以人工智慧(AI)輔助自學程式編碼於醫療之應用初探

AI-Supported Self-learning for Enhancing Medical Programming Skills: A Preliminary Study

目的：功能強大的人工智慧(AI) ChatGPT 3.5於2022年底發佈後舉世轟動，導致各行各業思考導入AI的可能，我們也嘗試以AI輔助程式編碼於醫療的實例應用。

方法：以目前最新版本的DeepSeek、Microsoft Copilot、Google Gemini及ChatGPT等AI模型，輔助Excel 之Visual Basic For Application (VBA)及JavaScript編碼，前者選取工作表列之任何一個工作表/滑鼠右鍵/檢視程式碼或者Alt+F11開啟VBA編輯器，檔案儲存成.xlsm或.xlsb副檔名，後者存成.html。兩者以BMI 計算、VBA多加邏輯判斷If做舉例說明。使用電腦型號為ASUS筆記型LAPTOP-TSDMU3MP，配備 Intel Core i7-1165G7 2.80GHz 處理器、運行Windows 11及Microsoft 365軟體。

結果：免費AI有使用次數的限制或問不出結果時，可交替使用。本研究探討AI在醫療領域中，輔助程式編碼應用的可行性，發現AI可作為開發者的得力助手。以BMI計算及成人預防保健為提示詞(prompt)為例，發現給予AI提示，即可自動生成VBA及JavaScript編碼，會有使用步驟、功能說明、程式碼說明、範例運行及注意事項，且會除錯及優化程式碼等。對於沒有程式語言基礎的初學者可輕易上手，再者，透過不同的輸入條件，及於不同AI，可產生不同的結果以資比較，可達到程式語言的自我學習。If使用法之外，詢問AI，發現還可使用Select case及IIF，用法更簡潔。

結論：以AI輔助，可大大降低進入程式編碼的門檻，因為會有註解及程式碼解釋，因此也可達到初學程式語言者極有價值的學習參考。

 

Purpose: Since the groundbreaking release of the powerful ChatGPT-3.5 model in late 2022, various industries have been exploring ways to incorporate AI into their operations. Our study aimed to assess whether AI assistance could effectively support medical professionals in learning and implementing programming skills for healthcare applications.

Methods: The latest versions of AI models—DeepSeek, Microsoft Copilot, Google Gemini, and ChatGPT—were used to assist in coding with Excel’s Visual Basic for Applications (VBA) and JavaScript. For VBA, select any worksheet/ right-click / View Code or Alt+F11 to open the VBA editor. The file was then saved with the extensions .xlsm or .xlsb. For JavaScript, the file was saved as .html. Both examples demonstrated body mass index (BMI) calculation, with the VBA version including an additional logical If statement for illustration. All programs were executed on an ASUS laptop (LAPTOP-TSDMU3MP) equipped with an Intel Core i7-1165G7 2.80GHz processor, running Windows 11 and Microsoft 365 software.

Results: When an AI tool reached its usage limit or failed to generate satisfactory results, other models were used as alternatives. This study demonstrated the feasibility of AI-assisted programming applications in the medical field, showing that AI can serve as a valuable assistant for developers. Using "BMI calculation" and "adult preventive healthcare" as example prompts, we found that providing these instructions to AI-enabled automatic generation of both VBA and JavaScript code, complete with implementation steps, functional descriptions, code explanations, execution examples, and important notes. The AI systems additionally performed debugging and code optimization. This approach allowed beginners without any programming background to get started easily. Moreover, by applying different input conditions across various AI platforms, users could obtain diverse outputs for comparison, thereby facilitating self-directed learning of programming languages. Beyond the conventional If-Then method, consultation with AI revealed alternative approaches using Select Case and IIF statements, which provided more concise coding solutions.

Conclusion: AI assistance can significantly lower the barrier to entry for programming. The inclusion of comments and code explanations offers valuable learning support and reference materials for programming novices.