Patent Issued for Device and method for sequential posture identification and autonomic function information acquisition (USPTO 11350879): Nippon Telegraph and Telephone Corporation

2022 JUN 27 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- From Alexandria, Virginia, NewsRx journalists report that a patent by the inventors Horikawa, Keitaro (Musashino, JP), Nakamura, Yoshitaka (Musashino, JP), Sawada, Masato (Musashino, JP), Tsukada, Shingo (Atsugi, JP), Yamada, Toshiya (Yokohama, JP), Yamanaka, Akihiro (Musashino, JP), filed on February 16, 2016, was published online on June 7, 2022.

The patent’s assignee for patent number 11350879 is Nippon Telegraph and Telephone Corporation (Chiyoda-ku, Japan).

News editors obtained the following quote from the background information supplied by the inventors: “In recent years, utilization of so-called wearable devices, i.e., information processing terminals which can be attached to human bodies (hereinafter, wearable devices) has expanded. Because it can be worn and carried by a user on a daily basis, a wearable device can be utilized in continuous and long-term monitoring of health conditions and lifestyle habits of the user. Further, technology for collecting, on a large scale, information related to users’ health conditions and lifestyle habits collected by wearable devices is being popularized. Examples of such information on user’s health conditions and lifestyle habits collectable by utilization of a wearable device include times of the user’s sleep, exercise, work, commuting, meals, and the like.

“Along with aging of society, countermeasures for increase of chronic diseases, such as high blood pressure and diabetes, and dealing promptly with heart attacks and cerebrovascular diseases (apoplexy) have become social challenges. Since these diseases are closely related to lifestyle habits, such as exercise and diet, there is a demand for prevention of these diseases and prevention of increase of these diseases by appropriate health management according to individual lifestyle habits.

“Action estimation technology may be utilized for lifestyle habits, action patterns, and the like of an individual to be perceived. For example, according to an action estimation technique described in Non-Patent Literature 1, a single triaxial acceleration sensor is attached to a subject and acceleration data are acquired. Based on the acquired acceleration data, types of activity of the subject are estimated.”

As a supplement to the background information on this patent, NewsRx correspondents also obtained the inventors’ summary information for this patent: “Technical Problem

“However, when action estimation for a subject is performed by the conventional technique, in-depth estimation of posture and states of activity is difficult. For example, when an acceleration sensor is attached to the arm, movement of the torso and movement of the arm are not necessarily the same, and thus the posture of the subject may not be detected accurately. Further, just performing action estimation of a subject by use of an acceleration sensor may not directly lead to health management and disease prevention for the subject.

“Embodiments disclosed herein have been made in view of the above, and aim to provide a technique for detecting a posture and a state of activity of a subject and enabling health management and disease prevention.

“Solution to Problem

“Sequential posture identification device, method, and program disclosed herein receive respectively, from an acceleration information measurement unit and a biological signal information measurement unit that are provided in a wearable device, acceleration information on motion of a subject whom the wearable device is attached to, and biological signal information of the subject. Further, the sequential posture identification device, method, and program extract, from the acceleration information and the biological signal information, first feature data corresponding to a first predetermined period and second feature data corresponding to a second predetermined period. Further, the sequential posture identification device, method, and program generate, by machine learning based on the first feature data, a dynamic/static activity identification model for identification of whether the subject is involved in a dynamic activity or in a static activity. The sequential posture identification device, method, and program generate, by machine learning based on the first feature data, a dynamic-activity identification model for identification of plural dynamic-activity patterns. The sequential posture identification device, method, and program generate, by machine learning based on the first feature data, a static-activity identification model for identification of plural static-activity patterns. The sequential posture identification device, method, and program determine, based on the dynamic/static activity identification model and the second feature data, whether the subject is involved in a dynamic activity or in a static activity in the second predetermined period. The sequential posture identification device, method, and program determine, based on the dynamic-activity identification model and the second feature data, a dynamic-activity pattern of the subject in the second predetermined period. The sequential posture identification device, method, and program determine, based on the static-activity identification model and the second feature data, a static-activity pattern of the subject in the second predetermined period. The sequential posture identification device, method, and program identify a posture and an activity of the subject in the second predetermined period, by combining results of the determination by first, second, and third determination units. The sequential posture identification device, method, and program generate correspondence information associating between the posture and the activity identified by an identification unit and biological signal information of the subject in the second predetermined period.”

The claims supplied by the inventors are:

“1. A sequential posture identification device, comprising: a wearable device that is configured to be attached to a trunk of a subject; an acceleration sensor that is provided in the wearable device, and measures acceleration information of motion of the subject whom the wearable device is attached to; a biological signal information sensing device that is provided in the wearable device, and measures biological signal information of the subject; and processing circuitry configured to extract, from the acceleration information and the biological signal information, first feature data corresponding to a first predetermined period and second feature data corresponding to a second predetermined period; perform a first generation process that generates, by machine learning based on the first feature data, a dynamic/static activity identification model for identification of whether the subject is involved in a dynamic activity or in a static activity; perform a second generation process that generates, by machine learning based on the first feature data, a dynamic-activity identification model for identification of plural dynamic-activity patterns; perform a third generation process that generates, by machine learning based on the first feature data, a static-activity identification model for identification of plural static-activity patterns; perform a first determination process that determines, based on the dynamic/static activity identification model and the second feature data, whether the subject is involved in a dynamic activity or in a static activity in the second predetermined period; perform a second determination process that determines, based on the dynamic-activity identification model and the second feature data, a dynamic-activity pattern of the subject in the second predetermined period; perform a third determination process that determines, based on the static-activity identification model and the second feature data, a static-activity pattern of the subject in the second predetermined period; identify a posture and an activity of the subject in the second predetermined period, by combining together results of the determination by the first, second, and third determination units; and generate correspondence information associating between the posture and activity identified by the identification unit and biological signal information of the subject in the second predetermined period.

