WO2020236091A2

WO2020236091A2 - Method for detecting falls by using relative barometric pressure signals

Info

Publication number: WO2020236091A2
Application number: PCT/TH2020/000031
Authority: WO
Inventors: Surapa Thiemjarus; Ping Lai Benny LO; Yingnan SUN; Po Wen LO
Original assignee: National Science And Technology Development Agency
Priority date: 2019-05-17
Filing date: 2020-05-15
Publication date: 2020-11-26
Also published as: WO2020236091A3

Abstract

The present invention relates to the method for detecting falls by using relative barometric pressure signals, wherein comprising: the step of receiving signal, the step of processing, and the step of displaying, wherein the step of receiving signal comprises the signals from at least two barometric pressure sensors or the barometers attached to the positions at different or same height, wherein at least one barometric pressure sensor is attached on any position on the user's body and another barometric pressure sensor is attached on the reference position to be used as the reference at another position on the body at different height or any position in surroundings. When there is the postural transition or change in prefall and postfall activities, there is the change in the difference between the barometric pressure value from the barometric pressure sensor at any position to the barometric pressure value from the barometric pressure sensor at the reference position at least one pair when comparing values obtained from two sensors prior to the postural transition. For the step of processing, using the change in the barometric pressure values to indicate the change in the height of the device attached to the position for the detecting falls, activities and/or prefall activities, wherein the value obtained from the reference barometric pressure sensor causes the calculation of difference in the barometric pressure values to indicate the motion of the body more accurately as it could confute the change in the barometric pressure values altered based on time, place, and surroundings, wherein the step of processing includes the step of signal inference for the conversion of signals obtained from the step of receiving signals to the information for monitoring system for patients and elders.

Description

METHOD FOR DETECTING FALLS BY USING RELATIVE BAROMETRIC PRESSURE SIGNALS

Field of the invention

Engineering relates to the method for detecting falls by using relative barometric pressure signals.

Background of the invention

Falling is a serious accident leading to death for elders. Examples of possible harms include head injury, abrasion, bone fracture, and joint dislocation which affect daily life of elders. For inpatients, falling might occur in postoperative patients when they try to stand up by themselves as they assume that they can walk by themselves or in patients having drug-induced vertigo or drug-induced myopathy. Therefore, the device having ability to detect falls effectively is necessary. If the device could detect falls and send an alert signal in time, it could reduce harm caused by falling in patients and elders. The prototype of a fall detection device had been developed since the early of 1970s, wherein the device could only send a longdistance alarm when the user presses the button. In 1990s, there was the study of the automatic fall detection. The prototype of the fall detection device had been developed by William et al by using piezoelectric shock sensor and tilt switch to detect. Nowadays, the fall detection device has been continuously researched and developed as the tendency of the number of elders continues to increase every year leading to the increasing interest in the fall detection device.

Fall monitor and detection technology has been developed in various forms, varied from mostly in research to a product prototype sold in foreign countries. The method for detecting the behavior of falling depends on the use of device, for example, using the shock sensor for detecting impact force during the fall along with the mercury tilt switch for detecting the fall direction, using the video camera for capturing the fall that is advantageuous for some type of fall but causing elders lack of privacy with limitation in lighting and field of view constraint, as well as wearing the accelerometer on wrist or neck like accessories that seems convenient but providing low fall detection accuracy compared to waist-worn or chest-worn accelerator.

Although the fall detection device is various, the conventional device used in the study and the development of wearable fall detection system nowadays are motion sensors, such as, the accelerometer and/or the gyroscope which might be used in the form of a single sensor or multiple sensors to increase the efficiency of fall detection. The examples of fall detection devices developed in the prior art are as follows: 1. US Patent Application No. US20080129518 A1 titled“Method and system for fall detection” uses the tri-axial accelerometer and the impact detector, wherein the device measures the acceleration and the impact force obtained from the body-linked microphone to detect the fall, wherein the device could be worn on the wrist or attached to the chest. The fall detection depends on the analysis of acceleration signals in three phases— prefall phase, impact phase, and postfall phase. If the acceleration signal at any position and the duration of the acceleration signal are in the threshold-based standard, that is, there is the occurrence of fall.

