EP4432916A1 - Apprentissage machine en vue de la prévention de chute par l?intermédiaire de capteurs de pression - Google Patents

Apprentissage machine en vue de la prévention de chute par l?intermédiaire de capteurs de pression

Info

Publication number
EP4432916A1
EP4432916A1 EP22896686.7A EP22896686A EP4432916A1 EP 4432916 A1 EP4432916 A1 EP 4432916A1 EP 22896686 A EP22896686 A EP 22896686A EP 4432916 A1 EP4432916 A1 EP 4432916A1
Authority
EP
European Patent Office
Prior art keywords
machine learning
fall
user
pressure data
predicted
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
Application number
EP22896686.7A
Other languages
German (de)
English (en)
Inventor
Nate SUKHTIPYAROGE
Vivek Kumar
Adhiraj Ganpat Prajapati
Kedar Mangesh Kadam
Keegan Duane Dsouza
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Matrixcare Inc
MatrixCare Inc
Original Assignee
Matrixcare Inc
MatrixCare Inc
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Matrixcare Inc, MatrixCare Inc filed Critical Matrixcare Inc
Publication of EP4432916A1 publication Critical patent/EP4432916A1/fr
Pending legal-status Critical Current

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • 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
    • A61B5/1036—Measuring load distribution, e.g. podologic studies
    • A61B5/1038—Measuring plantar pressure during gait
    • 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
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
    • A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
    • A61B5/6802—Sensor mounted on worn items
    • A61B5/6804—Garments; Clothes
    • A61B5/6807—Footwear
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
    • 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
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • G06N3/08—Learning methods
    • G06N3/09—Supervised learning

