Japan's population began to age at a rapidly accelerating rate from the 1970s. This has become a major social problem. As Japan's population grows older, the amount spent on medical care rises year by year. Peoples' health has a major impact on the economy and on the activities of the companies that support it. Illness among people who are in their most productive years drives down productivity. Early retirement causes the contraction of the labor force. Health insurance premiums paid by companies are rising. This state of affairs has led to a greater interest in preventative medicine, and a growing number of companies are taking an active approach to health and productivity management.
Toshiba is engaging in various initiatives in the precision medicine field to support peoples' healthy lives. We are using the power of digital technology to support preventative medicine that also encompasses future generations. Let's look at one of those initiatives, Toshiba's disease risk prediction AI service, which leverages the power of AI technology.

Lifestyle-related diseases are becoming a societal problem

One of the major societal problems currently being faced by Japan is the rise in lifestyle-related disease and the resulting increase in medical costs. According to the Ministry of Health, Labour and Welfare's "2018 Overview of Medical Spending in Japan," medical spending related to lifestyle-related diseases accounts for 20% of total medical spending.

Lifestyle-related diseases have a higher likelihood of occurrence when there is an overlap of genomic factors, in which a person's constitution makes them more prone to contracting an illness, and environmental factors, such as the environment people grew up in or their current living environment. They are also more likely to happen the older a person is. Diabetes, for example, is difficult to permanently cure once it is contracted. It is also often accompanied by complications, such as retinopathy, nephropathy, and neurological disorders. Because of this, it significantly reduces the quality of life (QOL) of its sufferers. Providing health instruction that is tailored to individuals' circumstances early on, before symptoms appear, is important for preventing lifestyle-related diseases such as diabetes*. This is extremely effective in reducing medical costs and improving QOL for individuals.

* Diabetes is broadly divided into two types: Type 1 and Type 2. Type 2 diabetes is considered a lifestyle-related disease.

However, there are three difficult challenges involved in providing this health instruction: (1) accurately predicting the risk of lifestyle-related disease onset, (2) providing objective health instruction based on individuals' data, and (3) setting lifestyle habit improvement targets that are achievable for the person undergoing the health examination. The reality is that even when health instruction is given, many examinees do not make lasting behavioral changes that would prevent the future onset of the disease. 

How does Toshiba use AI technology to predict disease risk?

To solve these problems, the national government is leading efforts to implement data-driven health reforms by analyzing big data from the medical field and promoting the use of AI. AI is needed to provide support for complex decisions that are difficult to make when using conventional data analysis alone, and there are demands for improvements to be made to clinical practices to enable doctors to focus on more advanced decision-making tasks while providing objective health guidance to numerous patients.

Given this societal trend, Toshiba has decided to help address these issues by developing lifestyle-related disease-related data analysis technologies and AI. This includes disease risk prediction, in which customers are informed of dangers based on risk information that has already been provided to them, solutions for preventing diseases from becoming more severe, and the development of optimized prevention methods, which is still in the research and development stage (Fig. 1). 

"The disease risk prediction AI service" inputs one year's worth of health examination data (hereinafter referred to as "health examination data"), and by six years from now, diabetes, hypertension, dyslipidemia, renal dysfunction, liver dysfunction.  And it is a service that predicts the possibility of developing six lifestyle-related diseases called obesity.

One of the diseases that can be predicted by this service is diabetes. The AI model we have developed predicts diabetes with an over 90% accuracy rate. This high level of accuracy has been made possible by the broad-ranging AI technologies that Toshiba has accrued in its over 50 years of AI research, together with massive amounts of medical examination data and medical expense statement data for Toshiba Group employees, collected over the course of multiple years. This involves the handling of employees' precious personal information, so we use Toshiba's information security technologies such as data anonymization to appropriately manage and handle the data.

Toshiba's AI researchers have a tradition of placing great importance on field expertise and knowledge when utilizing data, and they value dialog with people working in the field. AI researchers, occupational health physicians, public health nurses, and other professionals have collaborated and discussed how this data should be used and the direction to be taken by the AI that is developed.

Through these efforts, Toshiba has created an AI prediction model based on machine learning and used it to perform analysis of the patterns shared by people who develop lifestyle-related diseases, their associated traits, and more. When creating an AI model, the quality of the data used for learning is extremely important. Machine learning requires a huge amount of learning data, so the quality of the data has a particularly large impact on the predictive accuracy of the AI model.

For this service, we used medical examination data and medical expense statement data as learning data. Changes in result values in the medical examination data and information regarding medications in the medical expense statement data were used to determine whether people had lifestyle-related diseases. The AI model then studied lifestyle habits and how medical examination values in the medical examination data changed over time.

Toshiba has a large number of employees and a low attrition rate. This means that we have a large amount of data covering multiple years for each employee. We were able to use this data in machine learning to develop an AI model that produces highly accurate predictions.

We also used the "random survival forest" method, a statistical machine learning method, and applied our own unique optimization technologies to achieve a high level of predictive accuracy. We used the AI model to predict who had a high risk of developing diabetes within the next three years. The results that were produced had a high predictive performance, with an AUC* of 0.96.

