CN112016619A - Fall detection method based on insoles - Google Patents
Fall detection method based on insoles Download PDFInfo
- Publication number
- CN112016619A CN112016619A CN202010884915.9A CN202010884915A CN112016619A CN 112016619 A CN112016619 A CN 112016619A CN 202010884915 A CN202010884915 A CN 202010884915A CN 112016619 A CN112016619 A CN 112016619A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- lstm
- falling
- insole
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 238000003062 neural network model Methods 0.000 claims abstract description 18
- 230000001133 acceleration Effects 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000013528 artificial neural network Methods 0.000 claims abstract description 9
- 238000010606 normalization Methods 0.000 claims abstract description 8
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 3
- 238000000034 method Methods 0.000 claims description 15
- 238000003491 array Methods 0.000 claims description 6
- 230000007613 environmental effect Effects 0.000 abstract 1
- 208000014674 injury Diseases 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 208000027418 Wounds and injury Diseases 0.000 description 4
- 230000006378 damage Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000032683 aging Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 206010008111 Cerebral haemorrhage Diseases 0.000 description 1
- 201000001429 Intracranial Thrombosis Diseases 0.000 description 1
- 206010071368 Psychological trauma Diseases 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000008733 trauma Effects 0.000 description 1
- 230000009978 visual deterioration Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- A—HUMAN NECESSITIES
- A43—FOOTWEAR
- A43B—CHARACTERISTIC FEATURES OF FOOTWEAR; PARTS OF FOOTWEAR
- A43B17/00—Insoles for insertion, e.g. footbeds or inlays, for attachment to the shoe after the upper has been joined
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1116—Determining posture transitions
- A61B5/1117—Fall detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Dentistry (AREA)
- Pathology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Evolutionary Biology (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a fall detection method based on insoles, which comprises two insoles, wherein each insole is internally provided with a pressure sensor, a three-axis acceleration sensor and a three-axis gyro sensor, and the fall detection method comprises the following steps: acquiring pressure, acceleration and angular speed data of a user in a falling state by utilizing the insole to obtain a falling data set; carrying out denoising and normalization processing on data in the falling data set, analyzing the data through PCA dimension reduction, and dividing the obvious data into an input form of a neural network; constructing an LSTM-FCN neural network model; inputting the training set into an LSTM-FCN neural network model for training; and inputting the test set into the trained LSTM-FCN neural network model to obtain a detection result. The environmental limitation is avoided, the user can be monitored at any time and any place, and the detection accuracy is high; whether the patient falls down can be detected in time, and the delay of treatment time caused by the fact that the patient does not find the patient in time after falling down is avoided.
Description
Technical Field
The invention belongs to the technical field of methods for monitoring human body states, and relates to a falling detection method based on insoles.
Background
China is facing to a severe aging problem of population at present, and along with rapid development and transition of society, children and women are difficult to live near parents, so that a lot of solitary old people and empty-nest old people appear. Elderly, teenagers and children are high risk groups of injuries, and among all causes of injuries to elderly, falls are "the chief culprit", and reports show that nearly half of women and one quarter of men have fallen among elderly residents aged 65 years and that the probability of falling is higher the older. With the aging of the old, the physiological structure of the old is changed in a series, the body function is reduced, the strain capacity is reduced, the resistance of the old is reduced, and diseases such as heart disease, cerebral hemorrhage, thrombus and the like are easy to occur. In addition, the visual deterioration hinders the elderly from exploring the surrounding environment and identifying surrounding obstacles, and certain drugs taken by the elderly reduce the mental sensitivity of the elderly, which greatly increases the probability and frequency of accidental falls of the elderly. The falling of the old can not only cause the physical trauma of the old, but also greatly cause a series of adverse effects on the psychology of the old, so that the old loses the confidence of independent life. However, the fact that a fall occurs is not itself the root cause of serious injury and disease consequences, since the elderly are not rescued in a timely manner after the fall occurs, thereby delaying the time of treatment and thus increasing the rate of injury and death. And may cause sequela and psychological trauma after falling for the elderly because the best treatment opportunity is missed. It has been demonstrated that the earlier the occurrence of a fall is found to be reported, the lower the morbidity and mortality of the elderly who fall will be. Therefore, after the old people fall down, it is of great significance to find the old people as soon as possible, report the old people as soon as possible, and inform relatives or medical monitoring institutions to help the old people.
