Methods and systems for using and training clinical support algorithms

By creating separate data streams for clinical decision algorithms and caregiver modifications, the variability in clinical support algorithms is reduced, enhancing precision and usability while maintaining caregiver flexibility in patient monitoring systems.

WO2024235618A1PCT designated stage expired Publication Date: 2024-11-21KONINKLIJKE PHILIPS NV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
PCT/EP2024/061704
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-15
Filing Date
2024-04-28
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

Current clinical support algorithms face variability issues due to caregiver modifications in patient monitoring data, leading to unpredictable outputs during both training and application, which reduces precision in predicting physiological conditions.

Method used

Generating two separate data streams: one containing the output of a clinical decision algorithm using raw physiological data and another with modified data for caregiver use, where the raw data is irreversibly modified using caregiver settings for user interface purposes, ensuring consistent input for the algorithm while allowing caregiver adjustments for visualization.

Benefits of technology

This approach reduces variability in clinical support algorithm outputs, improving precision without constraining caregiver modifications, thus enhancing clinical decision-making without impacting clinical care workflows.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure EP2024061704_21112024_PF_FP_ABST
    Figure EP2024061704_21112024_PF_FP_ABST
Patent Text Reader

Abstract

A method for collecting data indicative of the health of a subject. The method comprises obtaining raw physiological data of a subject and inputting the raw physiological data into a clinical support algorithm, where the clinical support algorithm is trained on prior raw physiological data of one or more subjects. A first data stream containing the output of the clinical support algorithm is generated. The raw physiological data is also separately modified using caregiver settings to generate a second data stream containing the modified physiological data.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] 1 20.03.2024

[0002] METHODS AND SYSTEMS FOR USING AND TRAINING CLINICAL SUPPORT ALGORITHMS

[0003] FIELD OF THE INVENTION

[0004] The invention relates to clinical support algorithms. In particular, the invention relates to a method for using clinical support algorithms and methods for training clinical support algorithms.

[0005] BACKGROUND OF THE INVENTION

[0006] The performance of Clinical Decision Support Algorithms and Analytics and their specificity depends on the quality and consistency of the input data. However, the training data for developing and for applying these algorithms is currently captured from existing data streams that are intended for patient monitoring and real-time alarming.

[0007] Patient monitoring systems allow the caregiver to adjust the behavior of these existing data streams to optimize them for the specific environment, personal or hospital preference and the specific patient status. These adjustments (caregiver modifications) can significantly impact the data values and therefore the same (physiological) situation might be represented by a wide spread of clinically collected data values, which impacts the algorithm’s development. The same uncertainty occurs again when such an algorithm is then used for a specific patient in a clinical context. It has been realized that avoiding this variability, which is not a physiological variability, would improve the performance of such algorithms.

[0008] The current approach for developing analytics is to collect data from a patient monitoring system by obtaining the data as it is provided in a data stream used for data interfaces. Data interfaces are used by caregivers to visualize physiological data on a corresponding user interface / display of the patient monitor or at the user interface of a connected system, such as a central station or an electrical medical record (EMR). This data stream is optimized by the caregiver for user interface purposes and for the alarming behavior of the patient monitoring system.

[0009] Some of the caregiver modifications can be mapped in a more or less unambiguous way to a standardized representation, e.g., the unit of measure. For analytics training purposes, this mapping can be done as part of the data collection process or at the time when the data is used for training.

[0010] However, conventionally, when the mapping takes place after the data was collected, the caregiver settings used to achieve the caregiver modifications are not stored with the data. Therefore, a correct and time specific mapping cannot always be achieved and thus the raw physiological data cannot be restored from the modified physiological data. When an analytics algorithm is used, such a mapping needs to take place such that the input data for the algorithm represents the format that the algorithm was trained with.

[0011] A major disadvantage of the current approach is that some caregiver modifications modify the data in such a way that the data cannot be mapped in an unambiguous way from one caregiver adjustments choice to another. Moreover, the mapping would require additional data that is not always available as part of the data collection process, e.g., ambient pressure. In these cases, the mappings are irreversible if there is no access to the necessary additional data.

[0012] This causes the data to deviate depending on the modifications that a caregiver has chosen, even for identical physiological conditions, thus providing modified data having “physiological uncertainty”. As such, training a clinical support algorithm with the modified data can negatively impact the outputs of the algorithm during its application.

[0013] Similarly, the application of a clinical support algorithm using modified data causes the output to be further impacted by the physiological uncertainty due to possible further adjustments that a caregiver might apply to the data stream in order to control how the data is provided at the user interface. In particular, the modified data used in the application of the analytics algorithm may not be adapted in the same manner as the training data used on existing clinical support algorithms.

[0014] In summary, the current approach of using clinical support algorithms adds unpredictable variability to the outputs twice, once during training of the algorithm and once during application of the algorithm. This variability reduces the achievable precision of clinical support algorithm to predict a specific physiological condition.

