Analysing Aspects of General Health in Scotland

Abstract

The Scottish Health Survey (SHS), recognised as National Statistics, provides an invaluable data source for understanding health trends and disparities in Scotland (Scottish Government, 2023). Meanwhile, the principles of machine learning, a facet of artificial intelligence grounded in mathematics, statistics, and computer science, offered us the tools to analyse these large datasets effectively. In this study, we set out to identify factors correlated to the general health of the Scottish population by using Cramér's V and to predict the general health within the 2021 data based on these factors from 2019 and 2020. Through our analysis, we found that a wide range of factors, from Longstanding illnesses and Lifestyle behaviours to Socioeconomic status, Cardiovascular Disease and Diabetes, Asthma, Adult physical activity, Height and weight and Education, had varying degrees of correlation with general health. Three different machine learning models were trained and tested for each year in this study. Our research revealed that the Logistic Regression model, with an accuracy of 60%, performed optimally in predicting the general health of the population for the year 2021, with data from 2019 proving to be more efficacious in prediction. This study has a few limitations that could have influenced the findings. Firstly, the 2020 dataset had fewer cases in specific health categories, which could have affected the prediction accuracy for 2021. Secondly, a significant overlap was observed between Good and Very Good health cases, which may be due to respondents' subjective evaluations of their general health. Expanding the research to include data from other years could provide a more comprehensive view of health trends over time. Also, investigating alternative machine learning models might offer opportunities to improve prediction accuracy. Lastly, conducting similar studies in different geographical areas could yield valuable comparisons and insights.