* AUC (Area Under the Curve):  An indicator of prediction quality. AUC is a number between 0.0 and 1.0. The closer to 1.0, the greater the predictive accuracy. An AUC of 0.5 is the equivalent of random prediction.

Not only does the model have high predictive accuracy, it is also exceptionally generalizable, effective even when used with unknown data. The same AI model was applied to a different group of samples, separate from those of Toshiba's health insurance union, and achieved an AUC of 0.94. These results were presented both at the Scientific Meeting of the Japan Society of Ningen Dock* and at the World Congress on Ningen Dock**.

* 60th Scientific Meeting of the Japan Society of Ningen Dock (2019) "Development of an Algorithm for Predicting the Onset of Diabetes through AI Analysis of Health Examination Data"
** Joint Meeting of the 27th International Health Evaluation and Promotion Association & 4th World Congress on Ningen Dock (2020) "Prediction of diabetes, dyslipidemia, hypertension, liver function & renal function using AI"

Our solutions for preventing complications use data such as medical examination and medical expense statement data to identify individuals at high risk of suffering diabetes complications and to warn them of this danger. We are also conducting research in conjunction with a university with the aim of using Toshiba AI technologies to prevent increases in the severity of diabetic nephropathy. Our AI model is being used to divide diabetic nephropathy patients into multiple categories and develop optimized preventative methods that are stratified and systematized for each individual risk. These preventative methods assist with providing patients with guidance regarding lifestyle habits, tailored to their individual risks and symptoms. We believe that this can help them enjoy a greater QOL. 

Contributing to lifestyle habit improvement recommendations for medical examinees

Let's look at a specific usage example of this disease risk prediction AI service, which can identify future disease risks.

When provided with a year's worth of medical examination results for a person, the service can predict the likelihood of their contracting any of six diseases, including diabetes and high blood pressure, within the next six years. Furthermore, when predicting these risks, feature quantities (lifestyle habits) can be extracted from the AI model and used to assist in providing data-based, concrete health guidance and advice. The ability to quantitatively assess current conditions and future disease risk make this guidance more convincing for those receiving medical examinations and contributes to their adoption of behavioral changes to improve their lifestyles.

Furthermore, this service is expected to assist with the creation of health guidance in which lifestyle improvement approaches are formulated in conjunction with examinees. For lifestyle improvements, in particular, it is a known fact that improvement targets set by examinees are more effective than those set by doctors. For example, when the likelihood of contracting diabetes is shown to examinees in the form of a numerical probability, examinees become more focused on and dedicated to the long-term prevention of contracting this chronic disease. Furthermore, by setting improvement targets that match their own lifestyles, examinees become more proactive with regards to making and maintaining lifestyle improvements. This has the potential to consistently lead to more positive outcomes. 

Visualizing employee health risks for use in corporate health and productivity management

A growing number of companies dedicated to health and productivity management are inquiring about our disease risk prediction AI service.

Health and productivity management (H&PM) is an approach that considers the health management of employees from a corporate management perspective and promotes it through strategic investment in employees' health. Greater attention is being paid to H&PM in recent years due to factors such as the shrinking of the workforce, and the number of companies introducing H&PM measures is rising. Companies which actively implement H&PM are seeing the benefits of employees enjoying greater health, vitality, and productivity. These benefits include lower rates of attrition due to illness, greater organizational vigor, and improved productivity. Recently, many companies have been publicizing their H&PM measures on their websites, so H&PM is also serving to improve their corporate image.

H&PM has numerous potential benefits, and Toshiba's disease risk prediction AI service can be used in H&PM to provide more detailed health management and more accurate health guidance. For example, it can identify job positions and geographical regions highly correlated with lifestyle factors such as insufficient exercise due to office work or commuting by car, insufficient sleep hours, or eating foods high in salt. It can also perform comparative analysis within the industry. It can then use the results of various analyses to visualize employees' health risks and assist with deliberations regarding concrete countermeasures.

Expanding support for health promotion by linking a variety of data

We also believe that this service can be linked with various other services and data to provide support for peoples' healthy lifestyles and make specific recommendations that promote greater health. For example, the predictions of the disease risk prediction AI service could be used to create services that provide guidance regarding meal contents or services that recommend dietary supplements and health and fitness products.

Toshiba has introduced the Toshiba Group's "Smartreceipt" electronic receipt service in its employee cafeterias to contribute to the improvement of employees' dietary habits. We are also planning and performing pilot tests of digital health services that use AI, such as combining the dietary information collected from Smartreceipts with medical examination data to provide health guidance based on the characteristics of individual employees. Our aim is to promote the health of employees and implement health and productivity management (Fig. 2). 

In the era of the "100-year life", we must create societal environments in which people can enjoy good health throughout their long lives.

We will help support peoples' healthy lives and contribute to the creation of societies of abundance and longevity through Toshiba's disease risk prediction AI service and the other services we offer by leveraging our AI technologies and expertise.  

  • The corporate names, organization names, job titles and other names and titles appearing in this article are those as of June 2021.

>> Related information

Related articles