ayesh@abdn.ac.uk
https://orcid.org/0000-0002-5883-6113
Keynote talk at IEEE Conference on Intelligent Methods,
Systems, and Applications (IMSA), Cairo, Egypt,
July 2023

2. User-Centric AI Analytics for Chronic Health Conditions Management
Outline
Introduction
Chronic Health Conditions
Nutrition Related
Physiological Related
Mental Related
Managing Chronic Conditions
In educational context
Ageing population and smart cities context
Personalising AI Models
What is next?
Conclusion
2 / 25

3. User-Centric AI Analytics for Chronic Health Conditions Management
Introduction
Introduction
The use of AI analytics in health applications has seen a rapid
growth in recent years. This is helped by advances in two areas:
▶ advances in sensor technologies including the wide spread use
of wearable devices, e.g. smart watches, which enable
continuous data gathering with minimum disruption to daily
life.
▶ advances in data analytic algorithms, e.g. deep learning, and
high performance computers that allowed the processing of
large datasets at speed that was not imaginable a decade ago.
3 / 25

4. User-Centric AI Analytics for Chronic Health Conditions Management
Introduction
Introduction
In this talk, we look at AI analytics use in managing chronic health
conditions with examples from my team projects.
Focus on the challenges in managing these conditions due to the
variations in individual circumstances.
These variations directed the research into user-centric approach
leading to variety of research questions. In the following sections,
we explore examples from our research work in this area.
Conclude with what, in our opinion, to be the next steps and
some remaining open research questions.
4 / 25

5. User-Centric AI Analytics for Chronic Health Conditions Management
Chronic Health Conditions
Chronic Health Conditions
Chronic health conditions can cover a wide range of diseases and
disorders, which share some common characteristics:
▶ improvement in treatments and health care turned these
diseases from terminal or seriously impairing to manageable
conditions.
▶ require active management plans to control their effects and
to maintain quality of life for the sufferer.
▶ their impact and re-occurring manifestation may differ from
person to person depending on several factors including
personal and environmental.
▶ have great psychological and emotional impact on their
sufferers with a wide spectrum of variations.
5 / 25

6. User-Centric AI Analytics for Chronic Health Conditions Management
Chronic Health Conditions
Nutrition Related
Nutrition Related
There is a growing host of applications aimed towards managing
nutrition related chronic health conditions, e.g. Diabetes Type 2.
However, many of these applications lack:
▶ Sufficient nutritional research limiting their scientific
validation for long term effectiveness.
▶ comprehensive user models for personalisation, applications
are often limited to preference models if any.
6 / 25

7. User-Centric AI Analytics for Chronic Health Conditions Management
Chronic Health Conditions
Nutrition Related
Nutrition Related
A key challenge in this area is capturing the right amount and type
of data from users and environment with minimum disruption to
daily life.
Techniques we have been exploring combine the use of
gamification, wearable sensors, and self-reporting, in addition to
nutritional dictionaries. The current work we are doing in this area
is still at early stages and follows multiple streams.
7 / 25

8. User-Centric AI Analytics for Chronic Health Conditions Management
Chronic Health Conditions
Physiological Related
Physiological Related
Other physiological health conditions may require more user-centric
approaches to fuse multi-modal data and provide explanations [1].
8 / 25

9. User-Centric AI Analytics for Chronic Health Conditions Management
Chronic Health Conditions
Mental Related
Mental Related
There is a growing recognition of the potential impact of AI
technologies on mental health [2, 3, 4] and well-being [5]. If this
impact is well managed, it has a huge potential in addressing a
large number of mental health issues and improving quality of life
for millions on a daily basis.
9 / 25

10. User-Centric AI Analytics for Chronic Health Conditions Management
Managing Chronic Conditions
Managing Chronic Conditions
10 / 25

11. User-Centric AI Analytics for Chronic Health Conditions Management
Managing Chronic Conditions
In educational context
In educational context
Educational systems have been in use for years and became
essential requirement during the COVID-19 pandemic.
Their wide spread use gives us the opportunities to address variety
of conditions and AI algorithms that can be generalised beyond the
original purpose of complementing educational systems
For example:
11 / 25

12. User-Centric AI Analytics for Chronic Health Conditions Management
Managing Chronic Conditions
In educational context
In educational context
Stress
Emotional stress is a common condition impact technology users.
Some acute forms can be observed during the use of educational
systems [2] which enabled us to provide adaptive mechanisms by
which the system responds to the user needs and reduce the
temporary stress. Our approach combined quantitative and
qualitative data analytics informing a rule-based system to perform
the adaptation.
12 / 25

13. User-Centric AI Analytics for Chronic Health Conditions Management
Managing Chronic Conditions
In educational context
In educational context
Dyslexia
Dyslexia is another condition that often impact student
performance both in class rooms and while using online
educational systems [3]. For this condition, a more sophisticated
optical sensor system, i.e. eye-gazing goggles, were necessary to
assess student’s motivation while using online educational systems.
The eye tracking data was combined with other sensory readings,
namely EEG [6], to develop a framework that enables personalised
services in the educational context.
13 / 25

