智慧醫療在保險業保單設計之研究－以糖尿病為例

RESEARCH ON THE DESIGN OF SMART MEDICAL INSURANCE POLICY IN INSURANCE INDUSTRY: A CASE STUDY ON DIABETES

楊淑美(Shu-Mei Yang) ； 謝邦昌(Ben-Chang Shia)

《Journal of Data Analysis》 13卷6期 (2018/12) Pp. 183-218

https://doi.org/10.6338/JDA.201812_13(6).0008

摘要

由於國人經濟結構的蓬勃發展，逐漸老化的人口結構，飲食型態及生活習慣逐漸西方化，運動機會減少而致肥胖盛行率增高，再加上良好的醫療照護而延長平均餘命等因素，使得糖尿病的發生率與盛行率逐年快速上升。目前保險公司對糖尿病族群，大多為拒保或重度加費，實在相當可惜。若保單設計能運用智慧醫療，針對糖尿病客戶提供壽險及健康險保障，則應該可以大幅降低該族群及家屬之醫療負擔。糖尿病智慧保單模型的建立是透過了機器學習方法，將糖尿病對應成三種程度: 輕度糖尿病、中度糖尿病、重度糖尿病。依據糖尿病不同類型，從大數據資料分析以模型驗證，將糖尿病模型分三期而設計糖尿病智慧保單。也就是未來依照糖尿病不同類型的併發症，照護的方式也隨之改變，使得大部分（本研究為66.67%）的糖尿病可以被照顧到，根據糖尿病控制程度分類，而相對能降低糖尿病理賠的風險。本研究根據糖尿病文獻定義輕度、中度與重度糖尿病，本研究運用羅吉斯迴歸模型分析，其中輕度糖尿病人數為0，因此分析中度與重度糖尿病，除了導入糖化血色素、HDL膽固醇、身體質量指數、血壓四個變數之外，我們再增加了喝酒、吸菸及心臟疾病、年齡、性別與泌尿腎臟疾病六個變數合計十項變數，但結果只有糖化血色素、HDL膽固醇、身體質量指數、血壓四個變數顯著，相當有助於糖尿病嚴重程度之判斷，也符合文獻之定義。但喝酒、吸菸及心臟疾病三項變數對糖尿病影響不顯著，另外，年齡、性別、泌尿腎臟疾病之變數，因為不顯著故沒有影響力。

關鍵字

智慧醫療 ； 保單設計 ； 糖尿病 ； 機器學習 ； 外溢型保單

並列摘要

In this study, we applied logistic regression model in the modeling of diabetes intelligence policy by machine learning. Four input variables are blood sugar, lipids, blood pressure, and BMI as predictive factors, and it will come up with three classifications including mild, moderate and severe diabetes. Then the insurance companies can design diabetes intelligence policy according to this classification from big data analyzing and then verified by the model. We got no data from mild diabetes (0 person), hence analyze moderate and severe diabetes. In addition of introducing four variables: HbA1C, HDL, BMI, Blood pressure, we add extra six variables thus becoming ten in total. Alcohol consumption, smoking, and cardiovascular diseases are of no significance, and age, sex, urogenital and renal disease are of no significance as well. To sum up, it turns out that only the first four variables are significant for the outcome of diabetes, which corresponds to the references. In order to break the boundaries of time and space, we encourage insurance companies to design intelligence policy and use cloud computing technology combining diabetes group and ＂Telehealth＂ care in Smart Healthcare, which automatically sends and records down the data of blood sugar and blood pressure measured at home to the telehealth care center via internet. The data will be 24/7 regularly followed up by health care managers who also accordingly provide individualized patient education.

並列關鍵字

Smart Healthcare ； Policy Design ； Diabetes ； Machine learning ； Spillover insurance policies

參考文獻

Anderson, J. G.(2007).Social, ethical and legal barriers to e-health.International journal of medical informatics.76(5-6),480-483.

Centers for Disease Control. Prevention (2011) National Diabetes Fact Sheet: National Estimates and General Information on Diabetes and Prediabetes in the United States. Department of Health and Human Services, Centers for Disease Control and Prevention.

Della Mea, V.(2001).What is e-health (2): the death of telemedicine?.Journal of medical Internet research.3(2)

Eysenbach, G.(2001).Journal of Medical Internet Research is now indexed in Medline.Journal of medical Internet research.3(3),e25.

Gerber, T.,Olazabal, V.,Brown, K.,Pablos-Mendez, A.(2010).An agenda for action on global e-health.Health affairs.29(2),233-236.