State extrapolation device, state extrapolation program, and state extrapolation method

Patent metadata

Field	Value
Publication number	US-11627918-B2
Application number	US-202017004809-A
Country	US
Kind code	B2
Filing date	Aug 27, 2020
Priority date	Aug 27, 2019
Publication date	Apr 18, 2023
Grant date	Apr 18, 2023

Abstract

An object of the present invention is to more appropriately extrapolate a sign of a state that may affect a movement of a user. A state extrapolation device according to the present invention includes a vital sign acquisition unit that acquires a vital sign of a user, and a sign detection unit that uses a learned model that has learned, as training data, sign data about the vital sign related to a predetermined physical condition abnormality, and detects a sign by determining whether or not the vital sign of the user corresponds to a sign of the predetermined physical condition abnormality.

First claim

What is claimed is:

1. A state extrapolation device comprising:
a vital sign acquisition unit that acquires a vital sign of a user;
a sign detection unit that uses a learned model that has learned, as training data, sign data about the vital sign related to a predetermined physical condition abnormality, and detects a sign by determining whether or not the vital sign of the user corresponds to a sign of the predetermined physical condition abnormality;
a training data generation unit that generates the training data as the training data including a vital sign of the user;
a sign model generation unit that causes the learned model to learn the training data generated by the training data generation unit; and
a state detection unit that detects an occurrence state of the predetermined physical condition abnormality of the user, wherein
the training data generation unit generates, when the state detection unit detects the occurrence state, the training data by using first information being the vital sign of the user provided with a label in a predetermined period before the occurrence state, and second information being the vital sign of the user in the predetermined period when the occurrence state is not detected.

2. The state extrapolation device according to claim 1 , wherein
the predetermined physical condition abnormality is a physical condition abnormality that affects an exercise movement of the user.

3. The state extrapolation device according to claim 2 , wherein
the physical condition abnormality that affects the exercise movement of the user is a state where the user feels sleepiness or a state where the user has an epileptic seizure.

4. The state extrapolation device according to claim 1 , wherein
the vital sign acquisition unit acquires information about a pulse wave as the vital sign.

5. The state extrapolation device according to claim 1 , wherein
the training data generation unit generates the training data so as to set the first information and the second information at a predetermined ratio.

6. The state extrapolation device according to claim 1 , wherein
the training data generation unit specifies the predetermined period according to the physical condition abnormality.

7. The state extrapolation device according to claim 1 , wherein,
when the training data generation unit receives an input indicating a disagreement with a sign detected by the sign detection unit, in a case where the state detection unit does not subsequently detect the occurrence state within a predetermined period, the training data generation unit generates training data in which a vital sign having the sign detected is set to second information.

8. The state extrapolation device according to claim 1 , further comprising
a notification unit that notifies an abnormality by a predetermined method, wherein
the vital sign acquisition unit causes the notification unit to make a notification when the acquired vital sign deviates from a predetermined range, and deletes the vital sign.

9. The state extrapolation device according to claim 1 , wherein
the sign detection unit detects a sign of a plurality of the predetermined physical condition abnormalities by using a plurality of learned models that each have learned the plurality of predetermined physical condition abnormalities.

10. A product comprising programs stored in a non-transitory computer readable medium causing a computer to execute a state extrapolation procedure, the program causing the computer to function as a control means, and
causing the control means to perform:
a vital sign acquisition step of acquiring a vital sign of a user;
a sign detection step of using a learned model that has learned, as training data, sign data about the vital sign related to a predetermined physical condition abnormality, and detecting a sign by determining whether or not the vital sign of the user corresponds to a sign of the predetermined physical condition abnormality;

a training data generation step that generates the training data as the training data including a vital sign of the user;
a sign model generation step that causes the learned model to learn the training data generated by the training data generation unit; and
a state detection step that detects an occurrence state of the predetermined physical condition abnormality of the user, wherein
the training data generation step generates, when the state detection step detects the occurrence state, the training data by using first information being the vital sign of the user provided with a label in a predetermined period before the occurrence state, and second information being the vital sign of the user in the predetermined period when the occurrence state is not detected.

11. A state extrapolation method causing a computer to execute a state extrapolation procedure, the computer including a control means, the state extrapolation method comprising:
by the control means,
a vital sign acquisition step of acquiring a vital sign of a user;
a sign detection step of using a learned model that has learned, as training data, sign data about the vital sign related to a predetermined physical condition abnormality, and detecting a sign by determining whether or not the vital sign of the user corresponds to a sign of the predetermined physical condition abnormality;
a training data generation step that generates the training data as the training data including a vital sign of the user;
a sign model generation step that causes the learned model to learn the training data generated by the training data generation unit; and
a state detection step that detects an occurrence state of the predetermined physical condition abnormality of the user, wherein
the training data generation step generates, when the state detection step detects the occurrence state, the training data by using first information being the vital sign of the user provided with a label in a predetermined period before the occurrence state, and second information being the vital sign of the user in the predetermined period when the occurrence state is not detected.

Assignees

• Clarion Co Ltd

Inventors

• Nakamura Tomohiro

Classifications

• A61B5/742

  using visual displays (displays for heart-related electrical signals, e.g. ECG, A61B5/339) · CPC title

• A61B5/4806

  Sleep evaluation (A61B5/4821 takes precedence; devices for inducing sleep A61M21/02) · CPC title

• A61B5/4094

  Diagnosing or monitoring seizure diseases, e.g. epilepsy · CPC title

• G16H50/30

  for calculating health indices; for individual health risk assessment · CPC title

• B60W2540/221

  Physiology, e.g. weight, heartbeat, health or special needs · CPC title