At present, there are two main methods for fall detection: one is a computer-based method for detecting the falling of a human body by installing a monitoring camera, the vision-based method has high requirements on the environment although the cost is low and the efficiency is high, and the other is a falling detection method based on embedded equipment, which detects the falling through equipment such as a smart watch and a smart phone, but the method has low accuracy and is easily classified by other human behaviors by mistake. Chinese patent (application No. CN202010266081.5, publication No. CN111461042A) discloses a method for detecting a fall of a human body by a computer vision method, but is affected by the environment, and if the user leaves the visual field range of a camera, the user cannot be monitored, and chinese patent (application No. CN201911204629.7, publication No. CN110916675A) discloses a method for detecting a fall of a head-mounted device, which detects whether the user falls by collecting GMR magnetic sensor signals, but the device is too complex and inconvenient to wear.
Disclosure of Invention
The invention aims to provide a fall detection method based on an insole, which solves the problem of low detection accuracy in the prior art.
The invention adopts the technical scheme that the falling detection method based on the insoles comprises two insoles, wherein each insole is internally provided with a pressure sensor, a three-axis acceleration sensor and a three-axis gyro sensor, and the falling detection method comprises the following steps:
step 1, acquiring pressure, acceleration and angular velocity data of a user in a falling state by utilizing an insole to obtain a falling data set, wherein the falling data set comprises a training set and a testing set;
step 2, denoising and normalizing the data in the falling data set, analyzing the data through PCA dimension reduction, selecting obvious data, and dividing the obvious data into input forms of a neural network;
step 3, constructing an LSTM-FCN neural network model;
step 4, inputting the training set into an LSTM-FCN neural network model for training, and adjusting model parameters to obtain a trained LSTM-FCN neural network model;
and 5, inputting the test set into the trained LSTM-FCN neural network model to obtain a detection result.
The invention is also characterized in that:
and 6, calculating the accuracy of the detection result, the F1-score value, and drawing an ROC curve to evaluate the model.
Each insole is provided with 8 pressure sensors, 1 triaxial acceleration sensor and 1 triaxial gyro sensor.
The step 2 specifically comprises the following steps:
step 2.1, carrying out average value calculation on the data in the falling data set, wherein the average value is calculated once every 5 times of collection;
step 2.2, performing normalization operation on the data processed in the step 2.1, converting the pressure sensor array, the triaxial acceleration sensor array and the triaxial gyro sensor array into arrays of 63 × 16, 63 × 6 and 63 × 6 respectively, and storing the arrays in the forms of 1008 × 1, 378 × 1 and 378 × 1;
step 2.3, splicing the normalized data of the pressure sensor, the triaxial acceleration sensor and the triaxial gyro sensor into txW, and using the txW as the input of the LSTM-FCN neural network; w is the number of sensors in the sensor array, and the value of W is (16+6+ 6); t is 63.
The LSTM-FCN neural network model comprises:
FCN layer, 3 convolution layers totally, use the normalization of the data after each convolution layer; the LSTM module, containing 5 cycles of LSTM units, is finally sorted by softmax.
The invention has the beneficial effects that:
according to the method for detecting the falling of the insole, the limitation of the environment is avoided, the user can be monitored at any time and any place, the data acquisition equipment is more convenient and comfortable to wear, and the detection accuracy is high; the detection mode of the LSTM-FCN neural network is adopted, a large amount of preprocessing is not needed to be carried out on data, the calculated amount is reduced, the characteristics are not needed to be extracted manually, the model fitting process is greatly simplified, and the method has high accuracy and generalization capability; when falling occurs, whether the patient is in a falling state or not can be detected in time, and the delay of treatment time caused by the fact that the patient does not find the patient in time after falling is avoided.
Drawings
Fig. 1 is a flow chart of a fall detection method based on an insole of the invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
A fall detection method based on insoles comprises two insoles, wherein each insole is internally provided with a pressure sensor, a three-axis acceleration sensor and a three-axis gyro sensor, and as shown in figure 1, the method specifically comprises the following steps:
step 1, acquiring pressure, acceleration and angular velocity data of a user in a falling state by utilizing an insole to obtain a falling data set, wherein the falling data set comprises a training set and a testing set;
step 2, denoising and normalizing the data in the falling data set, analyzing the data through PCA dimension reduction, selecting obvious data, and dividing the obvious data into input forms of a neural network; each insole is provided with 8 pressure sensors, 1 triaxial acceleration sensor and 1 triaxial gyro sensor.
Step 2.1, carrying out average value calculation on the data in the falling data set, wherein the average value is calculated once every 5 times of collection;
step 2.2, performing normalization operation on the data processed in the step 2.1, converting the pressure sensor array, the triaxial acceleration sensor array and the triaxial gyro sensor array into arrays of 63 × 16, 63 × 6 and 63 × 6 respectively, and storing the arrays in the forms of 1008 × 1, 378 × 1 and 378 × 1;
step 2.3, splicing the normalized data of the pressure sensor, the triaxial acceleration sensor and the triaxial gyro sensor into txW, and using the txW as the input of the LSTM-FCN neural network; w is the number of sensors in the sensor array, and the value of W is (16+6+ 6); t is 63.
