Search Results (3)

This study investigates the application of explainable AI (XAI) techniques to understand the deep learning models used for predicting urban conflict from satellite imagery. First, a ResNet18 convolutional neural network achieved 89% accuracy in distinguishing riot and non-riot urban areas. Using the Score-CAM technique, regions critical to the model’s predictions were identified, and masking these areas caused a 20.9% drop in the classification accuracy, highlighting their importance. However, Score-CAM’s ability to consistently localize key features was found to be limited, particularly in complex, multi-object urban environments. Analysis revealed minimal alignment between the model-identified features and traditional land use metrics, suggesting that deep learning captures unique patterns not represented in existing GIS datasets. These findings underscore the potential of deep learning to uncover previously unrecognized socio-spatial dynamics while revealing the need for improved interpretability methods. This work sets the stage for future research to enhance explainable AI techniques, bridging the gap between model performance and interpretability and advancing our understanding of urban conflict drivers. Full article

Land expropriation has always been a hot spot of social conflicts. The land expropriation policy of Merging Villages and Living Together (MVLT) in rural areas has intensified conflicts due to insufficient financial compensation and “demolishing old houses before building new ones”. The current research most simply assesses the degree of conflict and the influencing factors but rarely includes farmers, governments at all levels, the strength of policy tools, and policy perceptions in a unified quantitative research framework, which is not conducive to conflict resolution and policy improvement. This paper adopts the institutional credibility theory, incorporates the policy instruments of higher-level governments, administrative instruments of lower-level governments, and farmers’ credibility of policies into a unified accounting framework, constructs a conflict stress index, evaluates the role of each subject’s characteristics, policy perceptions, and policy instruments in the process of conflict generation and resolution, and analyzes the methods of conflict resolution from the perspective of different stakeholder conflicts. The theoretical analysis framework and the quantitative analysis of the indicators are verified by using a case study of MVLT policy in Shandong Province, China. The results show that the credibility of the policy of “village integration” is influenced by individual characteristics and varies significantly. The administrative means and different combinations of lower-level government are significantly related to an increase in farmers’ credibility, which can significantly improve the success rate of policy implementation. The effect of administrative means of higher-level government on the credibility of farmers is limited. The highest value of the conflict index was observed when the administrative instruments reached the maximum value without a marginal increase in farmers’ credibility. Based on the quantitative evaluation of conflict generation and resolution mechanisms, recommendations for policy implementation and improvement were made. Full article

The rise in non-communicable diseases (NCDs) has been attributed to economic growth in developing countries, shifts in societal norms, and behaviors such as dietary habits and physical activity. Up to 80% of NCDs could be prevented by eliminating shared risk factors, mainly tobacco use, unhealthy diets, physical inactivity, and the harmful use of alcohol. The South African government’s national strategic plan to control NCDs, which includes cardiovascular disease (CVD) prevention, places a strong emphasis on the need to improve the prevention, detection, early intervention, and management of NCDs. In line with the above recommendations, this study aimed to screen rural communities using the non-laboratory INTERHEART Risk Score tool (NLIRS) and develop relevant and suitable intervention strategies for a patient at moderate risk of developing a heart attack. A quantitative research approach applying a household-based design was used to conduct this study and the community action model (CAM). The difference between pre-intervention and post-intervention results were analyzed using a t-test and Analysis of covariance (ANCOVA) with age, smoke, hypertension, and diabetes as the covariates. The study found a significant difference in proportions between pre and post-intervention for raised Systole (SBP), obesity by body mass index (BMI), and waist circumference (WC). In rural communities, using CAM to improve knowledge and behavioral practices of NCD risk factors is feasible and effective. This basket of interventions will assist community members in reducing their risk of developing metabolic syndromes as well as their risk of developing CVDs. Continued investment and research in CVD prevention interventions are required to improve health, reduce costs, and have long-term benefits for conflict-affected individuals and communities. Full article