Definitions

  • Embodiments of the present disclosure relate to machine learning. More specifically, embodiments of the present disclosure relate to using machine learning to evaluate pressure sensor data.
  • a method of training machine learning models includes: receiving pressure data corresponding to one or more feet of a user; determining a label indicating whether, within a defined window of time after the pressure data was collected, the user fell; and training one or more machine learning models, based on the pressure data and the label, to predict fall risk.
  • a method of predicting events using machine learning models is provided.
  • the method includes: receiving pressure data corresponding to one or more feet of a user; generating a predicted fall risk by processing the pressure data using one or more trained machine learning models; and upon determining that the predicted fall risk satisfies one or more defined criteria: selecting an intervention for the user; and initiating application of the intervention.
  • FIG. 1 depicts an example environment for using machine learning to evaluate sensor data and predict fall events.
  • FIG. 2 depicts an example environment for using an intermediate communication device and machine learning to evaluate sensor data and predict fall events.
  • FIG. 3 depicts an example workflow for generating labeled data to train machine learning models to predict fall events.
  • FIG. 4 depicts an example floorplan with determined fall risk areas indicated.
  • FIG. 5 is a flow diagram depicting an example method for generating labeled data to train machine learning models to predict falls.
  • FIG. 6 is a flow diagram depicting an example method for training machine learning models based on sensor data to predict fall events.
  • FIG. 7 is a flow diagram depicting an example method for training a likelihood model and a severity model to predict fall events.
  • FIG. 8 is a flow diagram depicting an example method for using trained machine learning models to evaluate sensor data.
  • FIG. 9 is a flow diagram depicting an example method for predicting fall risk and potential severity using machine learning models.
  • FIG. 10 is a flow diagram depicting an example method for predicting fall events and severity using machine learning.
  • FIG. 11 is a flow diagram depicting an example method for initiating preventative and/or remedial actions based on machine learning model output.
  • FIG. 12 is a flow diagram depicting an example method for evaluating and selecting personalized interventions using machine learning.
  • FIG. 13 is a flow diagram depicting an example method for determining aggregated fall risks using machine learning.
  • FIG. 14 is a flow diagram depicting an example method for training one or more machine learning models to predict fall risk.
  • FIG. 15 is a flow diagram depicting an example method for predicting fall risk using machine learning.
  • FIG. 16 depicts an example computing device configured to perform various aspects of the present disclosure.
  • the system evaluates pressure data at discrete instances in time. In other embodiments, the system evaluates the pressure data over windows of time to analyze motion or walking patterns of the user. For example, the system may evaluate the data to determine whether the user is walking normally on one foot while dragging the other somewhat (which may indicate an increased risk of fall).
  • the data is evaluated using one or more machine learning (ML) and/or artificial intelligence (Al) models. These models may enable the system to determine what pressure data is normal and/or acceptable, and what may trigger further concerns.
  • ML machine learning
  • Al artificial intelligence
  • the system uses models to predict the fall severity of such a fall. For example, if the user falls forwards, they may be likely to suffer reduced harm, as compared to falling backwards.
  • the particular pressure data may be used by trained models to predict this severity. In some embodiments, therefore, the pressure data may be used to predict which direction the user will fall.
  • the potential results are also determined or generated (e.g., head trauma from falling backwards, broken wrist from falling forwards, broken hip from falling sideways, and the like).
  • the system uses labeled training data collected from one or more users while the users are engaging in ordinary activities (e.g., walking).
  • data is collected from a group of users over time, and each time a fall occurs or is reported, the sensor data from the fallen user may be labeled as preceding a fall.
  • the system may automatically label the previous N minutes of data as imminently preceding a fall, and/or label the previous M days of data to indicate that a fall followed some days later.
  • the system further labels the data based on the characteristics of the fall, such as the fall direction, fall severity, and the like.
  • the training data can further indicate various characteristics of the user, such as their age, gender, weight, height, blood pressure, pulse rate, any mobility restrictions, whether the user uses any assistive devices such as a can or walker, and the like.
  • characteristics of the user such as their age, gender, weight, height, blood pressure, pulse rate, any mobility restrictions, whether the user uses any assistive devices such as a can or walker, and the like.
  • these predictions can be used to drive a wide variety of patientspecific interventions.
  • the system may respond to an immediate risk by dispatching aid (e.g., alerting a nearby caregiver), instructing the user to stop, sit, or otherwise rest, and the like.
  • dispatching aid e.g., alerting a nearby caregiver
  • the system can additionally or alternatively initiate various preventative actions, such as modifying physical therapy plans (e.g., to strengthen the muscles of the patient that will help to prevent the particular type of fall that is predicted), assigning or providing assistive devices such as a cane or walker, and the like.
  • the pressure data includes multiple individual pressure points on each foot of each user.
  • the system may receive and evaluate, for each foot, the pressure at the front-left point of the foot, the front-right point of the foot, the rear-left point of the foot, the rear-right point of the foot, the center of the foot, the front-center of the foot, the rearcenter of the food, the center-left of the foot, the center-right of the foot, and the like.
  • the models can be used for diagnostic purposes, such as to detect when a user is putting extra pressure on one side (which may indicate pain or other issues on the other side).
  • the models can be used to not only predict imminent falls (e.g., falls that may occur in the next few seconds, minutes, or days), but also to evaluate more long-term trends (e.g., to determine whether the user is losing their balance or wavering more frequently now than they did before, even in the absence of an actual fall).
  • imminent falls e.g., falls that may occur in the next few seconds, minutes, or days
  • long-term trends e.g., to determine whether the user is losing their balance or wavering more frequently now than they did before, even in the absence of an actual fall.
  • the predictive models can also be used to identify and remediate potential hazards in the physical space, such as rugs, uneven floors, cords across the walkway, and the like.
  • the system may identify region(s) in a facility where fall events are more common, where near-fall events occur (e.g., detected stumbling, wavering, or tripping), where predicted falls are often flagged, and the like.
  • the system can generate a heat map for the facility, indicating the average fall risk and/or predicted severity across the floorplan, and across a number of users and/or times. By evaluating this data, the system can identify potential hazards that are causing (or may cause) falls, and thereby initiate remediation that can prevent such occurrences.
  • FIG. 1 depicts an example environment 100 for using machine learning to evaluate sensor data and predict fall events.
  • the pressure sensors 105 transmit or otherwise provide pressure data 110 to a machine learning system 115.
  • this pressure data 110 may be provided using any suitable technology, including wired or wireless communications.
  • the pressure sensors 105 can use cellular communication technology to transmit the pressure data 110 to the machine learning system 115.
  • the pressure sensors 105 use one or more local wireless networks (such as a WiFi network, a Near Field Communication (NFC) network, a Bluetooth network, etc.) to transmit the pressure data 110.
  • the pressure sensors 105 can transmit the pressure data 110 to one or more intermediary devices, which can then forward the data to the machine learning system 115.
  • the pressure sensors 105 may transmit the pressure data 110 to a smartphone, tablet, or other device associated with the user (e.g., via Bluetooth), and this user device can forward the data to the machine learning system 115 (e.g., via WiFi or a cellular connection).
  • a smartphone e.g., via Bluetooth
  • the machine learning system 115 e.g., via WiFi or a cellular connection
  • the pressure data 110 is transmitted over one or more networks including the Internet. That is, the machine learning system 115 may reside at a location remote from the user and pressure sensors 105 (e.g., in the cloud). Though a single set of pressure sensors 105 is illustrated for conceptual clarity, in embodiments, data from any number and variety of pressure sensors 105 (and any number of users) can similarly be provided to the machine learning system 115.
  • the machine learning system 115 can generally train one or more machine learning models to analyze the pressure data 110, and/or use trained models to evaluate the pressure data 110.
  • the machine learning system 115 can also optionally receive a set of user records 112.
  • the user records 112 can include information relating to various characteristics of the users associated with the pressure sensors 105 (e.g., the residents in a residential facility).
  • the user records 112 may indicate, for each user, which pressure sensor(s) 105 are associated with the given user, as well as characteristics such as their age, height, weight, whether they use an assistive device such as a cane or walker, whether they have suffered a fall (and if so, details relating to the fall, such as the time, severity, direction, cause, etc.), and the like.
  • the user record(s) 112 are used as input, alongside the pressure data 110, when training the model(s), as well as when using them to predict fall events. That is, the pressure data 110 from a user, along with one or more of the characteristics indicated in the corresponding user record(s) 112 for the user, may be used as input to the model(s) to generate predicted fall characteristics (e.g., timing, severity, and the like). During inferencing, this output can be used to drive a variety of interventions, as discussed in more detail below. During training, this output can be compared against a known label (e.g., indicating whether the user fell) in order to compute a loss used to refine the model(s).
  • a known label e.g., indicating whether the user fell
  • the loss generally corresponds to the difference between the actual model output and the target output.
  • the loss may be computed using a variety of algorithms or techniques, including cross-entropy loss, mean absolute error, and the like.
  • the magnitude of the difference is directly correlated with the magnitude of the loss.
  • the loss is used to refine the internal parameters of the model(s), such as the weights and/or biases.
  • the magnitude of the loss is directly correlated with the magnitude of the change needed for these parameters.
  • the machine learning system 115 can use machine learning to predict a variety of fall-related events, such as a likelihood of a future fall (which may include an imminent fall, such as due to stumbling, as well as falls further into the future, such as due to degrading balance over time), a potential severity of the future fall, and the like.
  • a separate machine learning model is trained for each such output. This may allow the model(s) to specialize for their particular target output. For example, predicting whether a fall will occur may involve different features (or differently-weighted features), as compared to predicting when the fall will occur and/or how severe the fall will be.
  • the machine learning system 115 can collect pressure data 110 and/or user record(s) 112 over time from any number of users.
  • each user in a facility e.g., a long term care unit
  • can have an associated set of pressure sensors 105 e.g., a pair of sensor-equipped insoles or shoes
  • the machine learning system 115 or another component can monitor the environment to tag this pressure data based on a variety of factors, including the characteristics of the corresponding user (e.g., age, weight, height, and the like).
  • the data is associated with a “ground-truth” label based on whether the corresponding user suffered a fall. For example, each time a participating user falls (e.g., as indicated in the user records 112), the machine learning system 115 (or another system) can retrieve the corresponding records of pressure data 110 from that user for one or more prior times (e.g., for N seconds, minutes, hours, or days prior to the fall), and label this data as indicative that a fall is imminent.
  • the label further indicates how much time elapsed between the sensor reading and the fall. For example, a first record may be labeled to indicate that the user fell five minutes after the data was recorded, while a second record indicates that the user fell thirty seconds later.
  • the pressure data 110 may additionally or alternatively be labeled to indicate various aspects of the fall, such as the direction of the fall (e.g., to the left, to the right, forward, backwards, and the like). Similarly, the data may be labeled to indicate the severity of the fall (e.g., whether it resulted in any broken bones or other injuries, the extent of these injuries, and the like). In at least one embodiment, the data can additionally or alternative be tagged using contextual data that may be relevant in predicting similar falls. For example, using video or images of the fall (captured by one or more cameras), the system may discern whether the patient tripped on an object, whether the user was actually using any assistive device such as a cane, and the like. This information can allow the system to generate more robust training data that more accurately accounts for the particular context of the fall.
