SYSTEM AND METHOD FOR PREVENTING AND PREDICTING THE RISK OF POSTURAL DROP

Abstract

The present invention patent application for a system and method for preventing and predicting the risk of postural drop is in particular aimed at the health innovation sector, hardware engineering, software engineering, medical engineering, biotechnology, and more specifically to the field of wearable electronic systems for preventive medicine. The inventive solution is in the composition of solutions involving method and system enhanced through the use of specific signal processing, providing the clinical environment (4) and daily environment (5) with a low-cost, easy-to-understand solution that can be measured using graphs and numbers, as opposed to the existing clinical approach based on subjective evaluation, or the existing prior art that involves preventing drops exclusively from a motor perspective, and not systemically as recommended by scientific evidence.

Claims

1) SYSTEM FOR THE PREVENTION AND PREDICTION OF POSTURAL FALL RISK to prevent falls through an objective tool in the clinical environment (4) and everyday environment (5), based on clinical and motor information for the assessment, screening and monitoring of the risk of falling and motor performance with encouragement for the practice of preventive physical rehabilitation, offering systemic and interdisciplinary coverage to support clinical decision-making and remote monitoring of the user, to impact fall prevention, improve the quality of life of the elderly and intelligent management of health resources due to the decrease in the rate of hospitalizations and costs with falls in the hospital, clinical and domestic environment, characterized by the fact that the system includes patient (P) for collecting clinical data (D) and motor data (D1) with the dedicated application and sensor (1) (cell or Inertial sensor), according to which the therapist (T) collects information processed by a binary method that classifies as positive or negative, the prevention or absence of the risk assessed in a questionnaire, said therapist (T) asks the patient (P) to perform tasks including anticipatory and compensatory reactions, sensory guidance and dynamic gait balance; the patient (P) uses the sensor (1) coupled to the body, which automatically processes the inertial signal of the tasks to generate the risk indexes of falling, walking and balance, and classifying the level of risk as low, moderate or high, through processing (PI), and part of the algorithms is embedded in the sensor (1) of the cell (2) and the other part is processed in the cloud (3), instantly in the presence of the wireless signal, via Bluetooth connection, or as soon as possible, as the system enters wireless coverage; the patient (P) has their risk constantly controlled by continuous and automatic calculation of the risk, with partial processing autonomy (PI) between the apparatus and the patient's (P) cell (2); for prevention, the Rosenstein index rate reduction trigger triggers the vibration or beep of the apparatus to request the patient's motor attention (P); the user's motor behavior can also be monitored remotely, from cloud processing (3), by the therapist (T) or family member (F).

2) SYSTEM FOR PREVENTION AND PREDICTION OF POSTURAL FALL RISK, according to claim 1, characterized by the fact that it includes a clinical environment (4) and a daily environment (S), and part of the data (4) is offline (6) and in the cloud (3); among the related data are the clinical data (D) APP (7), the inertial data 9D (8), the inertial data 9D merged by the signal processing by quaternions (9), the inertial data 9D (8) communicating with Rosenstein prediction of fall risk (10) and the data merged by the signal processing by quaternions (9) with the gait and balance algorithms (11), with a communication of these with the gait and balance indexes (12) available in real time to support preventive assessment and rehabilitation; in the clinical and domestic environment (4) there is also the screening (13) of fallers by risk level with the suggestion of clinical guidance, this communicator with the risk monitoring (14) in the everyday environment (S) with preventive vibration or beep biofeedback in the continuous use of a wearable device.

3) SYSTEM FOR PREVENTION AND PREDICTION OF POSTURAL FALL RISK, according to claims 1 and 2, characterized by the fact that the results are achieved from the raw motion data from accelerometers (A1), gyroscopes (GI) and magnetometers (M!) of the inertial sensor (1), (cellular (2) or sensor (1) inertial) coupled to the patient's body (P) in the clinical assessment of the risk of falling, or in the same way, continuously coupled to the domestic patient (P).

