Jinsi
Guo
Sponsor: Fumika Hamada, Ph.D. Neuro Physio & Behavior Sleep is a biological mechanism heavily influenced by the circadian clock and is vital for homeostasis. Our research aims to build upon previous sleep studies and introduce a new method of tracking sleep via temperature to illustrate the circadian mechanisms responsible for sleep. Given that Drosophila relies on the environment for temperature regulation, a temperature gradient, and AI tracking assay was conducted over a 2-day (12H:12H) light/dark cycle to track sleep patterns. Analysis was performed using proprietary software to compare locomotor activity and body temperature rhythms (BTR). Wild-type (W1118 and CS) Drosophila lines were controls for general sleep patterns compared to Per, Tim, and Clock gene mutants. Results demonstrate that W1118 had a robust sleep pattern with average sleep bouts at their greatest during ZT16 to 17 before decreasing as the light phase began. This contrasted our mutant lines, which displayed inconsistent but similar sleep durations throughout the dark cycle. These results corroborate past studies on Drosophila circadian rhythms, revealing the circadian clock's role in sleep regulation and illustrating that our software is a valid tool for tracking sleep. The insights gained from Drosophila can elucidate sleep mechanisms and potential implications for sleep-related disorders. Sensor-Driven Home Monitoring for Alzheimer's Patients
Abstract profile. Full document pending author claim.
Authors:
Jinsi Guo
Date Created:
Not specified
Course Title:
Professor:
Not specified
About Paper:
Seven million Americans live with Alzheimer's, a number that is expected to increase to thirteen million by 2050. Alzheimer's patients experience memory and speech degradation that worsens with the progress of the disease. Since memory degradation can leave subtle traces in daily activities, doctors are relying on remote patient monitoring (RPM) devices to track patient activities and provide proactive care. However, existing RPM solutions face significant issues: camera-based systems raise privacy concerns while sensor-based alternatives often struggle with accurate data labeling and preprocessing, limiting their effectiveness. In light of these challenges, this project proposes a machine learning-driven pipeline that utilizes computer vision and time series analysis to improve sensor-based activity recognition using vibrational data from fixed sensors installed within the home. By leveraging computer vision models and rule-based algorithms, videos filmed with the collection of vibrational data can be used to annotate this data with the corresponding activity and its duration. The annotated data, after preprocessing, will then be trained using a machine-learning model enabling activity recognition based on vibrational sensor patterns. Once trained and validated, the sensor and model can operate independently, offering an efficient, reliable, and privacy- preserving solution for remote patient monitoring (RPM) in Alzheimer's care. Need for Consistency and Recognition of Point- of-Care Testing across Critical Notifications Lists in the United States Tarini Guru
Source:
UC Davis / Elect & Comp Engr / 2025
Topics:
No topics listed
Co-authors:
Jinsi Guo