Details

• Actionable AI in ICU
  Supervisor: Parvin Mousavi
  Description:
  
  ICU environments are fast paced and high pressure and require decision making in the presence of high levels of uncertainty. In this project we will focus on designing actionable predictions using physiological signals collected continuously from ICU patients. This is a multi-disciplinary project in conjunction with Health Sciences and Business School at Queen's. Students are required to have a solid understanding of machine learning and experience with developing and applying deep learning approaches. Knowledge of signal processing is a plus.
• Teaching Agents About Desires and Intentions
  Supervisor: Christian Muise
  Description:
  
  Recent work on epistemic planning has demonstrated how we can practically solve problems where agents reason about the nested belief of other agents by enforcing syntactic restrictions on what they can represent. This project aims to extend the recent line of epistemic planning work to the notions of "desire" (what an agent want's to achieve) and "intent" (what the agent is hoping to do). This will open the door to solving problems such as having an agent be zen (i.e., they hold no desires) or adhere to FOMO (i.e., they adopt all intentions they believe others might have), and so on.
  
  Strong performance in CISC 352 (especially the planning component) will be essential for this project. The student is not expected to know about epistemic planning (this will come as part of the learning for the project). Implementation will be mostly in Python and a custom language for specifying agent behaviour.
• ML applied to robotic assessment of cognition in epilepsy
  Supervisor: Gavin Winston
  Description:
  
  Epilepsy is a common neurological condition characterised by recurrent seizures. Many people with epilepsy also have cognitive problems e.g. memory and planning. Different subtypes of epilepsy affect varying parts of the brain so have distinct patterns of impairment. Within a single subtype of epilepsy, there are different patterns (clusters) of impairment.
  
  You will use a database of around 100 people with epilepsy that includes both assessment with pen-and-paper tests as well as a variety of tasks with the Kinarm robot (https://www.kinarm.com) yielding many quantitative parameters of performance that can be used for machine learning (https://pubmed.ncbi.nlm.nih.gov/25571189/).
  
  In this project, you will explore the use of machine learning techniques in the clustering of these scores into different patterns of impairment, and the classification of the subtype of epilepsy from these cognitive data.
  
  This project will be supervised by Dr. Gavin Winston, with input also from Dr. Parvin Mousavi.