4) METHOD of integrating clinical and motor information obtained in the system of claims 1 to 3, characterized by the integration of clinical data (D) APP (7) (clinical information) and gait and balance algorithms (11) (objective gait indexes and balance), by application and processing of measured vectors of the inertial signal, used on the patient's body (P) for clinical testing—prediction of the level of risk of falling (10) (Rosenstein risk—performed by a technician in the clinical environment (4) referring the patient (P) to the most appropriate treatment; these combined items generate the calculation of the risk of falling, generating a result (R1) that may indicate high risk (1S), moderate risk (16), low risk (17) or no risk (18).

5) METHOD, according to claim 4, characterized by the fact that in the table clinical data (D) APP (7) the patient (P) is subjected to a test (19) based on clinical data (D) APP (7), generating the evaluation (20) and the results: risk of falling+=3 (21), risk of falling+=2 (22), risk of falling+=1 (23) and risk of falling+=0 (24); then, in the gait and balance algorithms (11), the patient (P) is subjected to a test2 (2S) based on these indicators, generating the “motor parameters” (26) and the results: risk of falling+=3 (21) and risk of falling+=0 (24); in the table prediction of the level of risk of falling (10) (Rosenstein) the patient (P) is submitted to a test3 (27) that leads to the risk prediction (10B), with the results: risk of falling+=3 (23) and risk of falling+=0 (24).

6) METHOD, according to claim 4, characterized by the fact that after the assessment and calculation of the risk of falls, the result (R1) generated produces a risk stratification, in the table shown as high risk (15) a value >6 where there are two options: high risk 12-9 (28) and high risk 8-6 (29), both stratification ranges followed by appropriate clinical guidelines.

7) METHOD, according to claim 4, characterized by the fact that in the moderate risk frame (16) a value is set between 5-3 and in the low risk frame (17) a value is set between 2-1, also followed by adequate guidelines to support clinical decision.

Description

DESCRIPTION OF THE FIGURES

[0041] The invention will be explained below in a preferred embodiment, and for better understanding, references will be made to the attached drawings, in which they are demonstrated:

[0042] FIG. 1: Flowchart of the system showing the flow and composition of the data of the invention and how the results are shared with the clinician, the user and the family member;

[0043] FIG. 2: Flow diagram of the system showing how the data is processed and composed to generate the results, the processing autonomy in the offline and cloud environment and how the results are intended for the clinical environment for professional use and everyday environment for the user biofeedback and monitoring of family members;

[0044] FIG. 3: Flow of data modeling according to Madgwick's method of signal error modeling from quaternions in relation to the inertial navigation system (SNI);

[0045] FIG. 4: Schematic representation of the stability of human gait, according to the author. In this Figure, “the undisturbed dynamics of the xe (t) gait is represented by the red orbit with a radius ε. The ε limit represents a small distance to the disturbed gait dynamics x (t) for a small perturbation of size d (0) <.sup.δ, which does not interfere with stability. (B) A schematic representation of ordinary stability (green line), asymptomatic stability (yellow line) and exponential stability (red line) within a section of time (blue dashed lines). Distance d (t) between the undisturbed xe (t) and the disturbed gait dynamics x (t) must be within radius c for the dynamics to be stable. In addition, the yellow arrow must approach the undisturbed path xe (t) (red central line) as t −>∞. In addition, the red arrow must approach the red exponential arrow, where d (t) ≤d (0) exp (λt) and the exponent of Lyapunov λ<0, when the dynamics are exponentially stable” (Bizovska L, Svoboda Z, Janura M, Bisi M C, Vuillerrne N Local dynamic stability during gait for predicting falls in elderly people: A one-year prospective study. PLoS ONE. 2018;13(5): and 0197091. https://doi.org/10.1371/joumal.pone.0197091). “The size of a disturbance is evaluated as the distance d (0)=II×(0)−xe (0) II between the disturbed point x (0) and the undisturbed point xe (0) in the reconstructed state space (see upper red vertical arrow in FIG. 1B of FIG. 4). The reaction to the disturbance is numerically defined as the temporal change, d (t)=II×(t)−xe (t) II, from the initial distance d (0) between the trajectory of the disturbed gait x (t) and the undisturbed trajectory of the gait xe (t)” (Bizovska L, Svoboda Z, Janura M, Bisi M C, Vuilletme N Local dynamic stability during gait for predicting falls in elderly people: A one-year prospective study. PLoS ONE. 2018;13(5): e0197091. https://doi.org/10.1371/joumal. pone. 0197091);