  • the machine learning system 115 By collecting such data over time and from a variety of users, the machine learning system 115 is able to automatically build a training data set that can be used to train or refine a variety of machine learning architectures to predict future falls, how far into the future the fall will occur, which direction the user will fall, how severe the fall may be, and the like. [0054] In at least one embodiment, if a given user does not fall (or does not fall within some defined window), the machine learning system 115 can automatically label the corresponding records as not indicative of an upcoming fall.
  • the machine learning system 115 may continuously or periodically identify unlabeled pressure records (e.g., records that have not already been labeled as preceding a fall) that are older than a user-defined threshold (e.g., older than a week), and label these records as not indicative of a fall (if no fall is reported in the user records 112). In this way, the machine learning system 115 can generate both positive training samples (e.g., pressure data 110 and/or user characteristics that indicate that a fall is imminent) as well as negative training data (e.g., pressure data 110 and/or user characteristics that indicate that the user is steady and/or a fall is not imminent).
  • a user-defined threshold e.g., older than a week
  • the machine learning system 115 can train a number of machine learning models to address a variety of desired predictions. For example, a first model may be trained to predict if a user will fall, while a second is used to predict how far into the future the fall will occur. In some aspects, a first model may be trained to predict imminent falls (e.g., within the next M seconds, based on a narrow window of input time), while others are trained using progressively longer intervals for more distant predictions (e.g., evaluating data collected over an hour to determine whether a fall is likely in the next hour). Other models may be used to predict fall direction, fall severity, and the like.
  • training the machine learning models includes providing some set of pressure data 110 from one or more users as input to the model (e.g., pressure data covering a defined window of time) to generate some predicted output.
  • the input also includes one or more characteristics of the user (such as age, whether they use an assistive device, and the like), as discussed above.
  • this output may be relatively random or unreliable (e.g., due to the random weights and/or biases used to initiate the model).
  • the generated output can then be compared against the relevant label for which the model is being trained (e.g., whether a fall occurred, how severe the fall was, and the like) to generate a loss, and the loss can be used to refine the model (e.g., using back propagation in the case of a neural network).
  • this refinement process may be performed individually for each user or window of time (e.g., using stochastic gradient descent) or in batches (e.g., using batch gradient descent).
  • the machine learning system 115 can train models to operate on windows of data (e.g., a collection of discrete or continuous pressure values) or based on single records (e.g., pressure data at a given moment in time).
  • the machine learning system 115 can deploy them for use in real-time.
  • the models may be trained on one or more systems and deployed to one or more other systems.
  • the machine learning system 115 can both train the models and use them for inferencing.
  • the machine learning system 115 receives pressure data 110 in real-time (or near real-time) from participating users. In at least one embodiment, the machine learning system 115 also receives user record(s) 112 for the specific user.
  • inferencing refers to the stage of model deployment where the model(s) have been trained and are deployed for use in evaluating actual input during runtime. As the model output may be referred to in some embodiments as “inferences,” this stage may be referred to as “inferencing.”
  • the machine learning system 115 can process this pressure data (along with, in some embodiments, user characteristics such as age) using one or more of the trained models in order to predict potential falls.
  • the machine learning system 115 can process the data sequentially using the models, rather than using all models in parallel. This may significantly reduce computational expense of evaluating the pressure data 110, thereby reducing the needed memory space and computational power, as well as reducing energy consumption and latency (e.g., if the system becomes overloaded from a large number of users).
  • the machine learning system 115 may first use an initial model (e.g., to determine whether a fall is imminent). If a fall is determined to be imminent, the machine learning system 115 may immediately initiate remedial or preventative actions, as discussed in more detail below. The fall severity, direction, and other characteristics of the potential fall may be predicted later, or may be bypassed entirely (e.g., as irrelevant once the fall occurs or does not occur). In one such embodiment, if a fall is not imminent but may occur at some point in the future, the machine learning system 115 can selectively use one or more other models to predict the nature of the fall, such as when it will occur, the direction the user will fall, the severity, and the like.
  • an initial model e.g., to determine whether a fall is imminent. If a fall is determined to be imminent, the machine learning system 115 may immediately initiate remedial or preventative actions, as discussed in more detail below. The fall severity, direction, and other characteristics of the potential fall may be predicted later, or may be bypassed entirely (e.g., as
  • the machine learning system 115 can refrain from initiating any further action. However, in the illustrated example, if a fall is predicted, the machine learning system 115 can initiate a wide variety of remedial and/or preventative actions. For example, if a fall is imminent (e.g., within a defined number of seconds or minutes), the machine learning system 115 may initiate actions such as instructing the user to sit or grasp an object for support, alerting nearby caregivers to attend to the user immediately, and the like.
  • the machine learning system 115 can initiate other remedial actions, such as suggesting or prescribing medical devices including canes, walkers, and wheelchairs.
  • the machine learning system 115 can suggest or prescribe one or more physical therapy plans or changes based on the prediction. For example, if the machine learning system 115 predicts that the user may fall towards their left, the machine learning system 115 may determine that physical therapy is needed to strengthen the user’s muscles on the left side of their body (or to otherwise correct any pains or other concerns on this side). In response, the machine learning system 115 can suggest or initiate physical therapy to correct these concerns.
  • the machine learning system 115 can additionally or alternatively provide facility -wide remedial or preventative actions 120. For example, if a number of falls (or predicted falls) occur in a given area (e.g., if the machine learning system 115 observes that users hike in a given place in the facility, such that predicted imminent falls increase on average in this place), the machine learning system 115 may flag this area for manual review. This may indicate, for example, that there is a hazard in the area (e.g., a rug pulling up or a cable across the floor).
  • a hazard in the area e.g., a rug pulling up or a cable across the floor.
  • the machine learning system 115 is able to dramatically reduce falls (and fall severity), thereby improving user results. Further, by iteratively using some machine learning models only when the output from a prior model satisfies certain criteria, the machine learning system 115 can provide this granular and specific analysis using reduced computational resources and expense.
  • FIG. 2 depicts an example environment 200 for using an intermediate communication device and machine learning to evaluate sensor data and predict fall events.
  • a set of pressure sensors 105 are configured to record pressure data from the feet of a user.
  • the pressure sensors 105 may be included in wearable devices, such as included in insoles that are inserted into the user’s shoes, or integrated into the shoes themselves.
  • the pressure sensors 105 are configured to transmit or otherwise provide pressure data 210 to a user device 215, which can then forward the pressure data 210 (and, in some embodiments, other relevant data) to the machine learning system 115. That is, while in the environment 100, the pressure sensors 105 may transmit the pressure data directly to the machine learning system 115, the environment 200 uses a user device 215 of the user.
  • the user device 215 is a smartphone. However, in embodiments, other devices may be used. Generally, the user device 215 corresponds to a portable computing device associated with the user (such as a phone, tablet, smart watch, and the like). The user (or an assisting person, such as a caregiver) may use the user device 215 to link the pressure sensors 105 to the corresponding user and user device 215. For example, the user may use short- range wireless technology such as Bluetooth to pair the user device 215 and the pressure sensors 105. The pressure sensors 105 can then transmit pressure data 210 to the user device 215, which can optionally tag it based on user characteristics such as the identity of the user, the user’s age, and the like.
  • a portable computing device associated with the user such as a phone, tablet, smart watch, and the like.
  • the user may use the user device 215 to link the pressure sensors 105 to the corresponding user and user device 215.
  • the user may use short- range wireless technology such as Bluetooth to pair the user device 215 and the pressure sensors 105
  • the user device 215 then forwards the pressure data 210 to the machine learning system 115 via a network 220 (e.g., the Internet).
  • the user records 112 are separately provided to the machine learning system 115 via the network 220.
  • the user records 112 are alternatively or additionally provided by the user device 215.
  • the user device 215 may tag the pressure data 210 with a unique identifier of the user, and the machine learning system 115 may use this identifier to retrieve or determine the corresponding user record(s) 112 containing information relevant to the user.
  • the user device 215 may directly include some or all of this information, such as the user’s age, height, weight, and the like.
  • the machine learning system 115 can generally train one or more machine learning models to analyze the pressure data 210 and user records 112, and/or use trained models to evaluate the pressure data 110 and user records 112.
  • the pressure data 210 and user records 112 may be used as input to generate a predicted output (e.g., a fall likelihood or a fall severity), which can be evaluated against the ground truth label (e.g., determined from the user records 112) to refine the model(s).
  • the machine learning system 115 may process the pressure data 210 and user records 112 using the model(s) to predict whether a fall is likely for each given user, the potential severity of the fall, and the like.
  • FIG. 3 depicts an example workflow 300 for generating labeled data to train machine learning models to predict fall events.
  • the machine learning system 115 receives data from pressure sensor(s) 105, as discussed above.
  • one or more pressure sensors 105 may transmit pressure data continuously, periodically (e.g., every second, every five seconds, etc.), and the like.
  • the machine learning system 115 can tag it or otherwise associate it with the corresponding user.
  • the machine learning system 115 can identify the corresponding user record(s) 305 (e.g., based on an identifier included in the pressure data), such that the pressure data can be associated with relevant user characteristics that are used as input to the models, such as age, weight, and the like.
  • the received pressure data already includes such characteristics.
  • the machine learning system 115 may store this (tagged) pressure data for future labeling, as discussed below in more detail.
  • the machine learning system 115 can evaluate the user records 305 periodically, continuously, or upon specified events in order to label the pressure data. For example, the machine learning system 115 may periodically evaluate the user record(s) 305 for the registered users (e.g., daily or weekly) in order to determine whether any of the user(s) suffered a fall.
  • the machine learning system 115 can receive fall information automatically (e.g., as a pushed alert or notification, rather than by actively checking the records for fall reports).
  • the machine learning system 115 can retrieve one or more corresponding records of pressure data, and label it to indicate the fall. For example, as discussed above, the machine learning system 115 can label each pressure record to indicate information such as the fact that a fall occurred, how far into the future (relative to the record) the fall occurred, the direction of the fall, the severity of the fall, and the like. In some embodiments, the machine learning system 115 can label a sequence of pressure records in this way. For example, one record may be labeled to indicate that it immediately preceded a fall (e.g., within a defined period of time), while a second is labeled to indicate that a fall occurred one hour after the pressure data was recorded. This can allow the machine learning system 115 to train a set of model(s) to predict when the fall will occur.
  • the machine learning system 115 can then store the pressure data as labeled training data 310.
