資料驅動形塑數位健康路徑：COVID-19與風濕疾病觀點

研究目的：2015年，聯合國發布「永續發展目標」，當中目標三係指「確保及促進各年齡層健康生活與福祉」，為響應此，同時因應數位發展進程，世界衛生組織進一步發布《2020-2025 全球數位健康策略》，藉由數位健康路徑，引領全球發展。健康資料無疑是驅動醫療照護典範轉移之關鍵，因此，本論文旨於建構以資料驅動為核心之數位健康路徑，期促進數位健康發展，實現全民健康目標。研究方法及資料：本論文以世界衛生組織所發布之數位健康路徑為基礎，綜整分析當前國際數位健康政策、規範等趨勢，進而提出以資料作為觀點的嶄新路徑構想，並藉由兩個實證案例，從廣義、狹義之健康資料進行構想驗證。廣義健康資料面向，通過台灣於COVID-19期間，一系列數位導入防疫策略之舉措進行探討；狹義健康資料面向，擇定風濕疾病，導入人工智慧(Artificial Intelligence, AI)開發僵直性脊椎炎(Ankylosing Spondylitis, AS)嚴重程度檢測模型，進行臨床決策輔助(Clinical Decision Support, CDS)探討。研究結果：本論文所提資料驅動之數位健康路徑構想，經實證案例驗證具可行性，更揭示其應用價值。廣義之實證案例，表明台灣有意義地將數位導入防疫策略，從QR Code到口罩分配，成功體現資料治理、跨域協作，有助多元創新發展；狹義之實證案例，基於改良式僵直性脊椎炎X光影像評分法(The modified Stoke Ankylosing Spondylitis Spinal Score, mSASSS)所開發之模型與專家所得之評估結果相互比較，均符合預期，其中，針對嚴重等級3分之影像，模型靈敏度(Sensitivity)為0.988、特異度(Specificity)為0.974，評估準確率可達九成，體現新興科技技術，有助於臨床照護變革。結論與建議：本論文重新詮釋廣義、狹義之健康資料，提出資料驅動之數位健康路徑；然，如要全面推展施行，仍尚有健康資料格式、管理，甚至是隱私及倫理等諸多挑戰需克服。健康是每個人的基本人權，順應數位發展趨勢，宜持續探索可兼顧個人隱私保護與健康資料利用之治理機制，促進被動轉主動之典範轉型，發展以人為本的體系，以塑造更加健康的未來。

Objective: In 2015, the United Nations released the Sustainable Development Goals (SDGs), of which Goal 3 refers to "Ensure and promote healthy life and well-being for all ages". In response to UN SDG 3 meanwhile echoing the digital development process, the World Health Organisation further released the Global Strategy on Digital Health 2020-2025, leading global development through the digital health path. Health data is undoubtedly the key to driving the paradigm shift in healthcare. Hence, this thesis aims to establish a digital health roadmap with data-driven as the core, with the expectation to promote digital health development and achieve the goal of universal health coverage.Methods and Materials: This thesis is based on the digital health roadmap published by the World Health Organisation, comprehensively analyses the current trend of international digital health strategies and norms, and further proposes a new hypothesis based on data perspective. Through two empirical cases, the hypothesis is verified from the broadly and narrowly defined health data. In terms of broadly defined health data, a series of actions involving digital integrated strategies to prevent the pandemic in Taiwan during the COVID-19 period are discussed; in terms of narrowly defined health data, rheumatic disease is selected, and artificial intelligence (AI) is introduced to develop ankylosing spondylitis (AS) severity inspection model to discuss clinical decision support (CDS) research.Results: The concept of the data-driven digital health roadmap proposed in this thesis has been proven to be feasible through empirical cases, and its application value has been revealed. In the case of broadly defined health data, Taiwan's meaningful employment of digital technologies in pandemic prevention strategies, from QR code to mask distribution, successfully reflects data governance and cross-domain collaboration, and promotes diversified innovation and development; in the case of narrowly defined health data, the results obtained by the model developed based on the modified Stoke Ankylosing Spondylitis Spinal Score (mSASSS) are compared with the evaluation by experts, and both are in line with expectations. Among them, for images assessed to have a severity level of 3, the model's sensitivity and specificity are 0.988 and 0.974, respectively, with an overall accuracy reaching 90%, reflecting that emerging technologies are conducive to clinical care reform.Conclusion and Suggestion: This thesis reinterprets health data from broad and narrow perspectives, and proposes a data-driven digital health roadmap. However, the full implementation of the roadmap still needs to address many challenges, such as health data format, management, and even privacy and ethics issues. Health is a fundamental human right. In line with the trend of digital development, we should continue to explore governance mechanisms that take into account both personal privacy protection and health data utilisation, promote the paradigm shift from reactive to proactive healthcare, and develop a human-centred system to shape a healthier future.