[0046] FIG. 5: Rosenstein fall prediction method, according to the calculation of the maximum Lyapunov exponent. The log of the expansion/contraction of the Euclidean distance between these points of the neighboring estimated trajectories of the march by max)′ 11, represents the rate of divergence or convergence of the march signal or the degree of local dynamic stability of the system;

[0047] FIG. 6: Fall risk level classification method involving clinical information data, motor information (objective gait and balance indexes) and Rosenstein risk prediction, in order to make a calculation of the risk framework and generate results involving the levels risk.

DETAILED DESCRIPTION OF THE INVENTION

[0048] The “SYSTEM AND METHOD FOR PREVENTING AND PREDICTING THE RISK OF POSTURAL FALL”, object of this patent application, describes a system and method for preventing falls that has the purpose of offering an objective tool in the clinical and domestic environment, based on clinical and motor information for the assessment, screening and monitoring of the risk of falling and motor performance with encouragement to the practice of preventive physical rehabilitation available as an objective tool for health professionals and as a tool for remote control of risk, in the domestic environment, by family members and biofeedback to support motor correction for the user. The main objective of the method is to offer a systemic and interdisciplinary coverage to support clinical decision-making and remote user monitoring, to impact fall prevention, improve the quality of life of the elderly and intelligent management of health resources by decreasing the rate of hospitalizations and costs with falls in the hospital, clinical and domestic environment.

[0049] In FIG. 1, the system flowchart presents the patient (P) in the collection of DONE clinical data (D) and motors (D1) with the dedicated application and sensor (1) (cell or inertial sensor). In the application, the therapist (T) collects information processed by a binary method that classifies the prevention or absence of the risk assessed in a questionnaire about the previous history of risk of falls with positive or negative, integrating questions about previous fall, use of risk medication, urinary incontinence, among others. Still in the application, the therapist (T) asks the patient (P) to perform 14 tasks that include sitting and standing up, standing with eyes closed, under an uneven surface, walking, sitting and going back, among others (clinical test based on MiniBestest); used to evaluate anticipatory and compensatory reactions, sensory guidance and dynamic gait balance. During the performance of these tasks, the patient (P) uses the sensor (1) attached to the body, which automatically processes the inertial signal of the tasks to generate the risk indexes of falling, gait and balance, and classification of the level of risk at low, moderate or high risk, according to processing autonomy (PI) shown in the figure (offline (6) and cloud (3)), and part of the algorithms are embedded in the sensor (1) of the cell (2) and the another part is processed in the cloud (3), instantly in the presence of the wireless signal, via Bluetooth connection, or as soon as possible, as soon as the system enters the wireless coverage. This clinical solution extends to physical rehabilitation, as a measure of the patient's performance result in physical rehabilitation sessions and for home use, if the therapist (T) understands that the patient (P) remains at risk and needs to continually use the apparatus wearable in their daily activities. In this case, the patient (P) has their risk constantly controlled by continuous and automatic calculation of the risk, with partial processing autonomy (P1) between the apparatus and the cell (2) of the patient (P). For prevention, the Rosenstein index rate reduction trigger will trigger the vibration or beep of the apparatus to request the patient's motor attention (P). The user's motor behavior can also be monitored remotely, from cloud processing (3), by the therapist (T) or family (F).

[0050] FIG. 2 shows a flow with a clinical environment (4) and a daily environment (5), with part of the clinical environment (4) being offline (6) and part in the cloud (3); among the related data are the clinical data (D) APP (7), the 9D inertial data (8), the quaternion fusion sensors (9), the 9D inertial data (8) communicating with the Rosenstein prediction of fall risk (10) and the quaternion fusion sensors (9) with the gait and balance algorithms (11), with a communication between these with the gait and balance indexes (12) available in real time to support preventive rehabilitation. In the clinical environment (4) there is also the screening (13) of fallers by risk level with the suggestion of clinical guidance, this communicator with the risk monitoring (14) in the domestic environment with preventive biofeedback by continuous use of a wearable device.