  • this labeled training data 310 can be used to train and/or refine one or more machine learning models. In various embodiments, this training may be performed periodically (e.g., weekly), or upon specified events. For example, in at least one embodiment, if a fall occurs, the machine learning system 115 can determine whether the model(s) predicted the fall (or would have predicted it, if they were not used to evaluate the pressure data). If the model(s) accurately predicted the fall and/or fall characteristics, the machine learning system 115 can label and store the data for subsequent refinement.
  • the machine learning system 115 may immediately use the data (and any other labeled training data 310 that has not-yet been used to train the model(s)) to refine the model(s) for more accurate predictions.
  • the machine learning system 115 in addition to positive labeled training data 310 (e.g., samples that preceded a fall), can also generate labeled training data 310 corresponding to negative examples. For example, the machine learning system 115 may periodically or continuously identify pressure data that is older than some defined threshold (e.g., more than a week old). If the user records 305 do not indicate that the user fell in that defined window, the machine learning system 115 may automatically label the pressure data as not indicative of an upcoming fall. These negative samples can then be stored as labeled training data 310 for future training and/or refinement, as discussed above.
  • some defined threshold e.g., more than a week old
  • these negative exemplars may also be used periodically (e.g., weekly), or upon specified events. For example, in at least one embodiment, if the model(s) predict a fall (or predict specific fall characteristics) but one does not occur (or the fall characteristics differ), the machine learning system 115 may immediately use the data (and any other labeled training data 310 that has not-yet been used to train the model(s)) to refine the model(s) for more accurate predictions.
  • the machine learning system 115 is able to automatically generate labeled training data 310 for training and refining of the machine learning model(s).
  • This automated label generation is efficient and rapid, and further can generally result in labeled training data 310 with reduced error (e.g., mislabeled data).
  • this automated workflow 300 can enable training more accurate models than would be possible using conventional systems (e.g., that rely on manually-curated data). These more accurate models, in turn, can be used for significantly improved fall prediction and intervention.
  • FIG. 4 depicts an example floorplan 400 with determined fall risk areas indicated.
  • the floorplan 400 may be generated and/or augmented by a machine learning system, such as the machine learning system 115 of FIG. 1.
  • the machine learning system (or another system or device) can aggregate the fall predictions and/or events in order to provide deeper insights regarding potential fall hazards in one or more physical spaces or locations.
  • the floorplan 400 corresponds to a residential facility (e.g., a nursing home or long term care facility) where users reside.
  • a residential facility e.g., a nursing home or long term care facility
  • the techniques described herein can be readily applied to a wide variety of locations and facilities.
  • the machine learning system can evaluate reported fall(s) and/or predicted fall(s) for the facility. For example, for each reported (actual) fall, the machine learning system can determine where, in the facility, the fall occurred. This data may be used to generate augmented floorplans (e.g., using a heat map) indicating area(s) where falls occur more frequently.
  • augmented floorplans e.g., using a heat map
  • the machine learning system can additionally or alternatively aggregate predicted fall data to indicate problem areas. For example, each time the machine learning model(s) predict that a fall is imminent (e.g., due to a user stumbling), the machine learning system can determine where the predicted fall occurred (even if no actual fall follows). By similarly aggregating these predicted falls across the facility, the machine learning system can identify area(s) where falls are frequently predicted or close to occurring, even if no fall occurs. This data may similarly be output, such as via an augmented floorplan (e.g., using a heat map) to indicate these problem areas.
  • an augmented floorplan e.g., using a heat map
  • an area 405A is indicated as potentially problematic (e.g., due to the number or frequency of actual falls or predicted falls in that area).
  • the particular technique used to indicate the areas 405 may vary depending on the particular implementation.
  • the machine learning system may generate a heat map, to be output via a graphical user interface (GUI).
  • GUI graphical user interface
  • the machine learning system can identify area(s) where the actual and/or predicted falls meet one or more defined criteria (e.g., a number of reported or predicted events, a frequency of the events, and the like), and indicate these areas specifically, such as by highlighting them, outlining them, and the like.
  • the machine learning system has also flagged the area 405C as potentially hazardous.
  • this area 405C is within a room assigned to a specific user. That is, while the areas 405A and 405B are in public regions that may be generally accessible to the users, the area 405C may be in a private room where one or more residents generally reside (e.g., where they sleep and store their items).
  • the indication of this area may generally allow a user or manager to investigate the area 405C to identify potential hazards (such as furniture or other objects, cords, rugs, and the like).
  • the machine learning system or user may infer that the particular user(s) associated with the area 405C may need assistance. That is, in addition to or rather than determining that the area itself is hazardous, the machine learning system may determine or infer that the corresponding user who lives in the room is frequently having difficulties with balance, such as when standing up from a chair or getting out of bed. In this way, the machine learning system can take preventative actions for the specific user(s), such as suggesting or instructing that a caregiver be present to assist in standing or sitting, or suggesting or assigning physical therapy tasks, as discussed above.
  • the machine learning system can derive deeper insights regarding the facility and its users, and thereby significantly improve the safety of the facility and reduce potential harm.
  • FIG. 5 is a flow diagram depicting an example method 500 for generating labeled data to train machine learning models to predict falls.
  • the method 500 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 500 provides additional detail for the workflow 300 of FIG. 3.
  • the machine learning system receives pressure data from one or more pressure sensors (e.g., pressure sensors 105 of FIG. 1).
  • the machine learning system may be configured to receive pressure data continuously or periodically from one or more pressure sensors associated with one or more users.
  • the pressure data can also include or be associated with various user data, such as a user identifier, characteristics of the user, and the like.
  • the machine learning system retrieves one or more corresponding user record(s) for the user associated with the received pressure data. For example, based on a user ID included with the pressure data, the machine learning system can query one or more repositories to retrieve user records for the user, where these records can indicate various characteristics of the user (e.g., their age, height, weight, whether they use assistive devices, and the like).
  • the user records can additionally or alternatively indicate events, such as falls, that occurred. For example, each time a user falls (or, in some embodiments, almost falls), the user record(s) may be updated (e.g., by a caregiver) to indicate the event.
  • the machine learning system performs block 510 when the pressure data is received. For example, if the pressure data is from one or more prior times (e.g., the data was not provided live, immediately after being recorded), the machine learning system may immediately retrieve the corresponding records. In some embodiments, if the pressure data was received live (e.g., shortly after being recorded), the machine learning system may store the data and subsequently retrieve the user records upon occurrence of one or more defined criteria (e.g., after a defined period of time has passed, upon being informed that a fall occurred and the like).
  • the machine learning system evaluates the retrieved user records to determine whether the user suffered a fall. If not, the method 500 continues to block 525, where the machine learning system labels the received pressure data. That is, the machine learning system can determine that the user did not fall (during the relevant window after the pressure data was recorded), and label the data accordingly. In some embodiments, as discussed above, the machine learning system may wait until defined criteria are met (e.g., a minimum period of time after the pressure data is recorded) before determining that a fall did not occur. [0095] In at least one embodiment, in addition to determining whether the user fell, the machine learning system may determine whether the user lowered. Though hikes may generally be less dangerous than actual falls, they may nevertheless be indicative of potential concerns or problems. By labeling the data to predict stumbles, the system may be able to improve the overall results for the users.
  • defined criteria e.g., a minimum period of time after the pressure data is recorded
  • the method 500 continues to block 520.
  • the machine learning system determines one or more characteristics of the fall, such as the direction of the fall (e.g., forwards, backwards, or sideways), the location of the fall (e.g., the physical area in a facility), the severity of the fall (e.g., whether it resulted in lacerations, broken bone(s), sprains, etc.), and the like. In an embodiments, as discussed above, these characteristics are indicated in the user record(s) corresponding to the fall.
  • these characteristics are stored in a machine-readable format, such as using defined values or labels to indicate directionality and/or severity of the fall.
  • the machine learning system may evaluate written records (e.g., written by a caregiver at the time of the fall), such as by using natural language processing, to determine these fall characteristics.
  • the method 500 then continues to block 525, where the machine learning system labels the received pressure data to indicate the presence of the fall and/or the fall characteristics. As discussed above, this labeled data can then be used to train or refine one or more machine learning models to predict fall events. The method 500 then returns to block 505 to receive additional pressure data.
  • the machine learning system can automatically generate labeled training data. As discussed above, this process can significantly improve the accuracy and reliability of the resulting machine learning models, as well as significantly reduce the cost and inaccuracies involved in manually-labeling data.
  • FIG. 6 is a flow diagram depicting an example method 600 for training machine learning models based on sensor data to predict fall events.
  • the method 600 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the machine learning system receives pressure data from one or more users.
  • this pressure data can generally include the pressure exerted by the user’s foot or feet at one or more points on the foot (or feet) while the user stands, walks, runs, and the like.
  • the machine learning system can collect this data from a number of users (e.g., all patients in a long-term care facility) over a period of time (e.g., over the course of a year).
  • the machine learning system can further receive biographical data of each user, such as their age, weight, height, mobility issues, and the like.
  • the machine learning system determines a set of label(s) for the received data. For example, as discussed above, the machine learning system (or another system) may identify falls (e.g., recorded or reported in the user’s patient data). The machine learning system (or other system) may then retrieve the corresponding pressure data records from one or more times prior to the fall, and label the data accordingly. For example, the machine learning system can retrieve records from a window of thirty seconds prior to the fall and label then as “fall imminent,” retrieve the records beginning five minutes prior and label them as “fall upcoming,” and so on.
  • falls e.g., recorded or reported in the user’s patient data
  • the machine learning system may then retrieve the corresponding pressure data records from one or more times prior to the fall, and label the data accordingly. For example, the machine learning system can retrieve records from a window of thirty seconds prior to the fall and label then as “fall imminent,” retrieve the records beginning five minutes prior and label them as “fall upcoming,” and so on.
  • the machine learning system can additionally or alternatively label the records based on other characteristics of the fall, such as the direction of the fall, severity of the fall, and the like. Additionally, in some embodiments, if a sufficient period of time has elapsed without the user falling (e.g., if the record is more than a day old, more than a week old, and the like), the machine learning system can label these records as not indicative of an upcoming fall.
  • a sufficient period of time has elapsed without the user falling (e.g., if the record is more than a day old, more than a week old, and the like)
  • the machine learning system can label these records as not indicative of an upcoming fall.
  • One example of determining the pressure data labels is described above in more detail with reference to FIG. 5.
