Open search panel Close search panel Open menu Close menu

May 16, 2018

NTT DATA Corporation

Free Trials Utilizing Lifestyle Disease Risk Prediction Technology Performed with Insurance Companies

NTT DATA Corporation (Head Office: Koto-ku, Tokyo, Japan; President and Chief Executive Officer: Toshio Iwamoto; hereinafter, "NTT DATA") and NIPPON TELEGRAPH AND TELEPHONE CORPORATION (Head Office: Chiyoda-ku, Tokyo, Japan; President and Chief Executive Officer: Hiroo Unoura; hereinafter, "NTT") are recruiting insurance companies that agree to a free trial to demonstrate the effectiveness of NTT's lifestyle disease risk prediction technology (*1) in insurance companies' product development, member joining assessments, post-joining health improvements, and so forth. The aim of NTT DATA and NTT is to utilize this technology, a part of NTT's AI-related technologies (corevo® (*2)), in the insurance industry.

1.Background & Details

Currently, various efforts toward prevention and health promotion are being carried out in Japan with the intention of extending the country's healthy life expectancy. Lifestyle diseases in particular have become a social problem that leads to such issues as increased medical costs and lower employee productivity at companies. In order to prevent these diseases, there is a need to improve the lifestyle habits and grasp the health risks of individuals.

Accordingly, NTT has established, as a part of its AI-related technologies (corevo®), a method that makes it possible to conduct highly precise analyses, even with uneven or scarce data. The company has realized a way to predict the probability of developing a lifestyle disease in the future based on data acquired from health examinations (hereinafter, "health examination data").

Insurance companies need calculations regarding future disease risks for health improvements and other purposes. By utilizing this disease risk prediction technology, insurance companies will no longer need to acquire and analyze large quantities of health examination data and medical receipt data, nor to deal with personal information requiring special consideration, as stipulated in the Amended Act on the Protection of Personal Information (*3). It is predicted that this technology will become vital as the insurance industry utilizes medical and health data to an even greater degree in the future.

2.Regarding the Disease Risk Prediction Technology

With the analysis of health examination data, there are cases of data omissions when patients do not continue to receive examinations, cases of data for limited periods when only examination results from a certain period are retained, and cases of scarce sample data when diseases with few outbreaks are handled. In such cases, it has been difficult to analyze data with a high degree of precision.

To address this, NTT Service Evolution Laboratories and NTT Communication Science Laboratories have carried out ranking learning (refer to Figure 1) that also uses data from individuals who have not developed a disease, which is retained for only a short period, in order to achieve highly precise analyses even with uneven or scarce data. In this way, the laboratories has realized a technology that predicts the risk of lifestyle diseases (diabetes, hypertension, and dyslipidemia) based on health examination data (refer to Figure 2). This disease risk prediction technology has been applied to the health examination data analysis business at a portion of the organizations using Health Data Bank® (*4), provided by NTT DATA (approximately 100,000 individuals for up to six years). It has been verified that highly precise predictions are able to be obtained (*5).

Figure 1: The ranking learning using data including individuals who have not developed a disease of interest. The data were collected for both short and long periods.

Figure 2: An image of lifestyle disease risk prediction.

3.Regarding the Trials with Insurance Companies

NTT DATA will provide an environment for testing the disease risk prediction technology. The effectiveness of the prediction technology will be tested with multiple insurance companies who request testing. Applications from insurance companies will be accepted until the end of June, 2018. After the applications have been received, the insurance companies and NTT DATA will adjust the testing plans, NTT DATA will be entrusted with the testing data prepared by the insurance companies, the prediction technology will be applied, and the prediction results will be returned to the insurance companies. Based on the prediction results, the effectiveness of the technology, the challenges for introducing that technology, and so forth will be tested together with the insurance companies.

Figure 3: Flow of the Trial

Roles of the Three Companies in the Trial

Insurance Companies

  • Consideration of testing plans
  • Preparation of testing data
  • Testing of prediction result effectiveness


  • Provision of prediction technology


  • Construction and operation of prediction environment
  • Advancement of testing

4.Future Developments

NTT and NTT DATA are conducting research on not only the prediction of lifestyle disease risks, but also the prevention of increased severity after diabetes has developed. Concerning disease risk prediction as well, based on this trial's results and the needs that are present, the companies plan to accelerate their research with the aim of targeting more diseases. NTT and NTT DATA will combine their technologies and aim to commence services for insurance companies within FY 2018. They plan to conduct service expansions following the start of the service as well.

The NTT Group will fuse the cutting edge technologies of its research institutes with the information and communication technologies of its companies, seek to create businesses that support the health of citizens, and promote the activities of such businesses.

  1. This is a technology that uses data from test values in health examinations, the results of patient interviews, and so forth to predict the risk of lifestyle diseases (diabetes, hypertension, and dyslipidemia).
  2. These are AI-related technologies of the NTT Group. "corevo®" is a trademark of NIPPON TELEGRAPH AND TELEPHONE CORPORATION. (
  3. However, when predicting the disease risks of individuals, it is necessary to use those individuals' health examination data, which is personal information requiring special consideration.
  4. "Health Data Bank®" is a cloud-based health management service provided by NTT DATA. It has been introduced to 1,500 organizations, and is being utilized for the health management of over 3.5 million people (as of the end of March, 2018).
  5. For the prediction of the likelihood of developing diabetes, 0.81 was achieved using the Kendall rank correlation coefficient, an index for evaluating rank. This indicates that in this case, the technology was able to distinguish individuals likely to develop the disease with an accuracy rate of 90%. Further, all predictions for the risk of developing diabetes in year N (1≤N≤6) achieved 0.9 or over in ROC AUC, an index for evaluating prediction accuracy. An ROC AUC score equal to or greater than 0.9 is generally said to possess excellent prediction accuracy.

Inquiries Regarding This Matter


Kitajo, R&D Produce Group,
Research and Development Planning Department

Innovative R&D by NTT NTT Has Instituted a Logo to Represent R&D Activities.

Information is current as of the date of issue of the individual press release.
Please be advised that information may be outdated after that point.