Step 3, constructing an LSTM-FCN neural network model; the FCN layer comprises 3 convolutional layers, and each convolutional layer is subjected to data normalization processing without using any pooling layer; and the LSTM module comprises 5 circulating LSTM units, performs feature extraction according to the time sequence of the data, and finally classifies the data through softmax, and outputs whether the data fall down.
Step 4, inputting the training set into an LSTM-FCN neural network model for training, and adjusting model parameters to obtain a trained LSTM-FCN neural network model;
and 5, inputting the test set into the trained LSTM-FCN neural network model to obtain a detection result.
And 6, calculating the accuracy of the detection result, F1-score value, and drawing an ROC curve to evaluate the model.
The trained LSTM-FCN neural network model is embedded into an APP of intelligent equipment, real-time falling detection is carried out in a short time when a continuous monitoring mode is adopted, then state monitoring is carried out for a long time period, when the human body state is monitored for the long time period, if falling action occurs, an early warning program is started, if normal activity of a user is detected to continue in a certain time period, the warning is cancelled, and if the normal activity of the user continues, the warning program is started to send a short message to a guardian for help. When an alarm is given, the user is required to confirm whether the alarm is cancelled or not, if the user cancels the alarm, the alarm program is cancelled, and if the user does not answer within the specified time, the alarm program is immediately started.
Through the mode, the falling detection method based on the insole avoids the limitation of the environment, can monitor the user at any time and any place, and is more convenient and comfortable to wear the data acquisition equipment; the detection mode of the LSTM-FCN neural network is adopted, a large amount of preprocessing is not needed to be carried out on data, the calculated amount is reduced, the characteristics are not needed to be extracted manually, the model fitting process is greatly simplified, and the method has high accuracy and generalization capability; when falling occurs, whether the patient is in a falling state or not can be detected in time, and the delay of treatment time caused by the fact that the patient does not find the patient in time after falling is avoided.
Claims (5)
1. A falling detection method based on insoles comprises two insoles, wherein each insole is internally provided with a pressure sensor, a three-axis acceleration sensor and a three-axis gyro sensor, and is characterized by comprising the following steps:
step 1, acquiring pressure, acceleration and angular velocity data of a user in a falling state by utilizing an insole to obtain a falling data set, wherein the falling data set comprises a training set and a testing set;
step 2, carrying out denoising and normalization processing on the data in the falling data set, analyzing the data through PCA dimension reduction, selecting obvious data, and dividing the obvious data into an input form of a neural network;
step 3, constructing an LSTM-FCN neural network model;
step 4, inputting the training set into an LSTM-FCN neural network model for training, and adjusting model parameters to obtain a trained LSTM-FCN neural network model;
and 5, inputting the test set into the trained LSTM-FCN neural network model to obtain a detection result.
2. The method for detecting falls based on insoles of claim 1, further comprising the steps of 6, calculating the accuracy of the detection result, F1-score value, and drawing ROC curve to evaluate the model.
3. An insole-based fall detection method according to claim 1, wherein 8 pressure sensors, 1 three-axis acceleration sensor and 1 three-axis gyro sensor are provided on each insole.
4. An insole-based fall detection method as claimed in claim 3, wherein the step 2 comprises the following steps:
2.1, calculating the average value of the data in the falling data set, wherein the average value is calculated once every 5 times of collection;
step 2.2, performing normalization operation on the data processed in the step 2.1, converting the pressure sensor array, the triaxial acceleration sensor array and the triaxial gyro sensor array into arrays of 63 × 16, 63 × 6 and 63 × 6 respectively, and storing the arrays in the forms of 1008 × 1, 378 × 1 and 378 × 1;
step 2.3, splicing the normalized data of the pressure sensor, the triaxial acceleration sensor and the triaxial gyro sensor into t multiplied by W, and using the t multiplied by W as the input of the LSTM-FCN neural network; w is the number of sensors in the sensor array, and the value of W is (16+6+ 6); t is 63.
