Risk Assessment of Medical Insurance

To reduce the risk of loss, the insurance companies intend to predict the individual medical cost based on the data history, which are also expected to predict the total medical cost of the policyholder in the future. Through the prediction model, each input variable represents an uncertain factor that affects the medical costs. Therefore, if it can fully know these uncertain factors, it may really control the risk of claims and work out the compensation schemes accurately.

#### Solution

Through the SPSS Statistics Monte Carlo simulation method, it can’t only establish an cost prediction model based on the data history, but also simulate the input variables of various uncertainties. The prediction model can be substituted to calculate and establish the overall medical costs. In this way, it could know all possible medical costs.

• Establish prediction model based on the data history, and determine the input variables that affect the costs of diabetes treatment.
• Match each input variable with appropriate probability distribution (Through the SPSS Statistics Monte Carlo simulation method, the variables can be easily matched with the appropriate probability distribution, as shown below). • The input variable value is generated randomly through the appropriate probability distribution.
• The input variable value generated randomly is substituted into the established prediction model to calculate the medical costs.
• Generate input variable values randomly in a repeated way (generally steps 3~4 are performed repeatedly for thousands of times); it will generate a large number of medical costs and their occurrence probability, so as to form a probability distribution function.
• • After the appropriate probability distribution, the observation value will be generated randomly from the simulation: a total of 19,153 entries

#### Application Effect

Monte Carlo simulation method is used to perform a large number of simulations, so as to generate input variables of various uncertainties (age, glucose, and income), and to calculate the medical cost by using the model. • The total medical cost is obtained through the simulation data, from which it could also know the information of medical cost in details  Through the probability density function, we can not only process the point estimation--the average of the total medical cost, but also know the probability distribution of the total medical cost. Therefore, we can know the upper/lower limit of the insured's total medical cost (for example: 95% of the insured are under the upper limit of NT\$22,866 for medical costs). It could facilitate in getting complete information on claims payment and evaluating the the claims risk.

Relevant Cases：

#### AsiaAnalytics Taiwan Ltd.

5F, No. 356, Sec. 1, Neihu Rd., Neihu Dist., Taipei City 11493, Taiwan (R.O.C.)

TEL：+886 2 7728 7958 FAX: +886 2 2627 0667

e-mail：service@asia-analytics.com.tw  