2. Patent Application No. W02008129452 A1 titled “Multi-sensory fall detection system” is the concept of fall detection by using at least two detection devices, wherein said device might be the accelerometer or the vibration sensor, etc. The installation position of the device might be waist, ankle, or wrist. In case of installing the devices at waist and ankle, the fall detection could be proceeded by detecting the information measured by the devices at both positions.

3. US Patent Application No. US2009076419 A1 titled “Loss-of-balance and fall detection system” is the use of the foot force sensor by installing on the ankle or the shoe along with the accelerometer and the gyroscope attached to the chest and both thighs in order to use for the fall detection, wherein the device attached to the chest is used for classifying the behaviour of bending down, stretching, and turning around while the devices attached to thighs are used for classifying between standing and sitting.

4. US Patent Application No. US20090292227 A1 titled“Fall detection utilizing a three-axis accelerometer” is the fall detection by considering degrees between the body, the gravity force, and the acceleration magnitude calculated from the tri-axial accelerometer attached to the chest.

5. US Patent Application No. US201 10230791 A1 titled “Fall detection and/or prevention system” is the fall detection relying on the tri-axial accelerometer attached to the waist, wherein the fall is considered from the features set, such as, the acceleration, the acceleration magnitude, etc. If they exceed the predetermined value, it is possible that the fall occurs. If the system detects the occurrence of fall, the wearer could respond to the situation by pressing the button. If the button is not pressed after the fall, the system would send the distress signal. Or, if the wearer presses the button, the system would not send the signal and record the value obtained from the signal to check the fall prediction next time. Hence, the wearer could press the help button without the occurrence of the fall.

6. US Patent Application No. US20140313036 A1 titled“ Fall detection system and method” suggests the fall detection system which could adjust the fall detection algorithm, wherein the fall detection device might be the tri-axial accelerometer attached to wrist, body, or neck. The information from the device about 1.3 seconds would be recorded and processed on the cloud. The fall detection is considered from the acceleration signal in three phases— prefall phase, impact phase, and postfall phase. Normally, the acceleration signal in the prefall phase is in the range of 0.0.6g about 0.4 seconds. The impact phase is the phase that the acceleration signal is more than 1.25g for about 0.3 seconds. And, the postfall phase is the phase that the acceleration signal is approximate to lg about 0.6 seconds. If the signal in three phases follows said conditions, the system would notify the fall.

7. US Patent Document No. US9005141 B1 titled“ Ambulatory system for measuring and monitoring physical activity and risk of falling and for automatic fall detection” suggests the monitoring system using the accelerometer attached to the chest (the upper body) comprising four modules, including the postural transition detection module, the gait analysis module, the risk of falling assessment module, and the automatic fall detection module. In the postural transition detection module, it considers the the change in degrees between the body and the gravity force to classify the activities, such as, sitting-standing, standing-sitting, sitting- lying down, etc. The gait analysis module considers the acceleration signal in three axes, such as, considering the stepping foot that it is left or right foot by analyzing the acceleration signal in the lateral axis, considering the forefoot or heel strike by analyzing the acceleration signal in the fronrtal axis and in the vertical axis for the footstep speed, etc. The risk of falling assessment module considers from three values measured from postural transition, including the mean of transition duration, the standard deviation of transition duration, and the successive transition, wherein three values are high in those who used to fall. The automatic fall detection module considers the signal strength from the square root of the norm of acceleration in frontal and lateral plane whether it exceeds the threshold or not. The activities and postures of the wearer in the prior-to-impact time to the impact time would confirm the occurrence of fall if the signal exceeds the threshold or the peak after the occurrence of behaviour of walking or turning around followed by any other behaviour to the behaviour of lying down. The system considers various values, including the peak width, the signal speed in vertical prior to the peak, the norm of acceleration in three axes at the impact time, the norm of acceleration in frontal and lateral plane at the impact time, the axial speed in each axis at the impact time, and the energy of the norm of acceleration in frontal and lateral plane at the impact time. If all values follow the conditions, there is the occurrence of fall. Or, if the peak does not occur after the behaviour of walking or turning around, the system notifies that the fall occurs after the postural transition from any activities to the behaviour of sitting or lying down.

8. US Patent Document No. US7714728 B2 titled“ Using RFID to prevent or detect falls, wandering, bed egress and medication errors” is the use of RFID device to detect the fall, wherein RFID tag is attached to wrist, ankle, or sock and RFID receiver is attached to floor, door, bed comer, or bed, wherein the system would report the fall when it is found that RFID tag approaches the floor.

