Predictive Analytics for Community Health: Enhancing Virginia's Quality of Life Using Machine Learning; Utilizing Algorithms for Healthcare Advancement: Exploring the Implementation of Machine Learning by Government Officials in Developing Countries

My technical study on developing a predictive model for the Health Opportunity Index (HOI) in Virginia and my STS essay investigating the integration of machine learning (ML) in addressing healthcare inequities in developing countries are connected through their shared focus on leveraging ML to enhance health equity. The technical report provides a practical application of ML techniques, such as linear regression and k-means clustering, to analyze social determinants of health and forecast HOI data. Additionally, it offers a model for identifying health disparities to inform targeted policy interventions. This approach mirrors my STS essay's exploration of how ML, supported by ICT and e-Government frameworks, can be strategically deployed to mitigate healthcare inequities in developing nations. Both essays emphasize the need for extensive data collection and collaboration between technology and policy to ensure effective implementation of ML solutions. Furthermore, the STS essay's use of Actor-Network Theory (ANT) and the Multilevel Perspective (MLP) to analyze sociotechnical interactions and systemic change complements the technical report’s practical findings. Both essays collectively provide a view of how emerging technologies such as ML can address health disparities in both the United States and other developing countries.

The technical study I performed aimed at enhancing health equity in Virginia by developing a predictive model for the HOI using ML algorithms, specifically linear regression and k-means clustering. The project began by compiling HOI data, which included a wide range of variables focused on social determinants of health, such as geographical location, population churning, and material deprivation. Using these datasets, my team trained and fine-tuned our models to identify which social factors were most significantly related to health outcomes. By applying k-means clustering, we were able to categorize communities into clusters based on the similarity of their health opportunity profiles. The linear regression model utilized a decision tree model to predict the HOI of various census tracts, achieving a notable root mean squared error (RMSE) of 0.020338. Principal Component Analysis (PCA) was also employed to simplify the dataset while maintaining data integrity. The findings revealed that material deprivation was the most critical factor affecting the HOI of a community, followed closely by community environment and affordability. These findings provide Virginia health officials with actionable data to identify areas of high health inequity and prioritize interventions. The predictive model’s success demonstrates its potential for guiding resource allocation to improve health outcomes. Future work involves refining the model for greater accuracy, incorporating additional variables like environmental quality and access to healthcare, and adapting the approach to other regions both within and outside the United States to validate its broader applicability. This research sets a precedent for using ML methods to address health disparities, ultimately leading to improved health equity in all communities.

My STS essay investigates the integration of ML algorithms by government officials as a strategic approach to address healthcare inequities in developing countries. The central research question guiding this study is: How can ML, supported by existing information and communication technologies (ICTs) and e-Government frameworks, enhance healthcare equity in developing countries? Analyzing this question through the lens of the MLP and ANT frameworks, the paper explores how technological changes at the niche, regime, and landscape levels can impact the interaction between human and non-human actors in healthcare systems. The paper expects to find that while ML has the potential to significantly improve healthcare equity, its effectiveness is contingent upon existing ICT infrastructure and data availability, particularly in rural areas. The significance of this research to the field of STS and engineering lies in its use of sociotechnical analysis in determining if highly technical ML solutions are feasible for developing countries. Additionally, it emphasizes the importance of understanding the social context and infrastructure needed for successful technology implementation. The findings aim to provide a better understanding of the challenges and opportunities in deploying ML technologies in healthcare systems, offering policy recommendations to help foster more equitable health outcomes.

Working on both the technical report and STS essay simultaneously allowed me to evaluate the feasibility of emerging technologies such as ML in real-world applications. This approach emphasized the technical challenges and potential of ML while also considering the economic and infrastructural limitations that could impact its implementation. By merging practical technical findings with sociotechnical network analysis, I gained an understanding of how ML can be used to address health inequities. This made me recognize the necessity of widespread ICT infrastructure and the critical role of policy and economic context in facilitating technological innovation at each level of a sociotechnical system. This dual perspective provided a broader understanding that would not have been possible by working on either project independently.

School of Engineering and Applied Science

Bachelor of Science in Computer Science

Technical Advisor: N. Rich Nguyen

STS Advisor: Bryn Seabrook

Technical Team Members: Jared Dutt, Anthony Ferguson