  • the method 600 continues to block 615, where the machine learning system trains one or more machine learning models based on the labeled data, as discussed above. Though the illustrated example depicts data collection, labeling, and training occurring sequentially for conceptual clarity, in some embodiments, the machine learning system may first collect and label data for some period of time, and move to model training after that period has elapsed (or after sufficient data has been collected). [0104] In an embodiment, as discussed above, training the models generally includes providing the pressure data (as well as user characteristics or other patient data) as input to the model, and using the label to compute a loss (based on the generated output). In this way, the machine learning system can iteratively refine the model(s) based on the labeled data in order to improve their predictions.
  • the machine learning system determines whether training is complete. In embodiments, this determination may be made based on a variety of termination criteria. For example, the machine learning system may determine whether a defined number of training cycles have been completed, whether a minimum number of training records have been used, whether a defined period of time has elapsed while training, whether a minimum preferred accuracy has been met (e.g., determined using labeled data as test data), and the like. If training is not complete, the method 600 returns to block 605 to receive a new set of pressure data.
  • the method 600 continues to block 625, where the machine learning system deploys the trained model(s) for inferencing. As discussed above, this may include using the models locally (on the machine learning system) to process new data, providing the models to one or more other systems or devices to predict falls, and the like.
  • FIG. 7 is a flow diagram depicting an example method 700 for training a likelihood model and a severity model to predict fall events.
  • the method 700 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 700 provides additional detail for block 615 of FIG. 6.
  • the machine learning system trains a likelihood model based on the pressure data.
  • the likelihood model is generally a machine learning model that learns to predict the likelihood of a user falling within some defined period.
  • the machine learning system predicts whether the user will fall within a fixed period. For example, a first model may predict imminent falls (e.g., within thirty seconds) while a second predicts upcoming falls (e.g., within a few hours or days), and so on.
  • a single model predicts both the likelihood or probability of a fall, as well as the expected time or delay until the fall.
  • training the likelihood model includes processing the pressure data (and, in some embodiments, various user characteristics such as age and whether the user uses a cane, walker, or other assistive device) with the likelihood model in order to generate an output fall probability or prediction.
  • This output can then be compared against the ground truth label for the pressure data (e.g., indicating whether a fall actually occurred, and if so, when).
  • This comparison can be used to compute a loss (e.g., a cross-entropy loss) based on the difference(s) between the prediction and the label, and the loss may be used to refine the likelihood model (e.g., using backpropagation).
  • stochastic gradient descent e.g., refining the model individually for each individual exemplar
  • the machine learning system may refine the likelihood model using batch gradient descent (based on a set of exemplars).
  • the machine learning system determines whether the label of the pressure data indicates that a fall occurred. If not, the method 600 terminates, and returns to block 620 of FIG. 6. That is, if the pressure data is not related to a fall, the machine learning system does not train or refine any other model(s), and the (negative) exemplar is used only to train the likelihood model.
  • the method 700 continues to block 725, where the machine learning system trains a severity model based on the data.
  • the severity model is a machine learning model that learns to predict the severity of a fall for a given patient, presuming a fall will occur.
  • the machine learning system uses only user data (e.g., age, weight, height, assistive devices, and the like) to refine the severity model.
  • the machine learning system uses pressure data for the user, along with the user characteristics, to train the severity model.
  • the severity model may be trained to receive the pressure data (and, in some embodiments, additional prior pressure data that may not have been used to train the likelihood model), along with one or more user characteristics, to generate a prediction as to how severe a fall would be, if one occurs.
  • this prediction includes consideration of the fall direction (e.g., forwards or backwards).
  • the severity model may be trained to predict which direction the user will fall, and this direction can be used to predict fall severity.
  • the fall direction can be implicitly learned internally by the model, and then used to predict the severity. That is, in some embodiments, the predicted severity is dependent on the predicted fall direction, regardless of whether the severity model itself outputs a predicted direction of the fall.
  • the severity model in a similar manner to the likelihood model, can be trained by generating a predicted severity using the relevant input data, and comparing the predicted severity with a ground-truth severity (indicated by the label of the pressure data). This difference can be used to compute a loss, which may be used to refine the severity model (e.g., using stochastic gradient descent and/or batch gradient descent). The method 700 then terminates and returns to block 620 of FIG. 6.
  • FIG. 8 is a flow diagram depicting an example method 800 for using trained machine learning models to evaluate sensor data.
  • the method 800 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 800 is performed using the machine learning models trained using the method 600 of FIG. 6
  • the machine learning system receives pressure data.
  • this pressure data can generally include the pressure exerted by the user’s foot or feet at one or more points on the foot (or feet) while the user stands, walks, runs, and the like.
  • the pressure data received at block 805 corresponds to data from a single user at a single point in time or over a window of time (e.g., over a thirty-second window).
  • the machine learning system can further receive biographical data of the user as discussed above, such as their age, weight, height, mobility issues, and the like.
  • the machine learning system processes the received data (which may include current pressure data as well as biographical data of the user) using one or more trained machine learning models, as discussed above. In some embodiments, this includes processing the data using all of the available models in parallel. For example, the machine learning system may use one model to predict whether a fall will occur imminently (e.g., within a defined (short) period of time), another model to predict whether a fall will occur in the future but not imminently, another model to predict the directionality and/or severity of the fall, and the like. [0117] In some embodiments, as discussed above, the machine learning system may use these models sequentially, only using some models if the output from one or more prior model(s) meets defined criteria. For example, the machine learning system may only use the fall direction and/or severity model(s) if the fall occurrence model(s) indicate that a future fall is likely.
  • the machine learning system determines whether one or more defined criteria are satisfied.
  • the criteria include determining whether a future fall is predicted with a sufficient level of confidence.
  • the criteria includes determining whether the predicted fall severity exceeds some defined thresholds (even if the fall likelihood does not). That is, the machine learning system may consider whether the fall would be particularly severe, even if the fall is not likely to occur. If the criteria are not satisfied, the method 800 returns to block 805 to receive new pressure data for evaluation. If one or more of the criteria are satisfied, the method 800 continues to block 820.
  • the machine learning system generates and/or initiates one or more remedial and/or preventative actions based on the generated predictions. For example, if the machine learning system determines that a fall is imminent, caregivers can be alerted and the patient can be instructed to sit or rest. If the predictions indicate that a fall is likely in the next few hours or days, the machine learning system may prescribe medical devices such as canes or walkers. If the fall is predicted to be severe (even if unlikely), more forward-looking interventions may include instructions like added physical therapy for the user.
  • the particular preventative actions selected may be determined based at least in part using the models themselves. For example, to determine whether to suggest a cane, walker, improved physical therapy, and the like, the machine learning system may modify one or more of the user’s characteristics, and use this modified data as input to the model to determine whether the predicted fall risk or likelihood has decreased. This can enable the machine learning system to not only determine whether intervention is needed, but also to identify the best intervention that is most likely to reduce the fall risk (or fall severity). One example of this predictive process is discussed below in more detail with reference to FIG. 12.
  • the machine learning system can also aggregate predicted falls and/or severities to provide facility-wide suggestions, such as to identify potential hazards like cords in a walkway, uneven ground, and the like. These suggestions can be used to improve the facility overall.
  • FIG. 9 is a flow diagram depicting an example method 900 for predicting fall risk and potential severity using machine learning models.
  • the method 900 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 900 provides additional detail for block 810 of FIG. 8.
  • the machine learning system generates a fall likelihood score by processing the user data (which may include pressure data, as well as biographical data) using a first machine learning model.
  • the likelihood score indicates a probability or likelihood that the user will suffer a fall at some point in the future.
  • the timeline of this prediction may vary.
  • the first model may be trained to predict whether a fall is likely within the next five minutes or five days.
  • the machine learning system uses a set of models, each configured to predict fall likelihood over a corresponding window of future time.
  • the machine learning system may selectively use these models depending on the output of the prior models. For example, the machine learning system may use a model that predicts falls over a five minute window only if a model predicting falls in the next thirty seconds does not indicate that a fall is likely. Similarly, the machine learning system may only use a model that predicts falls over the next five hours only if the five-minute model does not indicate that a fall is likely.
  • the machine learning system generates a severity score by processing the user data (e.g., pressure data and biographical data) using a second machine learning model.
  • the severity score indicates a probable, likely, or possible severity of a future fall (e.g., in terms of the types of injuries that are likely, the severity or intensity of those injuries, and the like).
  • the severity is based in part on a predicted fall direction (which may be predicted using a separate model, or using the second machine learning model).
  • the models include user data such as age, weight, and height as input, they inherently learn to adjust the predicted severity based on these contributing factors.
  • the machine learning system may selectively use this severity model depending on the output of the risk model(s). For example, if the likelihood model(s) do not indicate that a fall is likely, the machine learning system may refrain from using the severity model in order to reduce computational expense and latency. Similarly, in one embodiment, if the likelihood model(s) indicate that a fall is imminent (e.g., within thirty seconds), the machine learning system may refrain from using the severity model(s). In contrast, if the likelihood model(s) predict that a fall is likely, but not imminent, the machine learning system can use the severity model(s) to enable improved selection of the available targeted interventions.
  • the machine learning system may use the severity model regardless of the output of the likelihood model(s) (e.g., to allow for interventions with users that, though generally stable, could suffer catastrophic injury in the event of a fall). Additionally, in at least one embodiment, the likelihood model(s) and severity model(s) may collectively be referred to as fall risk models.
  • FIG. 10 is a flow diagram depicting an example method 1000 for predicting fall events and severity using machine learning.
  • the method 1000 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 1000 provides additional detail for block 810 of FIG. 8.
  • the method 1000 differs from the method 900 in that, using the method 1000, the machine learning system can determine whether to apply various models based on the output of prior models.
  • the machine learning system generates a likelihood score for the user using a first machine learning model. For example, as discussed above, the machine learning system may process the user’s pressure data (e.g., from a thirty second window) and/or the user’s characteristics or biographical data using one or more likelihood models to determine whether there is a likelihood that the user will fall at some point in the future (e.g.,. within one or more defined windows of time). [0129] At block 1010, the machine learning system determines whether the predicted fall probability or likelihood satisfies some defined threshold. For example, the machine learning system may determine whether a fall is predicted to occur, the generated confidence in the possible future fall, and the like. If the probability criteria are not satisfied, the method 1000 terminates and returns to block 815 (where the machine learning system will likely determine that the criteria are not satisfied, and no interventions should be initiated).