However, said technology has the limitations that it lacks accuracy for detecting the slow fall that often occurs when the falling person uses hand to hold on the object to reduce the impact force or when the slow fall occurs from unconsciousness . According to the research of Rubenstein et al., it is found that 22% of falls in elders are often the slow falls. Various research suggests that if the fall detection including slow fall is desired, the vital sign assessment is required. However, there is no confirmatory test result of the use of vital sign for detecting slow fall successfully.

Some research suggests to use one barometric pressure sensor for the fall detection along with the motion sensor to increase the accuracy of fall detection and to install the device at any one of positions chosen from, such as, head, wrist, or waist. For example, the research of Tolkiehn et al. could increase the accuracy from 81.48% to 86.97% by attaching the three- dimensional accelerometer and the barometric pressure sensor to the waist. Later, there is the improvement of algorithm by using the posture information after the impact detection. It is found that it could increase the accuracy to 90% by using only three-dimensional acceleration signal from the same information set. However, both methods could not detect the slow fall accurately (in the work of Tolkeihn, the slow fall in the posture“resting against a wall, then sliding down” is classified in the non-fall group in the work thereof). Although the use of one barometric pressure sensor to evaluate the height decrease along with the three-dimensional accelerometer could be seen sometime in the previous research, if the barometric pressure signals includes high-frequency noise caused by changes in barometric pressure all the time (or immediate change, such as, entering-exiting the elevator) allowing the surrounding air to directly affect the obtained signal. This decreases the accuracy in the height decrease evaluation that directly affects the slow fall detection.

The use of RFID technology to provide reference position to measure the height of any positions in the body has the limitation in the size of the reader which requires high energy level (so that it requires a large battery, frequent battery replacement, and energy supply tube) and the communication signal might not cover the area as expected.

To provide more flexible and effective fall detection, the present invention thus invents the method for detecting falls by using barometric pressure sensors installed at least two positions in different heights, wherein at least one sensor is provided on the user’s body, such as, waist and ankle, head and ankle, or waist and floor, etc. The method according to the present invention uses the relative barometric pressure signals to increase the accuracy in the fall detection, particulary the slow fall.

Summary of the invention

The present invention relates to the method for detecting falls by using relative barometric pressure signals, comprising step of receiving signal, step of processing, and step of displaying, wherein the step of receiving signal receives the barometric pressure signals from at least two barometric pressure sensors attached to the different height or on the same height, wherein at least one barometric pressure sensor is provided in any position on the user’s body and the other barometric pressure sensor is provided in the reference position to be used as the reference at the other position on the body in different height or any other position in the surroundings. When there is the postural transition or in prefall or postfall phase, there is the change in the difference between the barometric pressure value from the barometric pressure sensor at any position and the barometric pressure value from the barometric pressure sensor at the reference position at least one pair when comparing two values obtained from two sensors prior to the postural transition. For the step of processing, using the change in the barometric pressure values to indicate the change in height of the device position to detect falls, activities and/or prefall activities. The obtained value from the reference barometric pressure sensor could calculate the difference in the barometric pressure values illustrating the motion of body more accurately as it could confute the change in the barometric pressure values altered based on time, place, and surroundings, wherein the step of processing comprises the step of signal inference in the change of signal obtained from the step of receiving signal as the information for the monitoring for patients and elders.

The barometric pressure information from at least two barometric pressure sensors could help increase the accuracy of the fall detection, particularly the slow fall, and could be used to detect the prefall activities of the wearer. The information obtained from the information analyzing system according to the present invention would be gathered to apply to various portions of the monitoring system for patients and elders continuously and suitably. This helps protect, detect and/or analyze the cause of the fall accurately and allows the patients and elders to live freely and safely.

Brief description of the drawings

FIG. 1 shows the components of the method for detecting falls by using relative barometric pressure signals.

FIG. 2 shows the components of the method for converting the signals by using relative barometric pressure signals.

FIG. 3 shows the steps of the method for signal inference.

FIG. 4 shows the signal graph of the barometric pressure signals obtained from the sensors at different positions including wall (upper left portion), ankle (upper right portion), waist (lower portion) while the wearer falls and performs various activities.

FIG. 5 shows the signal graph of the three-dimensional acceleration signals obtained from the sensors at different positions including wall (upper left portion), ankle (upper right portion), waist (lower portion) while the wearer falls and performs various activities.

