Browsing by Subject "public health informatics"

Wide range of psychological distress risk factors in older adults have been identified, however there is room for improvement in the comprehension of population-level disparity among risk factors for subgroups. Understanding risk factor variance would help public health professionals recognise the diversity of at-risk populations and arrive at more effective health policy. Older adults have unique risk factor profiles for psychological distress since they experience more frequently neurological disorders, diminished physical capabilities and insomnia along with adverse life events like bereavement and socio-economic changes. In this thesis, an approach combining computational statistics with methods of observational epidemiology were used with Helsinki Health Study cohort of City of Helsinki municipal workers to analyse novel risk factor disparities and evaluate the suitability of model-based recursive partitioning method. The method combines decision trees with generalised linear models to identify subgroups with the most significant disparity among risk factors. The risk factors investigated for psychological distress in older adults included sleep disturbance, history of depression, childhood adversities and lack of social support. Sleep disturbance showed relatively low disparity, while social support and childhood adversities showed more disparity across subgroups. Model-based recursive partitioning offers promising modelling technique for exploratory subgroup analysis with focus on interaction effects and model interpretability.