A document search was performed for the listed publication years of 2000 to 2013 and the search terms of “active aging” and “healthy life expectancy” using the Web edition of the “Japana Centra Revuo Medicina” [7] which is a database of medical publications originating in Japan. It is a representative medical literature database of Japan that enables to retrieve on the web as many as approximately 7.5 million cases of information on academic articles and medical literature collected from 5600 periodicals in the fields of medicine, dentistry, pharmacy, nursing science and other related academic fields published in Japan. This is a well-established service that is used by almost all the universities in the fields of medicine, dentistry and nursing science in Japan.

Four articles were extracted using with the term, “active aging” and 1072 with “healthy life expectancy” (accessed at 8pm on 13 April 2014). Next, the conditions of “original articles” and “excl. conference proceedings” were applied to the extracted papers, narrowing the search to original articles, and to commentaries, reviews, and features.

Ultimately, the search yielded two original articles and one review/commentary/feature using the term, “active aging,” and 118 original articles and 212 reviews/commentaries/features using “healthy life expectancy.”

We combined the search results of “active aging” and “healthy life expectancy,” and analyzed 120 original articles and 213 reviews, commentaries, and features, using a text mining technique to analyze the abstracts of all of the articles. IBM SPSS Text Analytics for Surveys 4.0 were used to analyze the data. All of the abstracts were read, and the researchers determined whether the extracted literature was appropriate.

A “frequency basis” was employed to create the categories, with a lower limit of ten times for the frequency of occurrence. All of the abstracts were inspected in order to organize the categories, even those with a low frequency of occurrence.

The extracted categories and number of records were, in descending order were health statistics (73), gender (61), age (34), research district (25), long-term care insurance (22), disease (20), exercise (9), and lifestyle (5). Table 1 shows the content included in each category.

The health statistics category included content on disability-adjusted life expectancy (DALE), weighted disability prevalence (WDP), and other indicators of healthy life expectancy, such as population and, mean lifespan. Gender included men, women, gender differences, and men and women. Age included late-stage elderly, 85- years-old, 65 years of age or older, and 65 years old. Research district included region, nation, and prefectural and city governments, and revealed that there are some research investigations that have used national data associated with regional initiatives. Long-term care insurance included requiring long-term care and care prevention. Disease included state of requiring long-term care, osteoporosis, impairment, and cerebrovascular disease.

An association between the categories of exercise and lifestyle was observed using the software’s “Category web” function. This function makes it possible to visualize the collocations among the categories. Exercise was strongly associated with Indicators of healthy life expectancy in Health statistics (Figure 1). Lifestyle was strongly associated with men, gender differences, and men and women in gender; district and prefecture and city governments in research area and; indicators of healthy life expectancy in health statistics (Figure 2).

The extracted categories and number of records were, in descending order, orientation (134), disease (111), living (70), health statistics (57), QOL (55), exercise (40), care (31), gender (27), medical care (26), research area (19), musculoskeletal (18), falls (16), fitness (15), evidence (14), concept (12), and intervention 11. Table 2 shows the content that was included in each category.

Orientation included independence, maintenance, prevention, development, inhibition, lengthening, and extension. Disease included lifestyle-related disease, menopause, diagnosis, bedridden, impairment, pathology, and treatment. Health statistics included lifespan, longevity, life prognosis, and indicators. QOL included QOL and quality; care included care prevention, requiring long-term care, and care; gender included men and women; and research district included world and nation.

For evidence, intervention, falling, fitness, and medical care, the links between categories were observed using the software’s “Category web” function. The results indicated that evidence was strongly associated with life prognosis in health statistics, prevention in orientation, and lifestyle in living. Intervention was most apparently linked to lifestyle in living and to inhibition in orientation (Figure 3). Falls were most strongly linked to prevention in orientation and to musculoskeletal (Figure 4). Fitness was clearly associated with extension in orientation and indicators in health statistics. Medical care was strongly linked to inhibition, prevention, and extension in orientation; lifestyle-related disease in disease; and to care, exercise, living, QOL, and evidence.

The content reported in the original articles often related to health statistics, and there also was a substantial amount of content relating to gender, because men and women have different average life spans. In terms of the age groups, the early-stage elderly and the late-stage elderly were found to have been participants in research studies.

Osteoporosis and cerebrovascular disease received attention. On September 5, 2013, the Ministry of Health, Labour and Welfare reported that the leading cause of death by age group in Japan was malignant neoplasm for

people ages 65 and older, followed by heart disease in second place, pneumonia in third, and cerebrovascular disease in fourth place. The search that was conducted did not extract malignant neoplasm, heart disease, and pneumonia, which were the leading causes of death at the time of the search. This is thought to occur because treatment is a major component of malignant neoplasm, heart disease, and pneumonia, and therefore they were less likely to be extracted by the techniques used in this study. Cerebrovascular disease and osteoporosis, however, are more likely to lead to being bedridden and to having a decrease in QOL, and this is presumably the reason for the substantial amount of attention to research on “active aging” and “healthy life expectancy.” In the case of malignant neoplasms, it has been reported from the fit of the graph of cause-specific mortality to the Gompertz curve, that primary prevention of lifestyle-related diseases is important [8] . This is an issue that should be addressed by future research.

The fact that the “Category web” function showed a strong association between exercise and indicators of healthy life expectancy in health statistics demonstrates that the research on exercise is associated with indicators of healthy life expectancy. In the Japana Centra Revuo Medicina database, the search using “training and healthy life expectancy” extracted three original articles and 26 reviews, commentaries, and features. The “National Survey on Active Aging 2009” [9] , performed by the Japan Health Promotion & Fitness Foundation, reported “time to enjoy exercise and sports as a realistic goal to increase the amount of time spent in these activities that was most desired by the middle-aged and elderly (21.6%). This finding showed that about one in five middle-aged and elderly adults put into practice, to the goals they thought they would achieve, during the one- year period. We feel that in the future, research on exercise/training and healthy life expectancy is an important topic that needs to be addressed.

The extracted category of disease included lifestyle-related diseases, menopause, diagnosis, bedridden, impairment, pathology, and treatment, and did not generate osteoporosis and cerebrovascular disease unlike the original articles. The diseases referred to as lifestyle-related diseases and menopause were extracted. The categories of Orientation, QOL, falls, fitness, evidence, concept, and intervention, which were not generated in search of the original articles, were generated here. The differences in categories from those of the original articles may be attributed to the possibility that Japanese researchers are not publishing their research studies as original articles domestically but rather are publishing them abroad. Another possibility is that Japanese researchers are selecting, and submitting studies on research topics that are almost guaranteed to be published both at home and abroad. In the links between the categories in the “Category web” function, intervention was most strongly linked to lifestyle-related disease in living, and falls was most strongly linked to prevention in orientation as well as exercise; the focus of the research is inferred to be oriented toward intervention in lifestyle and fall prevention and exercise. This finding seems to be a research topic that needs further attention in future studies, as well.