Detailed description of the invention

The description of the present invention is carried out by giving the example and referring to the drawings as the example to clearly illustrate. Moreover, the identical objects shown in these drawings are represented by the identical numbers. Hence, it does not limit, and the scope of the invention follows the attached claims.

The method for detecting falls by using relative barometric pressure signals according to the present invention is illustrated according to FIG. 1 , comprising:

Step of receiving signals 1 functions to receive the barometric pressure signals from the barometric pressure sensor or the barometer, wherein the step of receiving signals 1 receives the barometric pressure signals from at least two barometric pressure sensors attached to different or the same height, wherein at least one barometric pressure sensor is attached to any position on the user’s body and the other one barometric pressure sensor is attached to the reference position to be used as the reference on another position on the body at different height or any position in the surroundings. When there is the postural transition or in prefall and postfall phase, there is a change in difference between the barometric pressure value from the barometric pressure sensor at any position and the barometric pressure value from the barometric pressure sensor at the reference position at least one pair compared to two barometric pressure values obtained before the postural transition, wherein the barometric pressure sensor is attached on any position on the body at the height from the floor of at least 20 cm when the user is attached with the sensor in the standing posture;

Step of processing 2 functions to receive the signals from the step of receiving signals 1 and convert the signal to the information for the monitoring system for patients and elders 16, wherein the step of processing 2 uses the change in the barometric pressure values to indicate the change in height of the position of the device attached to detect the fall, the activities and/or the prefall activities, wherein the value obtained from the reference barometric pressure sensor allows the calculation of the difference in the barometric pressure values indicating the motion of the body more accurately as it could confute the change of barometric pressure altered based on time, place, and surroundings, wherein the step of processing 2 comprises the step of signal inference 14 in the conversion of the signal obtained from the step of receiving signals 1 to the information for the monitoring system for patients and elders 16.

Also, the step of processing 2 could further comprises the model 13 generated from the training dataset or the expert knowledge; and

Step of displaying 3 functions to display the information for monitoring system for patients and elders 16. The example of the step of displaying 3 is, such as, displaying in the form of the alert sound, the computer monitor, the wireless data transmission to another device, the automatic fall protection device, the alert message to the individual, or the record in the database to used as the information for healthcare data analysis of the user at least one or the combination thereof.

Moreover, the method for detecting falls by using relative barometric pressure signals according to the present invention further comprises the step of model learning 12. For the step of signal inference 14 as mentioned in the step of processing, there is steps in the step of signal inference 14 that could be processed within the same processing device or in different processing devices, wherein the results of each step would be transmitted to the other device via the wired or wireless network, wherein the step of signal inference 14 comprises the steps as follows:

- Step of data preprocessing 21 functions to receive the barometric pressure signals 15 from the barometric pressure sensors or the barometers and convert the signals into the usable form for the inference,

wherein, in the step of data preprocessing 21, the signal information might be converted into the usable form for the inference by the method of noise filtering, data transformation, data reduction, or data normalization at least one or the combination thereof;

- Step of feature extraction 22 functions to convert the signal obtained from the step of data preprocessing 21 into different attributes of signals,

wherein, in the step of feature extraction 22, there might be the attributes of signal obtained from the raw signals or the attributes of signal in any time window, wherein the example of the attributes of signal in any time frame is, such as, mean, standard deviation, magnitude, standard deviation magnitude (SVM), slope of SVM, minimum,, maximum, root mean square, Shannon entropy, energy, rate of change, or temporal change at least one or the combination thereof of the barometric pressure signals, the difference of the barometric pressure values at two positions, or the ratio between the barometric pressure values at two positions;

- Step of classifying activities, postures, and falls 23 functions to convert the attributes obtained from the step of feature extraction 22 to the information for monitoring system for patients and elders 16 by the technique of machine learning, expert system, or hybrid thereof, wherein, in the step of classifying activities, postures, and falls 23, there is the sample information of activities and postures including lying down, sitting, standing, walking, running, jumping, static activity, dynamic activity, fall, prior-to-fall activity, or post-fall activity at least one or the combination thereof.

According to the abovementioned method for detecting falls by using relative barometric pressure signals, apart from the use of information from the barometric pressure sensor in the processing, it is also used to process with the motion sensor, such as, accelerometer, gyroscope, impact detector, or vibration sensor at least one or the combination thereof.