  • the machine learning system may process the user’s pressure data (e.g., from a thirty second window) and/or the user’s characteristics or biographical data using one or more likelihood
  • the method 1000 continues to block 1015, where the machine learning system determines whether one or more temporal criteria are satisfied.
  • these temporal criteria relate to the urgency of the potential fall (e.g., how imminently or soon it is predicted to occur). For example, if the fall is predicted to occur within some defined “imminent” timeframe (e.g., within seconds or minutes), the machine learning system may determine that the temporal criteria are satisfied, and therefore bypass use of the severity model.
  • the machine learning system may determine that processing data using the severity model is useless, as the fall is likely to occur before the predicted severity is relevant. For example, if the user is expected to fall in the next few seconds or minutes, the potential severity of the fall may generally be irrelevant to the appropriate intervention: immediate response by one or more caregivers. By refraining from using the severity model, therefore, the machine learning system can save significant computational resources that need not be spent.
  • the method 1000 continues to block 1020, where the machine learning system generates a severity score using a second trained machine learning model.
  • this severity score can generally indicate the likely severity of any future falls, such as whether they may result in broken bone(s) (and if so, which bone(s)), lacerations, or other complications.
  • the potential severity may be an important factor in selecting appropriate interventions and performing other actions.
  • FIG. 11 is a flow diagram depicting an example method 1100 for initiating preventative and/or remedial actions based on machine learning model output.
  • the method 1100 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 1100 provides additional detail for block 820 of FIG.
  • the method 1100 generally indicates a variety of optional interventions that can be selected and implemented by the machine learning system, depending on the particular implementation, as well as on the particular likelihood score(s) and/or severity score(s) that are generated.
  • the machine learning system can optionally select one or more assistive or preventative devices for the user.
  • the machine learning system may suggest a cane, a walker, a wheelchair, a mobility scooter, and the like. This selection may be determined using machine learning (e.g., by processing modified user data with the models, as discussed above), using a rules-based system (e.g., based on the user’s aged, weight, mobility status, and the like), and the like.
  • the machine learning system can further consider the potential severity score to determine the best preventative device. For example, if the predicted fall severity is below a threshold, the machine learning system may indicate that the user can try a cane to retain mobility without significant inconvenience. If the severity is above some threshold, however, the machine learning system may indicate that a mobility scooter should be used, because the harm from a potential fall (even with other devices such as a cane) could be significant.
  • the machine learning system can optionally generate one or more therapy plans for the user. For example, as discussed above, if the machine learning system predicts that the user will fall towards one side, the machine learning system can suggest therapies or exercises that will improve the user’s strength on this side and thereby reduce the chance of a fall in that direction. Similarly, if the fall severity is predicted to exceed some threshold, additional physical therapy may be useful to help reduce the likely harm from the fall.
  • the particular therapy may be selected using machine learning (e.g., by processing modified pressure data or user data using the models), a set of defined rules (e.g., instructing strength training on one side when the fall is likely to be towards that side), and the like.
  • the machine learning system optionally alerts one or more user(s) of the upcoming potential fall. This may include, for example, alerting the user that they may be about to suffer a fall (e.g., via a smartphone of the user, via one or more output devices in the vicinity, such as smart televisions or speakers, and the like). For example, the machine learning system may suggest that the user grasp an object for support, sit down, slow down, and the like.
  • the alerts can be provided to one or more caregivers (e.g., to caregivers responsible for the user, or to caregivers identified as being nearby, physically, to the user). This alert may indicate a variety of data depending on the generated risk and/or severity scores. For example, the alert may indicate an urgency of the risk (e.g., how quickly the user needs to respond, and/or how severe the fall may be).
  • the machine learning system can optionally indicate problem region(s) in the space.
  • the machine learning system can aggregate predicted falls across users and/or time to identify physical regions of the facility that are particularly risky (e.g., using a heat map of user locations at the time of the predicted or actual falls). These locations may include hazards that can or should be mitigated, as discussed above.
  • each of the possible interventions in the method 1100 are optional, and the actual interventions initiated by the machine learning system may vary according to a wide variety of concerns and implementation details.
  • FIG. 12 is a flow diagram depicting an example method 1200 for evaluating and selecting personalized interventions using machine learning.
  • the method 1200 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the method 1200 provides additional detail for block 820 of FIG. 8 (after the machine learning system determines that a fall is likely, and decides to initiate one or more interventions).
  • the machine learning system identifies a set of potential intervention(s) for the user. For example, as discussed above with reference to FIG. 11, the machine learning system may select various assistive or preventative medical devices (such as a cane or walker), generate a new or revised therapy plan, alert nearby user(s), and the like.
  • assistive or preventative medical devices such as a cane or walker
  • the potential interventions are selected based at least in part on the user’s current characteristics. For example, if the user does not currently use any assistive device, the machine learning system may select one or more to suggest. If the user uses a device such as a cane, the machine learning system may suggest a more robust device such as a walker. In some embodiments, the machine learning system may determine whether the user has assistance standing up and/or walking, and suggest such assistance.
  • the machine learning system modifies the user data based on the potential interventions. That is, the machine learning system may artificially adjust the user data to reflect at least one of the proposed interventions.
  • the machine learning system determines a revised fall risk for the user, based on this modified user data. As discussed above, the fall risk may generally correspond to the likelihood of a fall and/or the severity of a potential fall.
  • the machine learning system may reprocess the user’s data (and pressure data, in some aspects), indicating that the user does have a cane, using the machine learning model(s). This can allow the machine learning system to simulate the effect of various interventions. At least one embodiment, the machine learning system repeats blocks 1210 and 1215 for each of the potential interventions determined in block 1205. That is, the machine learning system can separately determine revised fall risks for each possible intervention (and, in some embodiments, for combinations of interventions).
  • the machine learning system determines whether the potential intervention(s) reduced the fall risk, as compared to the user’s unmodified data. If not, the method 1200 continues to block 1225, where the machine learning system initiates one or more default interventions. For example, even if the simulation does not indicate that a walker would reduce the fall risk, the machine learning system may nevertheless assign, provide, or facilitate provision of a walker for the user. [0145] If, at block 1220, the machine learning system determines that one or more potential intervention(s) would reduce the fall risk, the method 1200 continues to block 1230, where the machine learning system initiates one or more of these personalized interventions. As these interventions are specifically designed for the particular user, and are validated via one or more simulations using machine learning, they may be better able to reduce the risk to the user and improve the overall results.
  • FIG. 13 is a flow diagram depicting an example method 1300 for determining aggregated fall risks using machine learning.
  • the method 1300 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • the machine learning system generates a set of fall risk scores for a set of users.
  • the fall risk score can generally quantify, for each user, the likelihood of a fall in the future and/or the potential severity of such a fall.
  • the machine learning system can generate the fall risk score(s) for each user by processing data associated with each user (e.g., pressure data and/or user characteristics) using one or more trained machine learning models, as discussed above.
  • the machine learning system may generate multiple risk scores for each user, based in part on when the pressure data is collected. That is, the machine learning system may generate risk scores for a single user at multiple points in time, where the score at each point is based on pressure data from the user at or near that time. For example, the machine learning system may generate a first score at a first time (e.g., at noon) based on pressure data from one or more windows prior to noon, and generate a second score at a second time (e.g., at three in the afternoon) based on pressure data from windows prior to three in the afternoon.
  • a first time e.g., at noon
  • a second time e.g., at three in the afternoon
  • the machine learning system aggregates the generated fall risks based on the location(s), in the facility, where user was when the pressure data used to generate the risk score was collected. That is, for each generated fall risk score, the machine learning system can identify the physical location(s) of the user when the pressure data was collected, and aggregate the scores based on these locations. For example, the machine learning system may, for one or more physical spaces or locations in the facility, identify a set of risk scores that are associated with each location (e.g., where the pressure data was collected within a defined distance from the location).
  • this aggregation includes determining a number of likely falls in each space (e.g., the number of times the machine learning system predicted a fall with sufficient probability), determining a potential severity of falls in each space (e.g., the average or maximum predicted severity, among the risk scores), and the like.
  • the machine learning system may provide a textual summary of the aggregated risks (e.g., indicating the number of predicted falls, the predicted severity, identifying users and/or regions with a large number of predicted falls, and the like).
  • the machine learning system can optionally display the aggregated data on a facility map.
  • the machine learning system may generate and output a heat map showing the distribution of fall probabilities, fall severities, or a combination of the two.
  • a heat map showing the distribution of fall probabilities, fall severities, or a combination of the two.
  • Such an augmented floorplan or map can be particularly useful to help users quickly identify problematic or hazardous areas.
  • the machine learning system determines whether there are any hotspots in the facility. That is, the machine learning system can determine whether any of the locations or spaces satisfy one or more defined criteria relating to the number and/or frequency of predicted falls, the potential severity of falls, and the like. If no such hotspots are found, the method 1300 returns to block 1305.
  • the method 1300 continues to block 1330, where the machine learning system initiates one or more interventions.
  • the machine learning system may indicate the location to one or more users or caregivers, and instruct that they investigate the area to identify potential hazards.
  • the machine learning system can identify the user (or users) associated with that space, in order to enable more targeted interventions with these user(s).
  • the method 1300 then returns to block 1305.
  • FIG. 14 is a flow diagram depicting an example method 1400 for training one or more machine learning models to predict fall risk.
  • the method 1400 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • a label is determined, the label indicating whether, within a defined window of time after the pressure data was collected, the user fell.
  • one or more machine learning models are trained, based on the pressure data and the label, to predict fall risk.
  • the pressure data is collected via one or more wearable devices comprising a plurality of pressure sensors, and the pressure data is collected while the user walked in a physical space.
  • determining whether the user fell comprises determining a time when the pressure data was collected, retrieving a set of user records corresponding to the defined window of time after the determined time, wherein the set of user records indicate patient data for the user, and evaluating the set of user records to determine whether a fall was reported.
  • the method 1400 further includes receiving pressure data corresponding to a plurality of users, and training the one or more machine learning models based on the pressure data for the plurality of users.
