VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT REVOLUTIONIZING HEALTHCARE SEO: LEVERAGING MACHINE LEARNING AND CHATGPT API FOR ENHANCED DIGITAL WELLNESS OUTREACH Student’s name NGUYEN TUAN THANH Hanoi - 2024 VIETNAM NATIONAL UNIVERSITY, HANOI INTERNATIONAL SCHOOL GRADUATION PROJECT REVOLUTIONIZING HEALTHCARE SEO: LEVERAGING MACHINE LEARNING AND CHATGPT API FOR ENHANCED DIGITAL WELLNESS OUTREACH SUPERVISOR: DR. NGUYEN DOAN DONG STUDENT: NGUYEN TUAN THANH STUDENT ID: 20070980 COHORT: QH2020-Q SUBJECT CODE: INS401101 MAJOR: BUSINESS DATA ANALYTICS Hanoi - 2024 ACKNOWLEDGEMENT I extend my deepest gratitude to Dr. Nguyen Doan Dong, my supervisor, whose expert guidance and unwavering support have been pivotal throughout this research journey. Dong's profound knowledge and insightful critiques have significantly enriched my work, providing both technical and conceptual clarity.
Beyond his academic guidance, Dr. Dong has been a constant source of encouragement and motivation, helping me navigate through challenges with great wisdom and patience. His dedication to fostering a nurturing and intellectually stimulating environment has immensely contributed to my growth as a researcher and as a professional. I am also profoundly thankful to the International School of Vietnam National University (VNU), a prestigious institution that has provided me with an invaluable platform to explore and hone my skills in Business Data Analytics.
The opportunity to study in an environment that champions rigorous academic research and innovation has been fundamental to my personal and professional development. The support and resources provided by the university have been instrumental in allowing me to pursue my passions and to undertake this significant research endeavor. To all the faculty members, peers, and administrative staff at the International School - VNU, whose assistance and encouragement have been essential, I express my sincere appreciation. Your collective wisdom and support have not only enhanced my academic journey but have also prepared me for future challenges and opportunities.
This thesis stands as a testament to the collaborative effort and shared knowledge that my supervisor, university, and its community have generously offered. I am immensely grateful for their contributions and am proud to present this work as a reflection of our collective commitment to advancing the field of digital marketing through innovative research. GUARANTEE As the principal researcher, I am committed to upholding the highest standards of academic integrity in conducting and presenting my research. This thesis on leveraging XGBoost for analyzing impactful SEO features across various websites is entirely the product of my original research efforts.
It contains no plagiarism, and it respects intellectual property rights, adhering strictly to academic norms and legal requirements. I ensure that all methodologies and technologies applied, including the innovative use of the XGBoost framework to evaluate the effectiveness of SEO features, were chosen and developed based on their relevance to the research objectives and their scientific merit. The findings presented are a result of rigorous data analysis and are aimed at providing a deeper understanding of SEO impacts on webpage ranking, specifically tailored for improving digital marketing strategies. This thesis not only adheres to rigorous academic standards but also contributes new knowledge to the field of digital marketing by exploring advanced machine learning techniques.
The practical applications developed from this research, particularly in enhancing SEO through predictive modeling, highlight the potential for significant advancements in the way businesses optimize their digital presence. In conclusion, I am deeply invested in the integrity and utility of this research. I am eager to contribute to ongoing academic conversations and practical implementations in digital marketing and SEO. I look forward to the potential applications of my findings in real- world scenarios and their subsequent impact on improving web page visibility and ranking.
ABSTRACT Search Engine Optimization (SEO) is critically important in the digital marketing landscape, particularly within the healthcare sector. This thesis develops a robust decision model for search engine rankings aimed at optimizing website performance to better satisfy user demands. Utilizing two advanced gradient boosting models, Light Gradient Boosting Machine (LightGBM) and Extreme Gradient Boosting Decision Trees (XGBoost), this study assesses the relationships and relative importance of various SEO factors. Comparative analysis indicates that XGBoost supersedes LightGBM in predicting actual search engine rankings, achieving an average accuracy rate of 87.
A detailed feature analysis by using SHapley Additive exPlanations (SHAP) highlights the significance of internal links, consistent keyword presence across paragraphs, the quantity and length of H2 headings, and the presence of keywords within anchor texts as paramount for effective SEO in the healthcare domain. Conversely, keywords located in footers, URLs, and image alt attributes were found to be less influential. Furthermore, this research includes a practical evaluation of the 'Everyday Health' website through comprehensive webpage crawling. Based on the identified SEO insights, strategic recommendations are provided to enhance the website's search engine positioning.
This study not only contributes to the academic understanding of SEO but also offers practical solutions for real-world applications, emphasizing the transformative impact of machine learning in the healthcare sector's digital marketing strategies. Keywords: Search-Engine Optimization; SEO Optimization; Machine Learning; Rank Prediction; Online Healthcare Industry; Digital Marketing; Everyday Health. TABLE OF CONTENTS I. Collection of Articles 6 2.
Findings forms Related Works 6 III. Data Collection 11 IV. Search Engine Optimization (SEO) 18 1. Features importance analysis 28 4.
Application of SHAP in Our Study 30 4. In-Depth Analysis of SEO Feature Characteristics 35 V. Practical Implications to Marketers for SEO in the Healthcare Industry 40 2. Limitations and Future Works 43 VI.
REAL IMPLEMENTATION WITH EVERYDAY HEALTH WEBSITE 44 1. Everyday Health Introduction 44 2. Visualization and Comparison with Top Ranking Insights 46 4. SEO Strategies Proposal for Everyday Health 50 VII.