Hereinafter, the steps of the method for detecting falls by using the relative barometric pressure signals is described by using one embodiment (according to the drawings) as follows:

FIG. 1 shows the components of the method for detecting falls by using relative barometric pressure signals according to the present invention, wherein comprises:

Step of receiving signals 1 means the step comprises at least two barometric or pressure sensors attached to differenct positions by having at least one barometric pressure sensor attached on any position on the user’s body, such as, head, neck, ear, back, chest, arm, armpit, wrist, front waist, side waist, hip, leg, or foot etc. and another barometric pressure sensor attached on another position on the body or any position in the surroundings, such as, wall, door, or ceiling, wherein the barometric pressure sensors attached on the body are either wearable or implantable electronic devices. As the barometric pressure value at the lower position in the same surrounding is higher due to the gravity force, the barometric pressure signals 15 obtained from the barometric pressure sensor attached at different heights would be different and could be used to estimate the distance in height of two barometric pressure sensors any time;

Step of processing 2 functions to receive the signal from the step of receiving signal 1 and convert the signal to the information for monitoring system for patients and elders 16, wherein the step of processing 2 means the circuit or the method comprising the step of signal interference 14 that converts the signal obtained from the step of receiving signal 1 as the information for monitoring system for patients and elders 16 comprising phases of fall (including prefill, impact, and postfall), getup/recovery, fall direction, prefall and postfall activity at least one signal to transmit to the step of displaying 3. Moreover, the step of processing further comprises the model and in FIG. 2 shows the components of the method for converting barometric pressure signals 15 comprising two main components, including the step of model learning 12 and the step of signal inference 14, wherein the step of model learning 12 might be included or not included in the processing unit. The model 13 is generated from the training dataset comprising at least two barometric pressure signals. The step of model leamin 12 where the model 13 is used in the step of signal inference 14 from two barometric pressure signals 15 entering the system during the operation, wherein the step of processing 2 comprises the step of signal inference 14 that could detect the change in barometric pressure at two positions of the barometer 1 caused by the motion of the user. As this method could use the barometric pressure under the same surrounding as the reference, it could refer or estimate the information of change in the barometric pressure particularly the vertical motion. In the practical way, the change in barometric pressure values might be caused by other reasons, such as, elevating, entering the elevator, moving out of the building, opening/closing door or window, etc. At least two barometric pressure signals 15 under the same surrounding thus provide information to analyse the motion of activities at any positions and any phases to convert the signal to the information for monitoring system for patients and elders 16 and transmit to the step of displaying 3, wherein the step of signal inference 14 comprises the step of data preprocessing 21, the step of feature extraction 22, the step of classifying activities, postures, and falls 23; and

Step of displaying 3 functions to display the information for monitoring system for patients and elders 16, wherein the step of displaying 3 means a part of the patients and elders monitoring system that obtain the information for monitoring system for patients and elders 16 to use and could display in the form of the alert sound, the computer monitor, the wireless data transmission to another device, the automatic fall protection device, the alert message to the individual, or the record in the database for the healthcare data analysis of the user later.

FIG. 3 shows the steps of signal inference 14, wherein comprises of the steps as follows:

Step of data preprocessing 21 functions to receive the barometric pressure signals 15 from the barometers 1 and convert the signals to the usable form for the inference by using different methods, that is, noise filtering, data transformation, or data reduction, data normalization to reduce the difference in the useless signal for the step of signal inference or the step of signal classification (wherein the difference is caused by user, hardware, size of battery, surrounding etc.);

Step of feature extraction 22 functions to convert the signals obtained from the step of data preprocessing to various attributes of the signals which are raw signals, difference of raw signals, signals after the step of data preprocessing 21, or the attributes of signals at any time window, such as, mean, standard deviation, magnitude, standard deviation magnitude (SVM), slop of SVM, minimum, maximum, root mean square, Shannon entropy, or energy at least one or the combination thereof of the barometric pressure signal(s) and/or the difference of barometric pressure signals at least two positions or the ratio between the barometric pressure signals at least two positions;

