20 September 2017, 12:00 - 14:00
The Anglia Ruskin IT Research Institute (ARITI) invites you to two presentations at this research seminar.
We are delight to welcome Ifty Ahmed, CEO and founder of Pow Health, one of our partners in our TB-Test project. Ifty will present "A look at a patient self-management technology, striving for patient centricity".
We also present Md Mahmudul Hasan, one of our PhD researcher. Mahmud will discuss "Optimising decision in a multi-criteria based environment".
Both presentations followed by Q&A sessions will take place 12:00 - 13:00.
Lunch and networking will take place 13:00 - 14:00.
Ifty Ahmed will showcase the work being done at Pow Health to create a single technology platform to support the changing health needs of patients. Insights into how the technology is being used today and could be used in the future will be shared and discussed.
Ifty Ahmed has over 16 years’ experience as chief strategist in high tier internet technology companies. He has managed delivery of world-class global projects, created and run multiple agency teams, won and sustained senior client relationships for blue-chip brands including; Vodafone, Microsoft, Merck Sharp & Dohme, GSK, Reuters and Visa. Having co-founded and run an award-winning agency in the 1990s, Ifty became Head of Digital at IMG Media and Vivid Lime, before taking global responsibility for Collinson Group as Global Head of Digital and Customer Engagement. Ifty founded Pow Health in April 2011 aiming to create a secure and smarter way to manage 'all your health in one place'. Currently, the technology is benefiting people in over 88 countries around the world.
Ifty is a creative technologist, passionate about empowering people so that they can better manage their health, support one another and help create better treatments for everyone. This is the focus behind Pow Health.
Decision refers to a conclusion or the outcome of a situation or to resolve a particular question. We consider or re-consider many decisions in our everyday life. Some of them are direct or indirect while some of them depend on various criteria or sub-criteria in a constrained environment. In almost every case, we follow a process or some steps to reach a final decision or to answer a specific question. Moreover, we may observe some situations where there are similar and equally important decisions. Therefore, it is often required to select an optimised decision among several alternatives in a data-driven decision making scheme. In this study, a simulated environment has been created using Markov Decision Process in a reinforcement learning settings. The proposed algorithm is based on multi-objective triple Q learning (i.e. an off policy model free reinforcement learning) that communicates with a knowledge model. This knowledge model is based on Fuzzy Analytic Hierarchy Process (FAHP). The result shows, the proposed hybrid algorithm performs better compared to Monte-Carlo Tree Search (MCTS) and Dynamic Programming (DP). Within less elapsed time, it can achieve the goal state and return the expected value in a fully observable Markov Decision Process. In other words, the optimised decision can easily be achieved in a multi-criteria based environment with the help of the proposed MOQN-FAHP algorithm.
Md Mahmudul Hasan has worked as a senior lecturer in the Department of CSE at Daffodil International University, Dhaka, Bangladesh. He completed his MSc in Computer Science at the University of Essex, UK, with a specialisation in games and mobile apps development. Mahmud's research interests include machine learning, data science, decision support system, and personalized learning through games and gamification. He won the best poster prize at the Faculty of Science and Technology's 7th Research Conference.