The numerous surveillance videos recorded by a single stationary wide-angle-view camera persuade the use of a moving point as the representation of each small-size object in wide video scene. The sequence of the positions of each moving point can be used to generate a trajectory containing both spatial and temporal information of object's movement. In this study, we investigate how the relationship between two trajectories can be used to recognize multi-agent interactions. For this purpose, we present a simple set of qualitative atomic disjoint trajectory-segment relations which can be utilized to represent the relationships between two trajectories. Given a pair of adjacent concurrent trajectories, we segment the trajectory pair to get the ordered sequence of related trajectory-segments. Each pair of corresponding trajectory-segments then is assigned a token associated with the trajectory-segment relation, which leads to the generation of a string called a pairwise trajectory-segment relationship sequence. From a group of pairwise trajectory-segment relationship sequences, we utilize an unsupervised learning algorithm, particularly the k-medians clustering, to detect interesting patterns that can be used to classify lower-level multi-agent activities. We evaluate the effectiveness of the proposed approach by comparing the activity classes predicted by our method to the actual classes from the ground-truth set obtained using the crowdsourcing technique. The results show that the relationships between a pair of trajectories can signify the low-level multi-agent activities.

This work presents comprehensive analytics of trajectories of functional loss and recovery using sequence analysis and clustering techniques. The study focuses on a large dataset consisting of assessments of activities of daily living conducted among nursing home residents. The first main part of this research involves converting the assessments into sequences of disability combinations and utilizing graphical tools and various indicators to gain valuable insights into the trajectories of functional disabilities over time. In the second part of the research, a novel clustering approach is introduced that combines Markov models with distance-based techniques. This hybrid methodology results in 13 distinct clusters of trajectories. The clusters are thoroughly examined, and representative sets are carefully selected based on various criteria. This selection process ensures that the chosen sets accurately represent the characteristics of each cluster. The findings of this study have significant implications for healthcare systems, including developing predictive models which can be utilized to forecast the trajectory of individual patients based on their cluster membership. This enables healthcare providers to anticipate disease progression, tailor treatments, and dynamically adjust care plans, resulting in improved patient outcomes and the overall quality of care. Moreover, the information derived from the analytics can aid in optimizing healthcare systems by facilitating resource allocation and cost optimization. The insights gained can also guide policymakers and families in planning appropriate care for patients. This research advances healthcare decision-making and ensures appropriate support. / Thesis / Master of Science (MSc) / The ability to independently perform activities of daily living (ADLs) is a crucial indicator of an individual's health status, and the loss of this ability can have a profound impact on their overall quality of life. Our research focuses on analyzing the trajectories of patients as they experience functional decline and recovery. While various techniques have been utilized to explore ADL trajectories, this study stands out by employing clustering and sequence analysis approaches to examine different groups of trajectories. To overcome the computational challenges involved, we propose a combined clustering approach. This hybrid approach consists of two phases: applying a Markov model prior to distance-based algorithms. The findings derived from our research hold significant applications in optimizing healthcare systems, improving health outcomes, facilitating the development of targeted and effective interventions that support patients in preserving their independence, and enhancing the quality of care.