Step of classifying activies, postures, and falls 23 functions to convert the attributes obtained from the step of feature extraction 22 for the information for monitoring system for patients and elders 16 including the information of activities and postures, such as, lying down, sitting, standing, walking, running, jumping, lying down/sitting, static activity, dynamic activity, fall, prefall activity, or postfall activity at least one or the combination thereof, wherein this step could be analyzed by using the technique of machine learning, expert system, or hybrid thereof, wherein the parameters of the model or the threshold for classifying activities, postures, and falls use the value obtained from the model 13 generated in the step of model learning 12 for the information for monitoring system for patients and elders 16 obtained from this step to be transmitted to the step of displaying 3 directly or stored in the database or memory,

wherein the components of the step of signal inference 14 might be included in a single device or multiple devices, such as, the signal or the attributes calculated from the device with barometer 1 might be transmitted to calculate in the device with barometer 2 or the signal or the attributes calculated from the device of barometer 1 and the signal and the attributes calculated from the device of barometer 2 might be transmitted to another receiver or server, etc.

In the exemplary test of the method for converting barometric pressure signals 15, there is the fall test of 8 male volunteers and 2 female volunteers to attach two barometric pressure sensors to each volunteer, wherein one barometric pressure sensor is attached to the waist (waistband) or the belt and another barometric pressure sensor is attached to the ankle (to any one of shoe), wherein the second barometric pressure sensor is used as the reference and might be attached to the floor or the wall instead. Each volunteer performs five falling postures (including slowly falling forward, fast falling forward, slowly falling rearward, slowly falling from the chair, and slowly falling from the bed) and five daily activities (including walking, standing, sitting on the chair, lying down on the bed, and collecting object from the floor) three times per each volunteer. Then, comparing the model for detecting falls and prefall postures that is generated from the application of machine learning using four machines including the linear support vector machine, the Medium Gaussian support vector machine, the decision tree, and the K-nearest neighbor (KNN) by generating the model from the attributes of the barometric pressure signals in different phases of signal after the step of data preprocessing (such as, noise filtering, data normalization). In this test, we separate the signal into three phases, including prefall phase, impact phase, and postfall phase. In each phase of signal, six attributes are calculated from three phases of signals from both sensors providing total 36 attributes per one sample, wherein the attributes include standard deviation, minimum, maximum, root mean square, Shannon entropy, and energy of the signal phase. The accuracy of the fall detection is shown in Table 1.

Table 1 shows the results of the method of motion signal conversion 15

According to Table 1, it indicates that we could detect most of the slow fall up to 94% by using only the attributes calculated by two barometric pressure sensors. Moreover, there is the use of the model and the attributes to check the default value (standing, sitting, lying down) and it is found that the average detection accuracy is about 81.7%.

FIG. 4 and FIG 5 show the signal graph of barometric pressure signals and three- dimensional acceleration signals obtained from the sensors at different positions, such as, wall (upper left portion), ankle (upper right portion), and waist (lower portion) while the wearer is falling and performing 16 activities, including falling to kneel, falling to lie down, falling and trying to get up, falling to lie on the right side, falling to lie on the left side, falling from the chair, falling and getting up to walk, falling and getting up, falling and lying down, falling to sit on the chair, falling to sit on the floor by leaning against the wall, getting up from the chair, sitting on the chair, jumping at one place, bending down to collect object, and bending down to tie shoelace. It could be seen that the signal from the barometric pressure sensor fixed to the wall changes according to change in barometric pressure and the trend of the signal is similar to the signal from the barometric pressure sensor attached to the ankle, wherein this value could be used to adjust the bias in the signal of the barometric pressure sensor attached to the waist causing the obtained information to indicate the change in the barometric pressure caused by the change in the activities more clearly. Although the present invention is complete described by using the attached drawings, it could be comprehensible that any modifications or adjustments done by the person skilled in the prior art and related field under the scope and the objective of the invention specified in the attached claims and further covering the characteristics of the invention omitted from the attached claims specifically but providing the same advantages and results specified in the characteristics of the invention speficied in the attached claims.

Best mode of the invention

As mentioned in detailed description of the invention.