  • training the one or more machine learning models comprises training a first machine learning model to predict a likelihood that a user will fall, and training a second machine learning model to predict a severity of a potential fall.
  • the predicted severity is based at least in part on a predicted direction of the potential fall.
  • the method 1400 further includes receiving patient data indicating characteristics of the user, and training the one or more machine learning models based on the characteristics.
  • FIG. 15 is a flow diagram depicting an example method 1500 for predicting fall risk using machine learning.
  • the method 1500 is performed by a machine learning system, such as machine learning system 115 of FIG. 1.
  • a predicted fall risk is generated by processing the pressure data using one or more trained machine learning models.
  • the pressure data is collected via one or more wearable devices comprising a plurality of pressure sensors, and the pressure data is collected while the user walked in a physical space.
  • generating the predicted fall risk comprises generating a likelihood that the user will fall by processing the pressure data using a first machine learning model.
  • the method 1500 further includes, upon determining that the likelihood that the user will fall meets one or more temporal criteria, refraining from predicting a fall severity and alerting one or more nearby caregivers.
  • the method 1500 further includes generating a predicted severity of a potential fall by processing the pressure data using a second machine learning model.
  • generating the predicted severity is performed upon determining that the likelihood that the user will fall does not meet one or more temporal criteria.
  • the predicted severity is based at least in part on a predicted direction of the potential fall.
  • the method 1500 further includes receiving patient data indicating characteristics of the user, and generating the predicted fall risk by processing the characteristics of the user using one or more trained machine learning models.
  • the method 1500 further includes generating a plurality of predicted fall risks by, for each respective user of a plurality of users in a physical facility, processing respective pressure data using the one or more trained machine learning models, determining, for at least a subset of the plurality of predicted fall risks, a location in the physical facility where the predicted fall risk occurred, and upon determining that physical location satisfies one or more location criteria, selecting a facility intervention for the physical facility and initiating application of the facility intervention.
  • FIG. 16 depicts an example computing device 1600 configured to perform various aspects of the present disclosure. Although depicted as a physical device, in embodiments, the computing device 1600 may be implemented using virtual device(s), and/or across a number of devices (e.g., in a cloud environment). In one embodiment, the computing device 1600 corresponds to the machine learning system 115 of FIG. 1.
  • the computing device 1600 includes a CPU 1605, memory 1610, storage 1615, a network interface 1625, and one or more I/O interfaces 1620.
  • the CPU 1605 retrieves and executes programming instructions stored in memory 1610, as well as stores and retrieves application data residing in storage 1615.
  • the CPU 1605 is generally representative of a single CPU and/or GPU, multiple CPUs and/or GPUs, a single CPU and/or GPU having multiple processing cores, and the like.
  • the memory 1610 is generally included to be representative of a random access memory.
  • Storage 1615 may be any combination of disk drives, flash-based storage devices, and the like, and may include fixed and/or removable storage devices, such as fixed disk drives, removable memory cards, caches, optical storage, network attached storage (NAS), or storage area networks (SAN).
  • I/O devices 1635 are connected via the I/O interface(s) 1620.
  • the computing device 1600 can be communicatively coupled with one or more other devices and components (e.g., via a network, which may include the Internet, local network(s), and the like).
  • the CPU 1605, memory 1610, storage 1615, network interface(s) 1625, and I/O interface(s) 1620 are communicatively coupled by one or more buses 1630.
  • the memory 1610 includes a training component 1650, an inferencing component 1655, an action component 1660, which may perform one or more embodiments discussed above.
  • the training component 1650 is used to train the machine learning model(s), such as by using the method 600 of FIG. 6, the inferencing component 1655 may be configured to use the models to predict falls and/or severity using trained models, such as by using the method 800 of FIG. 8, and the action component 1660 may be configured to generate and/or initiate various responsive actions to the risks, such as by using the method 1100 of FIG. 11.
  • the storage 1615 includes historical data 1670 (which may correspond to labeled pressure data and/or user data used to train and/or evaluate the models), as well as one or more machine learning model(s) 1675. Although depicted as residing in storage 1615, the historical data 1670 and machine learning model(s) 1675 may be stored in any suitable location, including memory 1610. Generally, the historical data 1670 includes the previously- received (and labeled) pressure data, as well as any relevant user data (e.g., user age and weight data) used to train the machine learning models 1675.
  • user data e.g., user age and weight data
  • Clause 1 A method of training machine learning models, comprising: receiving pressure data corresponding to one or more feet of a user; determining a label indicating whether, within a defined window of time after the pressure data was collected, the user fell; and training one or more machine learning models, based on the pressure data and the label, to predict fall risk.
  • Clause 2 The method of Clause 1, wherein: the pressure data is collected via one or more wearable devices comprising a plurality of pressure sensors, and the pressure data is collected while the user walked in a physical space.
  • Clause 3 The method according to any one of Clauses 1-2, wherein determining whether the user fell comprises: determining a time when the pressure data was collected; retrieving a set of user records corresponding to the defined window of time after the determined time, wherein the set of user records indicate patient data for the user; and evaluating the set of user records to determine whether a fall was reported.
  • Clause 4 The method according to any one of Clauses 1-3, further comprising: receiving pressure data corresponding to a plurality of users; and training the one or more machine learning models based on the pressure data for the plurality of users.
  • Clause 5 The method according to any one of Clauses 1-4, wherein training the one or more machine learning models comprises: training a first machine learning model to predict a likelihood that a user will fall; and training a second machine learning model to predict a severity of a potential fall.
  • Clause 6 The method according to any one of Clauses 1-5, wherein the predicted severity is based at least in part on a predicted direction of the potential fall.
  • Clause 7 The method according to any one of Clauses 1-6, further comprising: receiving patient data indicating characteristics of the user; and training the one or more machine learning models based on the characteristics.
  • Clause 8 A method of predicting events using machine learning models, comprising: receiving pressure data corresponding to one or more feet of a user; generating a predicted fall risk by processing the pressure data using one or more trained machine learning models; and upon determining that the predicted fall risk satisfies one or more defined criteria: selecting an intervention for the user; and initiating application of the intervention.
  • Clause 9 The method according to Clause 8, wherein: the pressure data is collected via one or more wearable devices comprising a plurality of pressure sensors, and the pressure data is collected while the user walked in a physical space.
  • Clause 10 The method according to any one of Clauses 8-9, wherein generating the predicted fall risk comprises: generating a likelihood that the user will fall by processing the pressure data using a first machine learning model.
  • Clause 11 The method according to any one of Clauses 8-10, further comprising: upon determining that the likelihood that the user will fall meets one or more temporal criteria: refraining from predicting a fall severity; and alerting one or more nearby caregivers.
  • Clause 12 The method according to any one of Clauses 8-11, further comprising: generating a predicted severity of a potential fall by processing the pressure data using a second machine learning model.
  • Clause 13 The method according to any one of Clauses 8-12, wherein generating the predicted severity is performed upon determining that the likelihood that the user will fall does not meet one or more temporal criteria.
  • Clause 14 The method according to any one of Clauses 8-13, wherein the predicted severity is based at least in part on a predicted direction of the potential fall.
  • Clause 15 The method according to any one of Clauses 8-14, further comprising: receiving patient data indicating characteristics of the user; and generating the predicted fall risk by processing the characteristics of the user using one or more trained machine learning models.
  • Clause 16 The method according to any one of Clauses 8-15, further comprising: generating a plurality of predicted fall risks by, for each respective user of a plurality of users in a physical facility, processing respective pressure data using the one or more trained machine learning models; determining, for at least a subset of the plurality of predicted fall risks, a location in the physical facility where the predicted fall risk occurred; and upon determining that physical location satisfies one or more location criteria: selecting a facility intervention for the physical facility; and initiating application of the facility intervention.
  • an apparatus may be implemented or a method may be practiced using any number of the aspects set forth herein.
  • the scope of the disclosure is intended to cover such an apparatus or method that is practiced using other structure, functionality, or structure and functionality in addition to, or other than, the various aspects of the disclosure set forth herein. It should be understood that any aspect of the disclosure disclosed herein may be embodied by one or more elements of a claim.
  • exemplary means “serving as an example, instance, or illustration.” Any aspect described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects.
  • a phrase referring to “at least one of’ a list of items refers to any combination of those items, including single members.
  • “at least one of a, b, or c” is intended to cover a, b, c, a-b, a-c, b-c, and a-b-c, as well as any combination with multiples of the same element (e.g., a-a, a-a-a, a-a-b, a-a-c, a-b-b, a-c-c, b-b, b-b-b, b-b-c, c-c, and c-c-c or any other ordering of a, b, and c).
  • determining encompasses a wide variety of actions. For example, “determining” may include calculating, computing, processing, deriving, investigating, looking up (e.g., looking up in a table, a database or another data structure), ascertaining and the like. Also, “determining” may include receiving (e.g., receiving information), accessing (e.g., accessing data in a memory) and the like. Also, “determining” may include resolving, selecting, choosing, establishing and the like.
  • the methods disclosed herein comprise one or more steps or actions for achieving the methods.
  • the method steps and/or actions may be interchanged with one another without departing from the scope of the claims.
  • the order and/or use of specific steps and/or actions may be modified without departing from the scope of the claims.
  • the various operations of methods described above may be performed by any suitable means capable of performing the corresponding functions.
  • the means may include various hardware and/or software component(s) and/or module(s), including, but not limited to a circuit, an application specific integrated circuit (ASIC), or processor.
  • ASIC application specific integrated circuit
  • those operations may have corresponding counterpart means-plus- function components with similar numbering.
  • Embodiments of the invention may be provided to end users through a cloud computing infrastructure.
  • Cloud computing generally refers to the provision of scalable computing resources as a service over a network.
  • Cloud computing may be defined as a computing capability that provides an abstraction between the computing resource and its underlying technical architecture (e.g., servers, storage, networks), enabling convenient, on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction.
  • cloud computing allows a user to access virtual computing resources (e.g., storage, data, applications, and even complete virtualized computing systems) in “the cloud,” without regard for the underlying physical systems (or locations of those systems) used to provide the computing resources.
  • cloud computing resources are provided to a user on a pay-per-use basis, where users are charged only for the computing resources actually used (e.g. an amount of storage space consumed by a user or a number of virtualized systems instantiated by the user).
  • a user can access any of the resources that reside in the cloud at any time, and from anywhere across the Internet.
  • a user may access applications or systems (e.g., the machine learning system 115) or related data available in the cloud.
  • the machine learning system 115 could execute on a computing system in the cloud and train and/or use machine learning models. In such a case, the machine learning system 115 could train models to predict user falls, and store the models at a storage location in the cloud. Doing so allows a user to access this information from any computing system attached to a network connected to the cloud (e.g., the Internet).

