人工智慧（AI）運用於疾病風險的預測

醫療行為有太多的風險且常無法事先預測，尤其是在要面對高風險病人進行高風險處置時，不只病人擔心、家屬擔心，醫療人員更為擔心。這些擔心都源自於：目前常規醫療並無法對於個別病人就其自身帶有的風險因子，加上醫療所帶來的風險因子，做一個量身訂做的個人化綜合疾病風險預測，這是一個個人化精準醫療疾病風險預測的概念，也是目前醫學上最先進的發展方向。如果我們可以依照每個個人本身所帶有的不同風險因子，再合併醫療行為所帶來的風險作一個統合運算，借以得知病人於醫療過程中可能會發生那些併發症，其概率如何，如死亡、住加護病房、腦中風、心肌梗塞、大出血、休克...等等，並且知道會引發這些併發症的危險因子有那些？這些若能事先知曉、事先防範，將有機會大幅降低醫療療程中的危險性、增加病人安全性、減少醫療人員的壓力、減少醫糾…等。

在近幾年國際上有少數先驅研究證實，應用人工智慧（AI）於醫療大數據庫的運算上（以結構化的數據資料），有機會達到以上的目的，但實際上，在國際上能落地使用於臨床醫療的案例則少之又少，其原因是在這一系列的AI建置流程，需要許多專業醫療人士、資訊工程師、AI工程師的投入及眾多前置作業及單位主管的認同及大力協助，當然也須資金設備的投入。另，在國際上罕見有一家醫院能就醫院內所面臨的諸多疾病風險預測難題，集結各方面的人才，做一個整合性且可互動的AI運算平台建置。
 

Artificial Intelligence in Patients' Risk Prediction

Any medical treatment has potential risks and unpredictable results, especially when providing high-risk interventions to high-risk patients. Currently, personalized disease risk prediction based on the patient's disease risks and treatment risk factors is not available. Disease risk prediction for personalized and precision medicine is an emerging concept and approach. By calculating individual patient’s and treatment risk factors to predict probability of complications (such as death, ICU admission, stroke and hemorrhage), the treatment risks, stress in healthcare workers, and medical malpractice can be reduced while patient safety can be strengthened.

A few pioneer studies have shown that using artificial intelligence (AI) to analyze structured healthcare big data might have potential for developing the personalized risk prediction models. However, there is not yet a hospital in the world to develop an integrated, interactive AI platform since AI-based processing system requires great devotion from medical professionals, information technology specialists and AI engineers as well as a tremendous amount of investment and venture capital.