5. An insole-based fall detection method according to claim 1, wherein the LSTM-FCN neural network model comprises:
FCN layer, 3 convolution layers totally, use the normalization of the data after each convolution layer; the LSTM module, containing 5 cycles of LSTM units, is finally sorted by softmax.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010884915.9A CN112016619A (en) | 2020-08-28 | 2020-08-28 | Fall detection method based on insoles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010884915.9A CN112016619A (en) | 2020-08-28 | 2020-08-28 | Fall detection method based on insoles |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112016619A true CN112016619A (en) | 2020-12-01 |
Family
ID=73503357
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010884915.9A Pending CN112016619A (en) | 2020-08-28 | 2020-08-28 | Fall detection method based on insoles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112016619A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021212883A1 (en) * | 2020-04-20 | 2021-10-28 | 电子科技大学 | Fall detection method based on intelligent mobile terminal |
| CN114067436A (en) * | 2021-11-17 | 2022-02-18 | 山东大学 | Fall detection method and system based on wearable sensor and video monitoring |
| CN114595748A (en) * | 2022-02-21 | 2022-06-07 | 南昌大学 | Data segmentation method for fall protection system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107890166A (en) * | 2017-11-03 | 2018-04-10 | 温州鞋都网络科技有限公司 | A kind of Intelligent insole based on plantar pressure gesture recognition |
| CN110946585A (en) * | 2019-11-21 | 2020-04-03 | 上海理工大学 | Fall detection system and method based on data fusion and BP neural network |
| CN111436944A (en) * | 2020-04-20 | 2020-07-24 | 电子科技大学 | Falling detection method based on intelligent mobile terminal |
-
2020
- 2020-08-28 CN CN202010884915.9A patent/CN112016619A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107890166A (en) * | 2017-11-03 | 2018-04-10 | 温州鞋都网络科技有限公司 | A kind of Intelligent insole based on plantar pressure gesture recognition |
| CN110946585A (en) * | 2019-11-21 | 2020-04-03 | 上海理工大学 | Fall detection system and method based on data fusion and BP neural network |
| CN111436944A (en) * | 2020-04-20 | 2020-07-24 | 电子科技大学 | Falling detection method based on intelligent mobile terminal |
Non-Patent Citations (1)
| Title |
|---|
| 肖洪兵 等: "《跟我学用单片机》", 北京航空航天大学出版社, pages: 210 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021212883A1 (en) * | 2020-04-20 | 2021-10-28 | 电子科技大学 | Fall detection method based on intelligent mobile terminal |
| CN114067436A (en) * | 2021-11-17 | 2022-02-18 | 山东大学 | Fall detection method and system based on wearable sensor and video monitoring |
| CN114067436B (en) * | 2021-11-17 | 2024-03-05 | 山东大学 | Fall detection method and system based on wearable sensor and video monitoring |
| CN114595748A (en) * | 2022-02-21 | 2022-06-07 | 南昌大学 | Data segmentation method for fall protection system |
| CN114595748B (en) * | 2022-02-21 | 2024-02-13 | 南昌大学 | Data segmentation method for fall protection system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112016619A (en) | Fall detection method based on insoles | |
| CN205041401U (en) | Equipment is worn detection device and is had and wears guardianship device that detects function | |
| Di Rosa et al. | Concurrent validation of an index to estimate fall risk in community dwelling seniors through a wireless sensor insole system: A pilot study | |
| CN107106085A (en) | Apparatus and method for sleep monitor | |
| CN105125221A (en) | System and method for cloud service real-time falling-down detection | |
| CN102488512A (en) | Automatic alarm system for electrocatdiogram monitoring and alarm method thereof | |
| US20140324459A1 (en) | Automatic health monitoring alerts | |
| CN106725445B (en) | A kind of the portable body injury gained in sports monitor system and method for brain wave control | |
| CN205050303U (en) | Human paralysis of intelligence is monitoring devices | |
| US20110148630A1 (en) | Apparatus and method for sensing photoplethysmogram and fall | |
| Jatesiktat et al. | An elderly fall detection using a wrist-worn accelerometer and barometer | |
| CN109119172B (en) | Human behavior detection method based on bee colony algorithm | |
| CN107019501B (en) | Remote tumble detection method and system based on genetic algorithm and probabilistic neural network | |
| CN115862873A (en) | Method, system and device for quantifying and intervening sleep rhythm | |
| CN115662631A (en) | AI intelligence discrimination-based nursing home management system | |
| CN117133464B (en) | Intelligent monitoring system and monitoring method for health of old people | |
| CN111626273B (en) | Fall behavior recognition system and method based on atomic action time sequence characteristics | |
| CN113869179A (en) | Multi-mode feature attention fused home-based and aged-care intelligent monitoring method and system | |
| CN113080897B (en) | Sleep time assessment system and method based on physiological and environmental data analysis | |
| Ghazal et al. | An integrated caregiver-focused mHealth framework for elderly care | |
| CN113096792A (en) | Intelligent health monitoring and early warning method and system based on machine learning | |
| CN111588358A (en) | Intelligent old age care management system and application method | |
| Chen et al. | An auxiliary diagnostic system for Parkinson’s disease based on wearable sensors and genetic algorithm optimized random forest | |
| CN116115198A (en) | Low-power consumption snore automatic recording method and device based on physiological sign | |
| CN115770028A (en) | Blood pressure detection method, system, device and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201201 |