Claims

1. The method for detecting falls by using relative barometric pressure signals according to the present invention, wherein comprising: step of receiving signals (1) functions to receiving the barometric pressure signals from the barometric pressure sensors or the barometers; step of processing (2) functions to receive the signals from the step of receiving signals (1) and convert the signal to the information for monitoring system for patients and elders (16); and step of displaying (3) functions to display the information for monitoring system for patients and elders (16),

characterized in that: the step of receiving signals (1) receives the barometric pressure signals from at least two barometric pressure sensors or the barometers attached to the positions at different height or at the same height, wherein at least one barometric pressure sensor is attached to any position on the user’s body and another barometric pressure sensor is attached to the reference position to be used as the reference on any other position on the body at different height or any position in the surroundings, wherein, when there is the postural transisiton or change in prefall and postfall activities, there is a change in the difference between the barometric pressure value from the barometric pressure sensor at any position and the barometric pressure value from the balometric pressure sensor at the reference position at least one pair compared to the values obtained from two barometric pressure sensors prior to the postural transition; for the step of processing (2), using the change in the barometric pressure values to indicate the change in height of the position of the device attached to detect falls, activities and/or prefall activities, wherein the value obtained from the reference barometric pressure sensor could be used to calculate the difference of barometric pressure to indicate the motion of the body more accurately as it could confute the change in the barometric pressure values altered based on to time, place, and surroundings, wherein the step of processing (2) comprises the step of signal inference (14) to convert the signals obtained from the step of receiving signals (1) to the information for monitoring system for patients and elders (16).

2. The method for detecting falls by using relative barometric pressure signals according to claim 1, further comprises the step of model learning (12) that receives the signal from the step of receiving signals (1) to transmit to the step of processing (2).

3. The method for detecting falls by using relative barometric pressure signals according to claim 1, wherein the step of processing (2) further comprises the model (13) generated from the training dataset or the expert knowledge.

4. The method for detecting falls by using relative barometric pressure signals according to claim 1, wherein the step of signal inference (14) comprises:

step of data preprocessing (21) functions to receive the barometric pressure signals (15) from the barometric pressure sensors or the barometers and convert the signals to the usable form for the inference;

step of feature extraction (22) functions to convert the signals received from the step of data preprocessing (21) to attributes of signals;

step of classifying activities, postures, and falls (23) functions to convert the attributes obtained from the step of feature extraction (22) to the information for monitoring system for patients and elders (16).

5. The method for detecting falls by using relative barometric pressure signals according to claim 4, wherein the step of data preprocessing (21) is where the signals are converted to the usable form for the inference by the method of noise filtering, data transformation, data reduction, or data normalization at least one or the combination thereof.

6. The method for detecting falls by using relative barometric pressure signals according to claim 4, wherein the step of feature extraction (22) has the attributes of signals obtained from raw signals or the attributes of signals in any time window.

7. The method for detecting falls by using relative barometric pressure signals according to claim 6, wherein the attributes of signals in any time window ar mean, standard deviation, magnitude, standard deviation magnitude (SVM), slope of SVM, minimum, maximum, root mean square, Shannon entropy, energy, rate of change, or temporal change at least one or the combination thereof of the barometric pressure signal(s) or the difference in barometric pressure signals at two positions, or the ratio between barometric pressure signals at two positions.

8. The method for detecting falls by using relative barometric pressure signals according to any one of claims 1 to 7, wherein further comprising the processing with at least one of the motion sensor.

9. The method for detecting falls by using relative barometric pressure signals according to claim 8, wherein the motion sensor might be accelerometer, gyroscope, impact detector, or vibration sensor at least one or the combination thereof.

10. The method for detecting falls by using relative barometric pressure signals according claim 4, wherein the step of classifying activities, postures, and falls (23) has the activity and posture information including lying down, sitting, standing, walking, running, jumping, static activity, dynamic activity, falling, prefall activity, or postfall activity at least one or the combination thereof.

11. The method for detecting falls by using relative barometric pressure signals according to claims 4 to 10, wherein the step of classifying activities, postures, and falls (23) analyzes the information by using the technique of machine learning, expert system, or the hybrid thereof.

12. The method for detecting falls by using relative barometric pressure signals according any one of claims 1 to 11, wherein the step of signal inference (14) processes on the same processing device or various processing devices, wherein the result of each step is transmitted to another device via wired or wireless network.

13. The method for detecting falls by using relative barometric pressure signals according to claim 1, wherein the step of displaying (3) displays the result in the form of alert sound, computer monitor, wireless data transmission to another device, operation of automic fall protection device, alert messege to individual(s), or record in database for healthcare analysis for the wearer at least one or the combination thereof.

14. The method for detecting falls by using relative barometric pressure signals according to any one of claims 1 to 13, wherein the installation of a barometric pressure sensor on any position of the body at the height from the floor at least 20 cm when the user is attached with the sensor in the standing posture.