14. User-Centric AI Analytics for Chronic Health Conditions Management
Managing Chronic Conditions
Ageing population and smart cities context
Ageing population and smart cities context
Another context for chronic health conditions management is
assistive technologies that are becoming ever more necessary with
ageing populations and the emergent of smart cities. Two relevant
areas, which are currently focal topics are empathetic technologies
and neural interfaces. There are multiple IEEE standards working
groups developing standards and recommended practices in these
two areas, e.g. [5] and [1].
14 / 25

15. User-Centric AI Analytics for Chronic Health Conditions Management
Personalising AI Models
Personalising AI Models
We adopted the principle of personalisation in exploring and
developing AI models [4, 7] and not just the systems developed or
enabled by AI.
15 / 25

16. User-Centric AI Analytics for Chronic Health Conditions Management
Personalising AI Models
Personalising AI Models
So when machine learning and analytic algorithms are developed
and trained,
⋆ more personal factors would have been taken into consideration ⋆
=⇒ user-centric and individualised learned models
=⇒ produce more relatable results from a user perspective than
focusing on abstract metrics.
We found that the need for such personalised AI models is even
more evident in the context of chronic health conditions.
16 / 25

17. User-Centric AI Analytics for Chronic Health Conditions Management
Personalising AI Models
Personalising AI Models
17 / 25

18. User-Centric AI Analytics for Chronic Health Conditions Management
What is next?
What is next?
Integrated Framework
▶ Developing an integrated framework for mental and physical
health monitoring.
▶ Combining the various sensors already used in existing
wearable and other domestic devices.
▶ User specific customisable framework.
18 / 25

19. User-Centric AI Analytics for Chronic Health Conditions Management
What is next?
What is next?
Evaluation Methods
▶ Investigating better metrics and evaluation methods in
evaluating machine learning and analytic algorithms especially
for personalised AI models and User-Centric AI applications.
▶ Integrating aspects of explainable and responsible AI to be
embedded in practical applications to provide a better
oversight of algorithmic AI.
19 / 25

20. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Conclusion
In this talk, we have presented briefly:
▶ the personalisation requirements and potential benefits of AI
technologies in managing chronic health conditions.
▶ the case for more personalised AI models to take into
consideration the fine differences in personal and
environmental factors
We conclude with some still open research questions and our
current and immediate future research plans in this area.
20 / 25

21. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Open Research Questions
Still open questions:
▶ explainable AI from deep learning and feature selection
algorithms to allow processing of big data and multimodal
sensors
▶ responsible AI including privacy and well-being impact
▶ machine learning algorithms metrics and evaluation methods
especially for personalised AI models.
21 / 25

22. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Current Research Plans
Our current research plans in this area is to focus on
▶ Nutrition related, e.g. diabetes and obesity
▶ Alzheimer’s
Our research plans for the near future is to focus on
▶ Well-being and mental conditions
▶ Cardiovascular, e.g. heart conditions, blood pressure, etc.
22 / 25

23. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Selected References
D. Zapala, A. Hossaini, M. Kianpour, G. Sahonero-Alvarez, and A. Ayesh, “A functional bci model by the p2731 working group:
psychology,” Brain-Computer Interfaces, vol. 0, no. 0, pp. 1–10, 2021.
Y. M. Lim, A. Ayesh, and M. Stacey, “Continuous stress monitoring under varied demands using unobtrusive devices,”
International Journal of Human-Computer Interaction, vol. 36, no. 4, pp. 326–340, 2020.
R. Wang, L. Chen, I. Solheim, T. Schulz, and A. Ayesh, “Conceptual motivation modeling for students with dyslexia for enhanced
assistive learning,” in Proceedings of the 2017 ACM Workshop on Intelligent Interfaces for Ubiquitous and Smart Learning,
SmartLearn ’17, (New York, NY, USA), pp. 11–18, ACM, 2017.
A. Ayesh, M. Arevalillo-Herráez, and P. Arnau-González, “Class discovery from semi-structured eeg data for affective computing
and personalisation,” (Oxford, UK), pp. 96–101, IEEE, 2017.
D. Schiff, A. Ayesh, L. Musikanski, and J. C. Havens, “Ieee 7010: A new standard for assessing the well-being implications of
artificial intelligence,” in IEEE SMC2020 Special Session on Human Well-Being in the Context of Autonomous and Intelligent
Systems, 2020.
R. Wang, L. Chen, and A. Ayesh, “Multimodal motivation modelling and computing towards motivationally intelligent e-learning
systems,” CCF Transactions on Pervasive Computing and Interaction, 2022.
A. Ayesh, M. Arevalillo-Herráez, and F. J. Ferri, “Towards psychologically based personalised modelling of emotions using
associative classifiers,” International Journal of Cognitive Informatics and Natural Intelligence (IJCINI), vol. 10, no. 2, pp. 52–64,
2016.
23 / 25

24. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Q & A
Acknowledgement
▶ My thanks to the IMAS conference organizing chairs and
committees for giving me this opportunity to share my work
with everyone here today and for their hard work in making
this event a success.
▶ My thanks also go to The IEEE Computer Society
Distinguished Visitors Program (DVP) for providing the
platform to connect researchers and speakers worldwide.
▶ Finally my thanks and gratitude to all of you for attending
and for your interest.
24 / 25

25. User-Centric AI Analytics for Chronic Health Conditions Management
Conclusion
Q & A
Any Questions?
Contact
Please do not hesitate in contacting me:
aladdin.ayesh@abdn.ac.uk
25 / 25