[[alternative]]Converting Taiwan's National Health Insurance Data into FHIR Standard Data Model

[[abstract]]隨著醫療科技的進步，研究人員不斷的深入探討如何從研究結果中發展有效治療的方法。任何有效的治療都是需要經過實驗室研究或動物試驗，再透過嚴謹的人體臨床試驗來證實該研究的有效性和安全性，才能廣泛應用於常規醫療中。而臨床試驗則是針對臨床上或病人遇到的問題制定研究條件，設計一個合適的研究以探討新的方法是否能達到更好的結果。一般而言，隨機對照試驗是在臨床試驗中最常用的方法，為了達到內部有效性，在研究設計中提出一些限制，來控制可能會影響結果的因素，因而也侷限了資料適用性的範圍，若將研究結果應用於真實世界中，可能會出現差異。在21世紀治癒法案(21st Century Cures Act)的推動下，各國皆積極推行真實世界數據(Real-World Data, RWD)相關指引文件及計劃，並發表支持觀點，顯示各國對RWD的重視，RWD是在日常生活中蒐集資料，與傳統臨床試驗相比，不具嚴格的條件限制和監管機制，可以減少整個研究所需的時間和金錢，也能廣泛用於不同研究中，但在蒐集資料的過程中，會遇到資料分散、樣本代表性不足、資料完整度不夠、資料品質不一致等問題。臺灣的全民健保資料庫是一個具有樣本代表性和全面性的RWD資料庫，若能善用臺灣的健保資料庫則可達成RWD以低成本換取高價值的實現。為能建立一個真實世界數據資料庫，解決與醫療資訊系統間無法交換的問題，改善資料品質和整合流程，滿足資料之原始目的，本研究以全民健保資料作為轉換物件，根據國際醫療資訊交換標準HL7 FHIR(Fast Healthcare Interoperability Resources)格式，制定資料欄位對應表，利用Python撰寫格式轉換模組，並將由全民健保資料的200萬人抽樣資料轉換為FHIR標準之資料格式，將資料根據欄位對應表轉換成FHIR標準的格式，再透過符合FHIR標準的伺服器驗證資料格式之正確性，並將其儲存至資料庫中。資料轉換完成後，使用者可以RESTful API的方式進行資料查詢和管理，解決醫療系統間資料交換的問題，除了將應用RWD來提升研究的準確與實用性外，也能擴大資料之加值應用領域。[[abstract]]With the advancement of medical technology, researchers keep exploring the best treatment for diseases. However, any research involves the following steps. First, researchers conduct experiments in laboratory studies or animal tests. After that, confirm the effectiveness and safety of human subject research. And finally, the results can be used as a routine medical treatment. A clinical trial is to formulate research conditions for clinical or patient problems, design an appropriate study to explore whether the new method can achieve better results. In clinical trials, to achieve internal validity, randomized controlled trials have to set up many limitations, which may affect the results. Thus, the results may restrict data applicability, and they cannot reflect the true situation in the real world.In the promotion of the 21st Century Cures Act, many countries have promoted real-world data (RWD) guidelines and plans, showing that countries are acknowledged the importance of RWD. RWD is collect data in daily life, it has no conditional restrictions and supervision mechanism, which reduces the time and cost required for the entire study, and can be widely used in a different field. However, in the process of collecting data, there are problems such as different sources, unrepresentative, incompleteness, and inconsistent quality of the data. Taiwan’s National Health Insurance Database is a representative and comprehensive RWD database. This research used the national health insurance data as a converting object and created a filed mapping table based on HL7 Fast Healthcare Interoperability Resources (FHIR) filed format. The research implemented a converting module in Python and converted 2 million national health insurance data into FHIR standard format, then established an FHIR standard verification server to verify the correctness of the data format and save it onto a database. Once the data is converted completely, users can use the RESTful API technique to query and manage the data. The research result not only resolves the data interoperating and exchanging difficulties among different medical information systems and also improves the data quality and integration process, thus increasing the data added value in many fields application