[0051] Thus, FIGS. 2 and 3 demonstrate how the results are achieved from the raw motion data from the 3 accelerometers (A1), 3 gyroscopes (G1) and 3 magnetometers (M1) of the inertial sensor (1) cell (2) or inertial sensor (1) coupled to the patient's body (P) in the clinical assessment of the risk of falling, or in the same way, continuously coupled to the domestic patient (P), according to two distinct flows:

[0052] 1) Fusion of data from the 9 sensors of the MEMs plate of the cell (2) or sensor (1), by the Madgwick signal modeling method, by rotated quatemions (Yingyong Yuddha A, Saengsiri Suwon AT, Panichaporrt W, et al. The Mini-Balance Evaluation Systems Test (Mini-BESTest) Demonstrates Higher Accuracy in Identifying Leader Adult Participants With History of Falls Than Do the BESTest, Berg Balance Scale, or Timed Up and Go Test. J Geriatr Phys Ther. 2016;39(2):64-70. DOI: 10.1519/JPT 000000000). From the output of the modeled signal, there are the three motion vectors in the axes y (Pitch), x (Roll) and z (Yaw), agreed according to the initial leveling to the SIN rotated to the NED, forming the 3D-axes for data of angular movement, which provide information of attitude and orientation with better precision. The data is exported in real time by Bluetooth low energy for integration with the cell and application, which, in tum, has processing autonomy so as not to lose data in the absence of a wireless signal. At the same time, the inertial signal fused in the 9 degrees of freedom with autonomy for offline processing (6) exports the data for cloud processing (3) of the gait and balance calculations and Rosenstein for gait, balance and risk of falling. All information is integrated, clinical and motor, and made available for reading in real time in the application for the objective motor quantification of gait and balance indexes and classification of the level of risk of falling, with subsequent suggestion of adequate clinical guidance, biofeedback of vibration or beep, for the wearer of the device continuously wearable in the domestic and everyday environment, whenever their risk of falling, supervised by a continuous automatic risk assessment method, observes the dispersion of the signal from time to time by rendering the automatic algorithm, gait and balance suggesting risk. Given this information, the patient (P) can proactively correct their posture or seek specialized care. Family members (F) can also have access to the dedicated application to accompany their elderly family member, user of the wearable fall prevention device, to find out how long they spent sitting, lying down, if they did the requested walks or if they have received a risk alarm;

[0053] 2) Calculation of prediction of risk of fall by the method of Rosenstein from the acceleration information of the three accelerometers (A1). In other words, disturbances in the dynamic gait system can be represented by calculating the local dynamic stability by reproducing the influence of these disturbances, in the estimated neighboring trajectories, which deviate from the original trajectory of the gait represented in the state space. Gait stability is maintained by the control of active and passive stiffness of the corrective neuromuscular system, through the strategy of recruiting active muscles combined with the motor adjustment of the joint torques necessary to maintain dynamic balance (Lockhart T E, Liu J 2008 Differentiating fall prone and healthy adults using local dynamic stability. Ergonomics 51, 1860-1872. doi: 10.1080/00140130802567079 (doi: 10.1080/00140130802567079″. “The “stable” gait can be pragmatically defined as one that does not lead to falls, despite disturbances” (Bruijn S M, Meijer O G, Beek P J, van Dieen J H (2013) Assessing the stability of human locomotion: a review of current measures. J R Soc Interface 10:20120999). The measure of local dynamic stability has already been demonstrated in a clinical study to be able to differentiate gait between elderly fallers, healthy elderly and young people (Lockhart TE, Liu J 2008 Differentiating fall prone and healthy adults using local dynamic stability. Ergonomics 51, 1860-1872. doi: 10.1080/00140130802567079, doi: 10.1080/00140130802567079). The capacity of the neuromuscular system to attenuate gait disturbances is the target of measuring local dynamic stability based on nonlinear dynamic theory.