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Pathology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Physiology (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Dentistry (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Fuzzy Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

L'invention concerne des techniques d'apprentissage machine améliorées. Des données de pression correspondant à un ou plusieurs pieds d'un utilisateur sont reçues, et un risque de chute prédit est généré par traitement des données de pression à l'aide d'un ou plusieurs modèles d'apprentissage machine formés. Lors de la détermination que le risque de chute prédit satisfait à un ou plusieurs critères définis, une intervention est sélectionnée destinée à l'utilisateur, et l'application de l'intervention est initiée.
EP22896686.7A 2021-11-16 2022-11-16 Apprentissage machine en vue de la prévention de chute par l?intermédiaire de capteurs de pression Pending EP4432916A1 (fr)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US202163264138P 2021-11-16 2021-11-16
US202163265083P 2021-12-07 2021-12-07
PCT/US2022/079961 WO2023091946A1 (fr) 2021-11-16 2022-11-16 Apprentissage machine en vue de la prévention de chute par l'intermédiaire de capteurs de pression

Publications (1)

Publication Number Publication Date
EP4432916A1 true EP4432916A1 (fr) 2024-09-25

Family

ID=86397787

Family Applications (1)

Application Number Title Priority Date Filing Date
EP22896686.7A Pending EP4432916A1 (fr) 2021-11-16 2022-11-16 Apprentissage machine en vue de la prévention de chute par l?intermédiaire de capteurs de pression

Country Status (2)

Country Link
EP (1) EP4432916A1 (fr)
WO (1) WO2023091946A1 (fr)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10422814B2 (en) * 2013-07-18 2019-09-24 Vital Connect, Inc. Fall detection using machine learning
US10743811B2 (en) * 2013-10-18 2020-08-18 Carepredict, Inc. Fall prediction assessment
EP3317630A4 (fr) * 2015-06-30 2019-02-13 ISHOE, Inc Identification de risque de chute utilisant des algorithmes d'apprentissage-machine
US20180228405A1 (en) * 2017-02-13 2018-08-16 Starkey Laboratories, Inc. Fall prediction system including an accessory and method of using same
US10891843B2 (en) * 2019-01-03 2021-01-12 International Business Machines Corporation Methods and systems for managing hazard risk based on location and incident data

Also Published As

Publication number Publication date
WO2023091946A1 (fr) 2023-05-25

Similar Documents

Publication Publication Date Title
US11819344B2 (en) Systems for automatic assessment of fall risk
JP6909078B2 (ja) 疾病発症予測装置、疾病発症予測方法およびプログラム
RU2700498C2 (ru) Персональная система экстренного реагирования с прогностической оценкой риска экстренного вызова
KR102340984B1 (ko) 기계학습을 이용하여 상처 관리를 위한 장치 및 상처 관리를 위한 장치의 동작방법
US20150261924A1 (en) Healthcare system and method
CN104823195B (zh) 一种降低临床设定中的妨害性警报负荷的方法和系统
US11862319B2 (en) Wound management and treatment using computer vision and machine learning
CN114068022A (zh) 跌倒风险的预警方法及装置、存储介质、电子设备
Mohan et al. Artificial Intelligence and IoT in Elderly Fall Prevention: A Review
US12073945B2 (en) Patient ventilator asynchrony detection
EP4432916A1 (fr) Apprentissage machine en vue de la prévention de chute par l?intermédiaire de capteurs de pression
US20230215551A1 (en) Machine learning for resource allocation
EP4113535A1 (fr) Procédés et systèmes de surveillance à distance pour surveiller des patients souffrant de maladies inflammatoires chroniques
CN118555933A (zh) 通过压力传感器预防跌倒的机器学习
US20230148960A1 (en) Wearable sensors for fall detection and prevention
KR102510180B1 (ko) 사용자 맞춤형 건강 관리 장치 및 방법
US20230200724A1 (en) Predicting wound management treatment resources using machine learning
CN113302701A (zh) 用于确定个体转移到急诊科的风险的数据处理系统及方法
US20230307140A1 (en) Machine learning for effective patient planning
US20240157900A1 (en) Systems and methods for accident prediction
Rajalakshmi et al. Inducement of Multivariate factors in Cardiac Disease Prediction with Machine Learning Techniques substantiated with Analytics
EP3709308A1 (fr) Probabilité et gravité de la chute d'un patient
US20240161929A1 (en) Machine learning for healthcare service planning
US20230107394A1 (en) Machine learning to manage sensor use for patient monitoring
US20210369142A1 (en) Predicting the probability of a brain injury of a subject resulting from a fall

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

17P Request for examination filed

Effective date: 20240514

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC ME MK MT NL NO PL PT RO RS SE SI SK SM TR