Method and apparatus for context-adaptive personalized psychological state sampling for wearable device

11439,334

16/418012

May 21, 2019

Disclosed are a method and apparatus for sampling a context-adaptive personalized psychological state for a wearable device. The apparatus may include a sensor data collection unit configured to collect sensor data using a mobile and a wearable device, a physical activity classification unit configured to deduce a user's physical activities based on the collected sensor data, a sensor data unsupervised learning unit configured to group data based on extracted sensor data feature values for each deduced physical activity, a hierarchical psychological state classification unit configured to deduce a current psychological state as corresponding sensor data based on the data classified by the sensor data unsupervised learning unit, an information collection request determination unit configured to determine whether to request self-report information collection from a user based on at least one of the deduced current psychological state, collected label information, the sensor data or a degree of a user burden, and an information collection interface unit configured to receive self-report information from the user when the information collection request determination unit determines to collect the self-report information from the user.

Korea Advanced Institute of Science and Technology (Daejeon, KR)

Korea Advanced Institute of Science and Technology (Daejeon, KR)

1. An apparatus for context-adaptive personalized sampling of a psychological state for a wearable device, the apparatus comprising: a sensor data collection unit configured to collect sensor data using a mobile and a wearable device; a physical activity classification unit configured to deduce a user's physical activities based on the collected sensor data; a sensor data unsupervised learning unit configured to group data based on sensor data feature values extracted for each deduced physical activity; a hierarchical psychological state classification unit configured to deduce a current psychological state as corresponding sensor data based on the data classified by the sensor data unsupervised learning unit; an information collection request determination unit configured to determine whether to request self-report information collection from a user based on at least one of the deduced current psychological state, collected label information, the sensor data or a degree of a user burden; and an information collection interface unit configured to receive self-report information from the user when the information collection request determination unit determines to collect the self-report information from the user, wherein the hierarchical psychological state classification unit is configured to: perform hierarchical psychological state model training for assigning a label to each group trained by the sensor data unsupervised learning unit using a label self-reported by the user from the information collection interface unit, and perform classification on multiple labels when a plurality of labels is assigned to one group.

2. The apparatus of claim 1 , wherein: the information collection request determination unit is configured to determine a value of a self-report information request to be presented to the user based on at least one of the deduced current psychological state, the collected label information, the sensor data, the degree of a user burden, and the determination of the value of the self-report information request comprises at least one of a number of corresponding labels collected for a pre-defined time, a degree of model uncertainty improved, a stability of label classification results using sensor data collected for a pre-defined time or a degree of a user burden at a self-report time.

3. The apparatus of claim 1 , wherein the information collection interface unit is configured to: present the information collection interface unit to the user so that the user self-reports his or her own psychological state according to a pre-defined self-report format, and request the user to perform labeling on a pre-defined past period in current time or not to request self-report information collection for past data from the user when a measured change of a label is a predetermined threshold or more.

4. An apparatus for context-adaptive personalized sampling of a psychological state for a wearable device, the apparatus comprising: a sensor data collection unit configured to collect sensor data using a mobile and a wearable device; a physical activity classification unit configured to deduce a user's physical activities based on the collected sensor data; a sensor data unsupervised learning unit configured to group data based on sensor data feature values extracted for each deduced physical activity; a hierarchical psychological state classification unit configured to deduce a current psychological state as corresponding sensor data based on the data classified by the sensor data unsupervised learning unit; an information collection request determination unit configured to determine whether request self-report information collection from a user based on at least one of the deduced current psychological state, collected label information, the sensor data or a degree of a user burden; and an information collection interface unit configured to receive self-report information from the user when the information collection request determination unit determines to collect the self-report information from the user, wherein the hierarchical psychological state classification unit is configured to: select a group closest to corresponding sensor data in order to deduce the current psychological state using the sensor data, and perform hierarchical psychological state model classification for deducing the current psychological state based on a distribution of labels assigned to the selected group.

5. A method of sampling a context-adaptive personalized psychological state for a wearable device, the method comprising steps of: collecting sensor data using a mobile and a wearable device through a sensor data collection unit; deducing a user's physical activities based on the collected sensor data through a physical activity classification unit; grouping data based on a sensor data feature value extracted for each deduced physical activity through a sensor data unsupervised learning unit; deducing a current psychological state as corresponding sensor data based on the classified data through a hierarchical psychological state classification unit; determining whether to request information collection from the user based on at least one of the deduced current psychological state, collected label information, the sensor data or a degree of a user burden through an information collection request determination unit; and receiving self-report information from the user through the information collection interface unit when the information collection request determination unit determines to collect the self-report information from the user, wherein the step of deducing the current psychological state as the corresponding sensor data based on the classified data through the hierarchical psychological state classification unit comprises: performing hierarchical psychological state model training for assigning a label to each group trained by the sensor data unsupervised learning unit using a label self-reported by the user from the information collection interface unit, and performing classification on multiple labels when a plurality of labels is assigned to one group.

6. The method of claim 5 , wherein: the step of determining whether to request the information collection from the user based on at least one of the deduced current psychological state, the collected label information, the sensor data or the degree of a user burden through the information collection request determination unit comprises determining a value of a self-report information request to be presented to the user based on at least one of the deduced current psychological state, the collected label information, the sensor data, the degree of a user burden, and the determination of the value of the self-report information request comprises at least one of a number of corresponding labels collected for a pre-defined time, a degree of model uncertainty improved, a stability of label classification results using sensor data collected for a pre-defined time or a degree of a user burden at a self-report time.

7. The method of claim 5 , wherein the step of receiving the self-report information from the user through the information collection interface unit when the information collection request determination unit determines to collect the self-report information from the user comprises: presenting the information collection interface unit to the user so that the user self-reports his or her own psychological state according to a pre-defined self-report format, and requesting the user to perform labeling on a pre-defined past period in a current time or not requesting self-report information collection for past data from the user when a measured change of a label is a predetermined threshold or more.

8. A method of sampling a context-adaptive personalized psychological state for a wearable device, the method comprising steps of: collecting sensor data using a mobile and a wearable device through a sensor data collection unit; deducing a user's physical activities based on the collected sensor data through a physical activity classification unit; grouping data based on a sensor data feature value extracted for each deduced physical activity through a sensor data unsupervised learning unit; deducing a current psychological state as corresponding sensor data based on the classified data through a hierarchical psychological state classification unit; determining whether to request information collection from the user based on at least one of the deduced current psychological state, collected label information, the sensor data or a degree of a user burden through an information collection request determination unit; and receiving self-report information from the user through the information collection interface unit when the information collection request determination unit determines to collect the self-report information from the user, wherein the step of deducing the current psychological state as the corresponding sensor data based on the classified data through the hierarchical psychological state classification unit comprises: selecting a group closest to corresponding sensor data in order to deduce the current psychological state using the sensor data, and performing hierarchical psychological state model classification for deducing the current psychological state based on a distribution of labels assigned to the selected group.

11284834 March 2022 Milbert
2017/0172493 June 2017 Rahman
2020/0151641 May 2020 Avegliano
2022/0051773 February 2022 Appelbaum
20160097841 August 2016

English Abstract of KR20160097841A, Aug. 18, 2016. cited by applicant

edspgr.11439334