Predicting the elderly's quality of life based on dynamic neighborhood environment under diverse scenarios: An integrated approach of ANN, scenario analysis and Monte Carlo method

Abstract There is an increasing global population of older adults in recent years, and the trend will be more acute in the following decades. Owing to low mobility and physical impairment, the elderly are sensitive to their nearby neighborhood environment. However, it is challenging to accurately judge changes of the elderly’ quality of life (QoL) before conducting improvement strategies of neighborhood environment due to complicated environmental impacts. This study proposes a QoL prediction approach by integrating artificial neural network (ANN) model, scenario analysis and Monte Carlo experiment. The QoL of the elderly is measured from four domains, and the neighborhood environment is measured by 16 key indicators. Based on the measurement data collected from Nanjing, the ANN model is trained to fit the influence relationship between neighborhood environment and the elderly's QoL. Scenario analysis sets up potential scenarios for neighborhood environment under natural progressions and human interventions. Finally, Monte Carlo experiment is conducted to predict the probability distribution of the elderly's QoL values under potential scenarios by using the trained ANN model as functions. The predictive QoL values of the elderly show the change pattern of the elderly's QoL with dynamic neighborhood environment, reveal the independent and compound effects of natural progressions and human interventions, and confirm the mutual promotions between human interventions. Furthermore, the integrated prediction approach can be implemented in other cities and regions to forecast the local elderly's QoL under possible scenarios, and offer concise evidence for deciding improvement strategies of neighborhood environment to support aging-in-place.