“2. The sequential posture identification device according to claim 1, wherein the processing circuitry extracts, from the first predetermined period, plural periods overlapping one another and having different lengths, and extracts the first feature data for each of the extracted periods.

“3. The sequential posture identification device according to claim 2, wherein the processing circuitry extracts the first and second feature data, based on: at least one of a maximum value, a minimum value, an average value, and a variance value, in a set of time series of acceleration information in a first period of the plural periods overlapping one another and having different lengths; and at least one of a maximum value, a minimum value, an average value, and a variance value, of heartbeat intervals in a second period of the plural periods overlapping one another and having different lengths.

“4. The sequential posture identification device according to claim 3, wherein the processing circuitry extracts the first and second feature data, based further on at least one of: a vibration frequency along each axis of acceleration measured in the second period; an average value of vibration frequencies; and a variance value of the vibration frequencies.

“5. The sequential posture identification device according to claim 3, wherein the second period is longer than the first period.

“6. The sequential posture identification device according to claim 1, wherein processing by the first, second, and third generation processes is executed in parallel with processing by the first, second, and third determination processes.

“7. The sequential posture identification device according to claim 1, wherein the static-activity patterns are at least a standing position, a sitting position, and a recumbent position.

“8. The sequential posture identification device according to claim 1, wherein the dynamic-activity patterns are at least ambulating, jumping, and stepping in place.

“9. A sequential posture identification method, including: a reception step of receiving respectively, from an acceleration information measurement unit and a biological signal information measurement unit that are provided in a wearable device that is configured to be attached to a trunk of a subject, acceleration information on motion of the subject whom the wearable device is attached to, and biological signal information of the subject; a feature data extraction step of extracting, from the acceleration information and the biological signal information, first feature data corresponding to a first predetermined period and second feature data corresponding to a second predetermined period; a first generation step of generating, by machine learning based on the first feature data, a dynamic/static activity identification model for identification of whether the subject is involved in a dynamic activity or in a static activity; a second generation step of generating, by machine learning based on the first feature data, a dynamic-activity identification model for identification of plural dynamic-activity patterns; a third generation step of generating, by machine learning based on the first feature data, a static-activity identification model for identification of plural static-activity patterns; a first determination step of determining, based on the dynamic/static activity identification model and the second feature data, whether the subject is involved in a dynamic activity or in a static activity in the second predetermined period; a second determination step of determining, based on the dynamic-activity identification model and the second feature data, a dynamic-activity pattern of the subject in the second predetermined period; a third determination step of determining, based on the static-activity identification model and the second feature data, a static-activity pattern of the subject in the second predetermined period; an identification step of identifying a posture and an activity of the subject in the second predetermined period, by combining together results of the determination in the first, second, and third determination steps; and a generating step of generating correspondence information associating between the posture and activity identified in the identification step and biological signal information of the subject in the second predetermined period.

“10. A non-transitory computer-readable recording medium having stored therein a sequential posture identification program that causes a computer to execute: a reception process of receiving respectively, from an acceleration information measurement unit and a biological signal information measurement unit that are provided in a wearable device that is configured to be attached to a trunk of a subject, acceleration information on motion of the subject whom the wearable device is attached to, and biological signal information of the subject; a feature data extraction process of extracting, from the acceleration information and the biological signal information, first feature data corresponding to a first predetermined period and second feature data corresponding to a second predetermined period; a first generation process of generating, by machine learning based on the first feature data, a dynamic/static activity identification model for identification of whether the subject is involved in a dynamic activity or in a static activity; a second generation process of generating, by machine learning based on the first feature data, a dynamic-activity identification model for identification of plural dynamic-activity patterns; a third generation process of generating, by machine learning based on the first feature data, a static-activity identification model for identification of plural static-activity patterns; a first determination process of determining, based on the dynamic/static activity identification model and the second feature data, whether the subject is involved in a dynamic activity or in a static activity in the second predetermined period; a second determination process of determining, based on the dynamic-activity identification model and the second feature data, a dynamic-activity pattern of the subject in the second predetermined period; a third determination process of determining, based on the static-activity identification model and the second feature data, a static-activity pattern of the subject in the second predetermined period; an identification process of identifying a posture and an activity of the subject in the second predetermined period, by combining together results of the determination in the first, second, and third determination processes; and a generating process of generating correspondence information associating between the posture and activity identified in the identification process and biological signal information of the subject in the second predetermined period.”

For additional information on this patent, see: Horikawa, Keitaro. Device and method for sequential posture identification and autonomic function information acquisition. U.S. Patent Number 11350879, filed February 16, 2016, and published online on June 7, 2022. Patent URL: http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=11350879.PN.&OS=PN/11350879RS=PN/11350879

(Our reports deliver fact-based news of research and discoveries from around the world.)