近年來，台灣各大保險公司陸續推出多項外溢型保險商品，允許保戶透過運動行為降低個人風險，進而獲得保費減免。然而，此類商品在風險評估與定價方面，往往需仰賴大量醫學研究數據支撐。由於保險公司內部累積的經驗數據仍屬有限，現行實務多以同業的損率與回饋比作為參考，缺乏一套系統化且科學的風險評估與保費定價方法。
為解決上述問題，本研究蒐集65篇探討運動與死亡率關聯之相關文獻，並根據各研究中受試者的運動強度，採用代謝當量（METs）作為衡量指標，將樣本劃分為低、中、高三種運動水平。此外，亦納入多項潛在調節變數，包括性別比例、平均年齡、人均所得、運動測量方式等，透過元迴歸分析（Meta-regression）方法，深入探討運動強度與死亡率之間的真實關係。研究結果顯示，相較於低運動組別，中度運動組別可降低約35%的死亡率，高度運動組別則可降低約48%的死亡率。
基於此統計模型，本研究進一步量化運動對死亡風險降低的影響，進而估算出合理的保費減免幅度。隨後，我們將該模型應用於實際保險商品之定價架構，並與僅使用平均效應的傳統統合分析模型進行比較。結果顯示，元迴歸模型在不同繳費期間與保險期間下，所估計之信賴區間普遍減少約20%至30%，顯示本研究方法在估計精度與風險分級應用上具有明顯優勢。
本研究亦據此模型設計不同運動等級對應之保費回饋機制，並以現金回饋形式提供保戶實質誘因。同時，我們建構了一套量化運動強度與死亡風險減少效果之估算公式，作為保險公司設計外溢型商品時的依據。

This study looks at the challenge of pricing behavior-based insurance products that offer premium discounts to policyholders who stay active. Because insurers often lack enough internal data, they usually rely on industry averages instead of scientific models.
We reviewed 65 studies on the link between physical activity and death rates, using METs to group people into low, moderate, and high activity levels. A meta-regression analysis, which also considered factors like age, gender, income, and how activity was measured, found that moderate and high activity levels can lower the risk of death by about 35% and 48%, respectively.
Using this model, we estimated suitable premium discounts and compared the results with traditional meta-analysis methods. Our model gave more accurate estimates, with confidence intervals that were 20–30% narrower.
We also suggested a step-by-step discount system based on activity levels and created a formula that helps insurers link physical activity to lower death risk—giving them a more scientific way to design and price these products.