Background: Dementia caregiving is a demanding and often prolonged role that disproportionately affects older women. While critical for care recipients, caregiving can impose significant psychological, behavioral, and physiological strain on caregivers. This dissertation investigates how caregiving relates to physical activity patterns, autonomic function, and psychological and cardiovascular health, using a multi-method approach that integrates literature synthesis, wearable sensor data, and predictive modeling. Methods: The dissertation comprises three interconnected studies. Study 1 is a systematic review and meta-analysis of studies comparing physical activity between informal caregivers and non-caregivers. Random-effects meta-analyses estimated pooled odds of physical inactivity and insufficient moderate-to-vigorous physical activity (MVPA), with heterogeneity assessed via the I² statistic. Study 2 is an observational study comparing device-measured physical activity, sedentary behavior, and heart rate variability (HRV) between older women dementia caregivers (n = 15) and non-caregivers (n = 16), alongside survey-based assessments of psychological health and caregiving burden. Study 3 used logistic and stepwise regression models to evaluate associations between wearable-derived metrics and psychological and cardiovascular health, including bootstrapping for internal validation. Results: Study 1 found caregivers had slightly higher odds of physical inactivity (OR = 1.11, 95% CI: 0.80–1.54) and insufficient MVPA (OR = 1.13, 95% CI: 0.92–1.41), though associations were not statistically significant. Substantial heterogeneity (I² > 90%) highlighted variability in caregiving definitions and physical activity measures. In Study 2, caregivers reported significantly poor psychological health including depression (rₚb > 0.5), and anxiety. Caregivers also reported significantly higher stress and poor sleep quality and showed more fragmented activity patterns. Despite higher daily step counts, they accumulated fewer sustained walking bouts and more prolonged sedentary time. HRV day-to-night Root Mean Square of Successive Differences (RMSSD) metrics indicated impaired autonomic recovery, particularly among those with higher caregiving burden. In Study 3, several wearable-derived metrics including prolonged sedentary bouts and lower HRV were associated with poor psychological health. Time in activity bouts > 5 minutes, prolonged sedentary bouts and HRV predicted poor cardiovascular health in stepwise regression. However, predictive modeling performance declined after bootstrapping, underscoring the need for caution when interpreting these models due to the small sample size and limited generalizability. Conclusion: Older women and women caregivers for individuals with dementia experience measurable psychological and physiological strain, which can be feasibly and objectively captured using wearable technologies. While caregivers often accumulate more daily steps, they also demonstrate less favorable movement patterns, such as fewer sustained walking bouts, more prolonged sedentary periods, and greater fragmentation of activity. These findings support the utility of real-world monitoring psychological and physiological dysregulation. Wearable-derived metrics show promise as informative indicators of caregiver health; however, caution is warranted when applying predictive models in small samples, given their limited generalizability and reduced performance after bootstrapping. Collectively, these results highlight the need for scalable, flexible interventions that address caregiver burden, reduce risky sedentary accumulation, and enhance autonomic recovery to promote long-term health and resilience.