CONCLUSION 54 REFERENCES 55 LIST OF TABLES Table 1: Keywords Utilized for Crawling Top 10 Webpage Rankings Using Google 11 Search Engine Table 2: Features Extracted from Webpages 15 Table 3: SEO Strategies for Everyday Health Website 53 LIST OF FIGURES Figure 1: Average NDCG Scores of Models 27 Figure 2: SHAP beeswarm summary plot measured for XGBoost 32 Figure 3: In-Deep Analysis of Internal Links 35 Figure 4: In-Deep Analysis of Paragraph Keyword Count and Anchor Keyword 36 Count Figure 5: In-Analysis of Number of H2 tags and Length of H2 tags 37 Figure 6: Analysis of H3 Tag Length, Image Quantity, Meta Description Length, 39 External Links, H1, H2, H3 Tag Usage, Meta Keyword Count Figure 7: Distribution of SEO features in Everyday Health website 46 Figure 8: Additional Insights from SEO features mixed of Everyday Health website 49 LIST OF ABBREVIATIONS No. Feature Acronym Definition 1 Amount of text total_words Count the number of characters in paragraph and titles (<p> and <h> elements) 2 H1 count of titles h1_num Count the number of H1 titles on page 3 H1 length h1_len Count the average length of H1 titles on page 4 H2 count of titles h2_num Count the number of H2 titles on page 5 H2 length h2_len Count the average length of H2 titles on page 6 H3 count of titles h3_num Count the number of H3 titles on page 7 H3 length h3_len Count the average length of H3 titles on page 8 Header total header_total Count of all the headers on page 9 Image count img_count Count the number of images 10 Internal links internalLinks Count the number of internal links count (internal = linking to a page in the same domain) 11 External links externalLinks Count the number of external links count (external = linking to a page in the different domain) 12 Total links count total_link Count the number of total links (total links = internal links + external links) 13 Keyword count h1_kcount Count how many times the keyword H1 mentioned in all the H1s 14 Keyword count h2_kcount Count how many times the keyword H2 mentioned in all the H2s 15 Keyword count h3_kcount Count how many times the keyword H3 mentioned in all the H3s 16 Keyword count p p_kcount Count how many times the keyword mentioned in all the paragraphs 17 Keyword in a_kcount 0 if keyword not in anchor text of any anchor text link, 1 if keyword in anchor text of any link 18 Keyword in footer footer_kcount 0 if keyword not in footer, 1 if keyword in footer 19 Keyword in URL link_kcount 0 if keyword not in URL, 1 if keyword in URL 20 Keywords in imalt_kcount Count the number of times keyword image alt mentioned in alt tag of images 21 Meta desc length meta_desc_len Count the length of the meta description. If no meta description, length = 0 22 Meta keywords meta_kcount Count the number of meta keywords count used 23 Page title used ti_used 0 if no page title tag used, 1 if page title tag used I. INTRODUCTION The rapid advancement of digital technology has led to unprecedented transformations across numerous sectors, with the healthcare industry being one of the most notably impacted.
This digital revolution has democratized access to health-related information, thereby empowering patients and consumers to make informed decisions about their health. The widespread availability of internet resources has also intensified the competition among healthcare providers to capture and retain consumer attention online. In this context, Search Engine Optimization (SEO) emerges as an indispensable element of digital marketing strategies aimed at enhancing visibility, user engagement, and ultimately, organizational success. The Growing Digitization of Healthcare The global healthcare market has experienced robust growth, driven by technological advancements and increasing health awareness among populations.
According to The Business Research Company, the global healthcare market is expected to grow from $8.45 trillion in 2018 to $11.9 trillion by 2022, reflecting a compound annual growth rate of 8. This growth is mirrored in the digital realm, where health-related searches form a significant portion of internet activity. Google reports that one in every twenty searches is related to health [2], underscoring the critical role of digital platforms in the dissemination of health information. The Crucial Role of SEO in Healthcare Despite the potential of digital platforms, many healthcare providers struggle to navigate the complexities of online marketing.
The sheer volume of information and the dynamics of search engine algorithms can be daunting, often leaving valuable content obscured in the vastness of the web. SEO, therefore, becomes a critical tool for healthcare entities to enhance their online presence. Effective SEO strategies can lead to improved search rankings, greater website traffic, and increased patient engagement, which are essential for 1 healthcare providers in an increasingly competitive market. A study by Moz in 2021 highlighted that the top five organic search results on Google receive approximately 67.6% of all clicks, with a significant drop in visibility beyond the first page of results [3].
This statistic underscores the importance of a strong SEO strategy; visibility equates to accessibility in the digital age. Challenges and Strategic Imperatives Despite its importance, the application of SEO in the healthcare industry faces unique challenges. These include maintaining compliance with health information privacy laws, such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States, and ensuring the accuracy and sensitivity of the content, given the potential implications on users' health behaviors and outcomes. Moreover, the rapidly evolving nature of search algorithms necessitates continuous learning and adaptation among digital marketers.
There remains a gap in the specific knowledge and strategic implementation of SEO tailored to the healthcare sector. Many organizations continue to employ generic SEO strategies that may not account for the unique aspects of healthcare services and patient engagement. This thesis aims to bridge this gap by developing a nuanced understanding of SEO factors specifically influential in the healthcare industry and proposing targeted strategies that can lead to more effective web presence management. Objective and Structure of the Thesis This research is structured to address these complexities through a detailed investigation of SEO effectiveness, employing advanced machine learning models: LightGBM and XGBoost to predict and improve webpage rankings.
By analyzing the Everyday Health website, this thesis not only provides theoretical insights but also offers practical, actionable strategies tailored to real-world needs in the healthcare sector.