Dashboards on Urban Health Using the MedSat Dataset

Urban planners and public health officials are increasingly investing in cycling infrastructure to promote sustainable transportation and improve population health. However, quantifying the relationship between cycling infrastructure and health outcomes at a local level remains challenging, particularly due to the uneven distribution of biking amenities, such as cycle lanes, shared-use paths, and surface quality. This project develops a set of cycling related metrics, including commute rate, crash rate and infrastructure quality metrics, aggregated to the Medium Super Output Area level. These metrics are then analysed alongside health outcome data to explore potential correlations between cycling uptake, infrastructure quality and health prescriptions. The results indicate that even a small increase in active commuting is associated with notable improvements in health outcomes, particularly reducing hypertension rates. While cycling infrastructure quality is not as strong a predictor on its own, it still shows a meaningful relationship with health indicators. However, this pattern is not consistent across all UK cities, suggesting that local context may significantly influence the strength of these associations. Although the benefits of active commuting are already well established in existing literature, and this project supports those findings, the more novel contribution lies in investigating the link between infrastructure quality and health. By doing so, the project moves beyond commute focused census data to consider how infrastructure may encourage not only active commuting, but also other forms of active travel - such as trips for shopping or leisure - that are less frequently captured in traditional analyses.

Green spaces such as parks are recognised to promote public health, offering opportunities for physical activ-ity, social interaction, stress reduction, and general well-being. Recent research goes beyond simply looking at the benefits of park presence and emphasizes the importance of park quality, providing a consistent development in new methods to quantify park attributes and understand their impact on various health outcomes. However, gaps remain in linking park quality to spatially resolved real-world health data. Currently, many studies rely on self-reported results or use oversimplified spatial park usage models that hinder the precision of their findings. This work aims to address these gaps by building on a recently proposed framework and further quantifying the health-promoting qualities of parks in the five most populous cities in England. First, park quality health scores were calculated and applied to Lower Super Output Areas (LSOAs) to derive the corresponding LSOA-level health scores. They were determined by the scores of intersecting parks, weighted by the degree of spatial overlap between the LSOAs and the park influence zones. The scores were also correlated with the five MedSat health outcomes, including anxiety, asthma, depression, diabetes, and hypertension. The findings were presented through various visualisations and a case study was conducted to analyse the strength of the relationship between LSOA scores and prescription rates while controlling for key socioeconomic covariates. The results of the analysis showed that the LSOA scores had the strongest correlation with prescription rates compared to any of the other socioeconomic variables. It was also discovered that London consistently exhibited the strongest negative correlations in all health outcomes, suggesting a clearer desirable link between LSOA scores and prescription rates. Manchester and Liverpool showed consistent but weaker associations, while Leeds presented a unique variance in relationships. In contrast, Birmingham did not show any clear correlations. Although hypertension in London exhibited the strongest correlation, anxiety and asthma presented the most consistent positive associations overall. Ultimately, the findings suggest some promising links; however, the predictive strength of park quality on health outcomes is context-dependent and varies by city.

The United Kingdom faces longstanding and persistent public health challenges linked to physical inactivity. Current evidence suggests that neighbourhood environmental characteristics – including walkability, greenspace availability, and air quality – substantially influence health outcomes. However, previous research has often been limited by coarse spatial units and the use of aggregated statistics, as well as by the reliance on restricted sets of urban and environmental variables. This underscores a need for fine-grained, multi-dimensional data to better determine these associations. This investigation will examine how neighbourhood walkability impacts the incidence of obesity and obesity-related illnesses within urban environments. It will capitalise on the recently published MEDSAT dataset, which offers high-resolution spatial data on health and environmental indicators across all Lower Layer Super Output Areas (LSOAs) in England. By integrating detailed socio-demographic data with satellite observation-derived environmental metrics, MEDSAT enables an in-depth analysis of how walkability correlates with community health outcomes in London. By making use of this dataset, this investigation aims to bridge the gap between broad epidemiological trends and the local-level determinants of health. The methodology will involve developing a walkability index for London’s LSOAs by combining MEDSAT variables with additional geospatial data, including street network connectivity, availability of dedicated pedestrian walkways, and accessibility to fundamental amenities, including supermarkets and pharmacies. I will then analyse the relationships between walkability and health outcomes of interest, including proxies for obesity, diabetes, and cardiovascular conditions, whilst accounting for secondary environmental factors, such as green space access and NO2 pollution levels. Finally, the findings will be visualised through an interactive dashboard, enabling urban planners and public health stakeholders to explore spatial patterns and identify areas where improving walkability and environmental conditions could produce significant health benefits.

This project investigates whether neighbourhood greenness reduced increases in anxiety and depression prescription rates across England in 2020, after accounting for NO2 pollution, population density, income and baseline health. It also tests whether these patterns hold in London, Birmingham, Liverpool, Manchester and Leeds. Using the MEDSAT dataset for Sentinel-2 NDVI and satellite NO2 data, alongside NHS prescription counts and demographic controls, a spatial panel for 31 799 LSOAs was assembled via Apache Spark. Data were stored in PostGIS and visualized through a React and Deck.gl dashboard. Analyses included ordinary least-squares, spatial-lag and spatial-error regressions, and NDVI-bin models. Key contributions are a reproducible ETL pipeline, an interactive mapping interface and a comprehensive modelling framework. Main findings show that greener areas experienced larger prescription increases in simple OLS, while polluted areas showed smaller rises, likely reflecting under-treatment in deprived districts. After adding socio-demographic controls, local greenness effects became non-significant in OLS and spatial-error models, but spatial-lag models for depression revealed strong neighbourhood spill-over of greenness. City-level multivariate models replicated these results once density, income and health were included. Environmental variables explained under 10 percent of prescribing variation; population density and baseline health were the strongest predictors. The principal conclusion is that greenness alone did not buffer mental-health prescribing surges during the pandemic, socio-economic context and healthcare access were the dominant factors.