Top > Publications >  Widening World of Statistical Science > Maintenance of good health
Maintenance of good health

Maintenance of good health has always been an important social issue. In health science researches, variation from individual to individual is great and the data obtained from human subjects is complicated, but effective information can be obtained by using methods based on statistical science. Statistical science is widely employed in varieties of areas relevant to health, including prediction and prevention of diseases, improvement in the accuracy of diagnoses, determination of safety standards for potentially hazardous substances such as air pollutants and food additives, examination of drug efficacy, and the evaluation of living environments.

§1 Automatic measurement of cardiac functions and computer-aided diagnosis

Photo by Department of Clinical Functional Physiology,
Toho University School of Medicine

The photo shows the automatic measurement of human cardiac functions during exercise such as heart rate, blood pressures, expulsion time from left ventricle, and pulse rate.

The time series analysis method and multivariate analysis method are used for this automatic measuring method, and in numerical diagnoses such as the efficacy of medicines, disease determination, and cardiac function evaluation method using measured characteristics.


§2 Research into medical care and public welfare

Now that we are entering the 21st century, it is more important than ever to make medical care and public welfare satisfactory. It is said that in Japan patients remain hospitalized longer than other countries on average. We analyses the hospitalized period in co-operation with Toshihiko Hasegawa Md, Department Director, The National Institute of Health Services Management.

As a result of the analysis, we found that the mean hospitalization period of 93% of patients was 2.5 weeks, and the mean hospitalization period of the remaining 7% of patients was 21.3 weeks. This research showed for the first time that there were two types of inpatients. This example explicitly shows the usefulness of statistical models.


The figure shows the change in hospitalization periods in Japan. The data show that the number of inpatients decreased from 100% to 24.9% in five weeks. As a result, it was shown that short-term inpatients decreased from 93% to 19%, and long-term inpatients decreased from 7% to 5.8%, thus clearly showing the reason for the decrease. After 20 weeks it was shown that there were very few short-term inpatients, and almost all that remained there were long-term inpatients.


§3 Risk assessment

Photo by Mr. Makoto Hayashi,
National Institute of Health Sciences

The photo shows an abnormal mitosis observed in cells cultured in a chemical substance. For evaluating safety chemical compounds, an epidemiological survey, an animal experiment or mutagenicity test is conducted. Since the data obtained varies according to individual differences, as well as survey and experimental error, statistical analysis is indispensable.


§4 Analysis of dental diseases

The "8020 (eighty-twenty)campaign" has been proposed for dental health in Japan. This is a campaign aimed at keeping 20 teeth until the age of 80.

It is natural to some extent that the number of teeth decreases as one advances in age. But how do the number of teeth differ by generation or age?

By analyzing the data of a nationwide survey, the Survey of Dental Diseases, conducted by the Ministry of Health and Welfare, using the Bayesian cohort model developed by the Institute, the change in present (remaining) teeth by age, period(era), and cohort(generation) can be shown, and thereby past situations can be grasped and predictions on future trends can be made.

(Joint research with Nihon University School of Dentistry at Matsudo. The figure shows the results of analysis of the rate at which the lower first molars remain in place.)

top of this page