02.04.2013 Views

Special Issue on Information Explosion

Special Issue on Information Explosion

Special Issue on Information Explosion

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

212 M. Kitsuregawa and T. Nishida<br />

must be made in the daily life of patients out of hospitals. We c<strong>on</strong>ducted two<br />

verificati<strong>on</strong> studies: <strong>on</strong>e for metabolic syndrome, which was shown as the main<br />

cause of diabetes in recent studies as obesity with visceral fat accumulati<strong>on</strong>, and<br />

another for diabetes.<br />

The first study aimed at managing metabolic syndromes. We asked <strong>on</strong>e<br />

hundred m<strong>on</strong>itors to wear activity sensors and pulse rate sensors, and to use<br />

wireless weight scales and blood pressure sensors for 11 days. On the 6th day,<br />

health instructors suggested lifestyle problems, up<strong>on</strong> the sensor data, to the m<strong>on</strong>itors.<br />

The sensed data were automatically uploaded to the server via Bluetooth<br />

or wireless LAN. The result showed that 91.6 % of the m<strong>on</strong>itors increased their<br />

amount of exercise (9.55 % increase in average (n = 83)) after the healthcare<br />

instructi<strong>on</strong>. In the analysis of the kinds of activities, “standing” and “walking”<br />

were increased, and “using elevator” was apparently decreased in daily behaviors.<br />

Additi<strong>on</strong>ally, we could estimate the optimizati<strong>on</strong> of health instructi<strong>on</strong>s<br />

in both time and accuracy, which should lead to enormous cost reducti<strong>on</strong> of<br />

Particular Health Checkup System held all over Japan.<br />

The sec<strong>on</strong>d study targeted <strong>on</strong> diabetics. We used activity sensors, blood<br />

sugar sensors, weight scales, and blood pressure sensors. The former two were<br />

mobile, and all the data were uploaded via mobile ph<strong>on</strong>es or wireless LAN. The<br />

mobile ph<strong>on</strong>e had an applicati<strong>on</strong> to show the recommendati<strong>on</strong> message either<br />

triggered by the sensor data or sent daily. As the result, we found a critical<br />

case of reverse white coat hypertensi<strong>on</strong>, where the blood pressure is high <strong>on</strong>ly at<br />

out of hospital. Moreover, according to the analysis of recommendati<strong>on</strong> effect,<br />

the frequency of measuring vital informati<strong>on</strong> increased after a recommendati<strong>on</strong>,<br />

and several messages certainly increased the amount of activities, and several<br />

decreased. These results have not been found in our studies so far without<br />

such an envir<strong>on</strong>ment. In additi<strong>on</strong>, we developed a collecting system of activity<br />

informati<strong>on</strong> using smart ph<strong>on</strong>es, and gathered well-labeled 3D accelerometer<br />

data of 150 people. Since smart ph<strong>on</strong>es have been deployed tens of milli<strong>on</strong>s in<br />

the world, they can be a global platform for progress of activity recogniti<strong>on</strong>.<br />

Papers in This <str<strong>on</strong>g>Special</str<strong>on</strong>g> <str<strong>on</strong>g>Issue</str<strong>on</strong>g><br />

This special issue c<strong>on</strong>tains five papers. The first paper 9) addresses automated<br />

c<strong>on</strong>structi<strong>on</strong> of a semantic network by making use of the huge informati<strong>on</strong><br />

resource of the Web. The challenge of this work is to go bey<strong>on</strong>d the c<strong>on</strong>venti<strong>on</strong>al<br />

keyword-based Web search, to enable the user to find informati<strong>on</strong> that <strong>on</strong>e might<br />

not even know. The main c<strong>on</strong>tributi<strong>on</strong> of this work is the development of the<br />

collective body of knowledge acquisiti<strong>on</strong> methods for automatically generating<br />

TORISHIKI-KAI, a machinery that maps out the c<strong>on</strong>text of use or handling of<br />

the topic the user inputs, and classifies topically related search terms according<br />

to semantic categories.<br />

The sec<strong>on</strong>d paper 10) presents a model-based approach to fault localizati<strong>on</strong><br />

that allows the human analyst to narrow down the fault localizati<strong>on</strong> in largescale<br />

computing envir<strong>on</strong>ments. Localizati<strong>on</strong> is regarded as anomaly detecti<strong>on</strong><br />

problems in system behaviors by automatically analyzing data resulting from

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!