[0054] In FIG. 4 the author explains, through the schematic drawing, how gait stability is understood. That is, the flow (F4) shows the gait dynamics of a patient (P). Below, the graph (G2) illustrates the curves obtained for ordinary stability (E), asymptotic stability (E1)) and exponential stability (E2), where x(t) represents disturbed trajectory and x.sup.e(t) represents imperturbable trajectory.

[0055] Lyapunov exponents of a trajectory measure the average rate of dispersion or conversion of nearby trajectories, in the context of human gait. If there is convergence of the signal, the stability of the system can be understood as good. However, the greater the divergence of the signal, the more chaotic the system is considered. Studies show that the Lyapunov Maximum Exponent measurement in short time series is ideal for calculating gait stability, to the detriment of samples over long time (World Health Organization [WHO). WHO global report on falls prevention in olderage. 2007. Available from: http://www.who.int/ageing/publications/Falls_prevention7March.pdf?ua=1). Rosenstein proposed an alternative approach to calculate the largest Lyapunov exponent (max λ1) from a short time series, with less noise influence. The Rosenstein method for calculating maxλ1 consists of the following steps, represented in the block flowchart of FIG. 5, where the block (B1) of original time series data (AP acceleration, 40 travel cycles) is represented; block (B2) (time delay—10 frames); block (B3) (incorporation dimension); block (B4) (reconstructed state space) submitted to the Rosenstein algorithm, which feeds the block (B5) (mean divergence between close trajectories) and from this to the walking stage (B6) (Lockhart TE, Liu J Differentiating fall-prone and healthy adults using local dynamic stability. Ergonomics. 2008; 51:1860-725): “State vectors are reconstructed from {xi} data. For each point i in the reconstructed space, the nearest neighbor j that is outside a time limit greater than the mean period of the series (estimated from the Fourier transform of the series), so that/i-j/ >average period. The temporal evolution of the distances between these points is followed up to a time limit Δt, accumulating the value of the logarithms of the distances and calculating the average of these values (dividing by Δt)” [19]. “The rocedure from (1) to (3) is repeated for different pairs of points, allocating the logarithm of the mean of these divergences as a function of the current value of Δt. The procedure is continued until Δt reaches a pre-established limit. Then a line is fitted to the resulting set of points—“( . . . ) the time versus the log of the Euclidean distance curve is calculated for all neighboring points of their respective trajectories” (Lockhart T E, Liu J Differentiating fall prone and healthy adults using localdynamic stability. Ergonomics. 2008;51:1860-72)—“( . . . ) the propensity of this line approaches the value of the greatest exponent of Lyapunov λ1 (Thielo M Análise e classifição de série temporais não estacionárias utilizando métodos não-lineares. Repositório digital UFRGS. 2000:13-98 http://hdl.handle.net/10183/12661 (accessed 2, Aug. 2018))” and “( . . . ) describes the rate at which the kinematic variability approaches the trajectory of the equilibrium movement or how much the slope curve of maxλ1 diverges from the average curve. The higher the maxλ1, the faster the divergence will increase and the worse the system's resistance to disturbances. Consequently, the higher maxλ1 indicates less local dynamic stability of the human motor control system” (Lockhart T E, Liu J Differentiating Jail prone and healthy adults using localdynamic stability. Ergonomics. 2008;51:1860-72) , that is, risk of falling”.

[0056] In recent clinical findings, time series dispersion measures representing gait cycle projection were considered to be potential for predicting the risk of falling when associated with clinical measures. The maximum Lyapunov exponent in short time series, computed from the linear acceleration of the torso in the medial-lateral direction, was associated with the decline in the rate of stability of falling individuals in a longitudinal study involving 131 elderly people. In this same study, the “pure” measure of)′1 was considered insufficient to predict the senile decline (Bizovska L, Svoboda Z, Janura M, Bisi Me, Vuilletme N Local dynamic stability during gait Jor predicting Jails in elderly people: A one-year prospective study. PLoS ONE. 2018;13 (5): and 0197091. https://doi.org/10.1371/joumal.pone.0197091), as proposed in document US 20110152727 A1, showing the relevance of this proposed method that includes the clinical variables measured in the application as validation for the physical-mathematical measures of risk, providing greater precision for these measures and promoting the interdisciplinary approach required for the prevention of falls.

[0057] FIG. 6 shows the method proposed here based on the integration of clinical data (D) APP (7) (clinical information) and gait and balance algorithms (11) (objective gait and balance indexes), by application and processing of measured vectors of the inertial signal, used on the patient's body (P) for clinical testing—prediction of the level of risk of falling (10) (Rosenstein risk)—performed by a technician in the clinical environment (4) (rehabilitation professional) which refers the patient (P) to the most appropriate treatment. These combined items generate the calculation of the risk of falling, generating a result (R1) that can indicate high risk (15), moderate risk (16), low risk (17) or no risk (18).

[0058] In the clinical data (D) APP (7) table, the patient (P) is subjected to a test 1 (19) based on the clinical data (D) APP (7), generating the evaluation (20) and the results: risk of falling+=3 (21), risk of falling+=2 (22), risk of falling+=1 (23) and risk of falling+=0 (24); then, in the gait and balance algorithms (11), the patient (P) is subjected to a test 2 (25) based on these indicators, generating the “motor parameters” (26) and the results: risk of falling+=3 (21) and risk of falling+=0 (24); in the table prediction of the level of risk of falling (10) (Rosenstein) the patient (P) is submitted to a test 3 (27) that leads to the risk prediction (10B), with the results: risk of falling+=3 (23) and risk of falling+=0 (24).

[0059] After assessing and calculating the risk of falls, the result (R1) generated specifies in the table high risk (15) a value >6 where there are two options: high risk 12-9 (28) and high risk 8-6 (29). The high risk 12-9 (28) has the following procedures as an indication: “preventive physiotherapy+use of continuous wearable apparatus in the daily environment+medication review and eye exam” (30); the high risk 8-6 is indicated by the following procedures: “preventive physiotherapy+motor test in activities of daily living, for one week, at the end of rehabilitation+review of medication and eye exam” (31).

[0060] In the moderate risk table (16), a value between 5-3 is established, indicating the following procedures: “preventive physiotherapy+repeat the test in 3 months and! or revision of medication and! or eye exam” (32).

[0061] In the low risk table (17), a value between 2-1is established, indicating t he following procedures: “walk three times a week+application to control the walk+repeat the test in three months” (33).

[0062] In the risk-free table (18), the value zero is established, indicating the following procedures: “walk three times a week+repeat the test in six months” (34).

[0063] If the indicated prevention includes preventive physical rehabilitation, the same test can be performed in physiotherapy sessions, the result of processing the motor parameters for risk (Rosenstein method) and gait and balance (Fusion Madgwick followed by processing gait and balance indexes), that is, prediction of risk of falling (10) can be used to quantify the clinical practice and evolution of preventive treatment in the clinical environment (4), in real time, at a low cost, determining whether the risk of falling is moderate (16) or low (17), or even if there is an absence (18) of risk of falling. Likewise, if the clinician, in the screening, understands that the risk is very high (15), and in addition to preventive rehabilitation, should include the user's home monitoring, then this user downloads the application on their own cell (2) and of a responsible family member (F) and uses the system continuously (the sensor (1) attached to the body or the cell itself attached to a strap). When the risk is measured in continuous monitoring by algorithm, the apparatus vibrates and beeps and the patient (P) has the opportunity to act before the fall. The family member (F) can also download the same application and, using a specific password, monitor, even from work, the movement of the elderly who stayed at home, monitor the activity performed by the elderly at that moment (lying, sitting, walking) or become aware of a preventive fall or fall alarm for that family member.

[0064] Therefore, according to the present invention, the inventive solution is in the composition of solutions involving method and system, improved with the use of specific signal processing, leading to the clinical environment (4) and everyday environment (5), a low cost solution and easy to understand, measurable in graphs and numbers to the detriment of the current clinical approach based on subjective evaluation, or the current state of the art that contemplates fall prevention only from the motor point of view, and not systemic as recommended in scientific evidence.