[0015] One way to overcome these limitations would be to prevent the caregiver modifications and force the caregivers to abstain from optimizing the individual physiological data e.g., for displaying or for real-time alarming. However, caregivers often want to make modifications to the data stream to control how the data is provided at the user interface. Additionally, caregivers do not want to be limited in their freedom to modify the data stream for the purpose of training and / or using a clinical support algorithm. Such a solution is unlikely to be acceptable to caregivers and may negatively impact clinical care.

[0016] As such, there is a need for a solution which provides reduced variability in the output of clinical support algorithms without significantly impacting a caregiver’s workflow or clinical care.

[0017] US 2021 / 027891 discloses a therapeutic management system using a therapeutic analytics engine. US 2022 / 084662 discloses a system for notifying a caregiver when clinical intervention is needed based on a machine learning model US 11 587682 discloses a care system for home based patients, which processes smart phone sensor data. US 2022 / 301666 discloses a system for capturing and recording patient treatment and physiological data. SUMMARY OF THE INVENTION

[0018] The invention is defined by the independent claims. The dependent claims define advantageous embodiments.

[0019] According to examples in accordance with an aspect of the invention, there is provided a computer implemented method for collecting data indicative of the health of a subject, the method comprising: obtaining raw physiological data of a subject; inputting the raw physiological data into a clinical support algorithm trained on prior physiological data of one or more subjects to generate a first data stream containing the output of the clinical support algorithm; and modifying the raw physiological data using caregiver settings or by applying adjustments for user interface purposes to generate a second data stream containing the modified physiological data.

[0020] The raw physiological data comprises raw sensor data, and / or raw waveform data, and / or raw parameter data and / or physiological data that has been adapted using standardized caregiverindependent processing to derive standardized physiological information.

[0021] The physiological information is displayed based on the second data stream. The obtaining and the modifying are performed by a patient monitoring system, and the second data stream comprises the monitoring output of the patient monitoring system.

[0022] When using a trained algorithm, it is important that the representation of the inputs is consistent and standardized. However, patient monitoring systems process physiological data in a way to achieve a representation (i.e. the monitoring output of the patient monitoring system) that is suited for use by caregivers via a user interface, which might eliminate important data content that might be beneficial for an algorithm like a clinical decision support algorithm. Furthermore, many caregivers want to control how the data is presented (e.g., at a user interface) by adapting / modifying the raw physiological data using caregiver settings. Many caregiver settings change the raw physiological to the extent that the modified physiological data can no longer be practically mapped to the representation preferred by the clinical support algorithm.

[0023] Caregiver settings are settings used to modify the raw physiological data for later use by the caregiver. For example, the modifications may be for the purpose of presenting / displaying the physiological data to caregivers and / or for the purpose of alarming. The caregiver settings may include settings set directly by the caregiver (i.e., caregiver-defined settings). In some cases, the caregiver settings may not be directly set by the caregiver. For example, some caregiver settings may be defined by safety and / or performance standards.

[0024] As such, it is proposed to generate two separate data streams using the raw physiological data. One of the data streams contains the output of the clinical decision algorithm, where the raw physiological data was used as an input for the clinical decision algorithm. The other data stream contains modified physiological data for use by the caregiver, where the modified physiological data is the raw physiological data modified according to the caregiver settings. For example, the raw physiological data may be modified according to the preferences of the caregiver (i.e., using caregiver-defined settings) and / or modified to achieve a particular representation of the data for ease of visualization by the caregiver (e.g., applying high-pass or low-pass filtering).

[0025] The output of the clinical support algorithm may then be displayed (e.g., to the caregiver). The output of the clinical support algorithm may also be modified according to the preferences of the caregiver (i.e., using caregiver-defined settings). Similarly, the modified physiological data (using the caregiver settings) may also be displayed to the caregiver. Of course, one or more of the outputs of the clinical support algorithm and the modified physiological data may also be stored for later use.

[0026] The preferred representation of the input data to the clinical support algorithm is based on the representation of the data which was used to train the clinical support algorithm. Said representation can be defined using data properties (e.g., unit of measurement, resolution, normalization method, correction method, offset, scale etc.) The representation / properties of the physiological data are often unknown from the physiological data per se unless explicitly specified in the physiological data.

[0027] Raw physiological data is data indicative of direct or indirect observations of variables attributable to the functioning of systems and / or subsystems in the subject’s body. Raw physiological data is generally physiological data, which has not been fully processed after being acquired. It must be understood that raw physiological data can still be the result of processing physical signals e.g., from sensors, i.e., raw sensor data. Such processing of physical signals (raw sensor data) can involve several steps, like digitizing analog signals, demodulating light or filtering electrical sensor signals to the relevant physiological bandwidth.

[0028] The result of this processing is typically the raw waveform data, which shall also be understood as raw physiological data in context of the invention.

[0029] For physiological parameters that need a physiological algorithm in order forthem to be determined based on a series of raw waveform data samples, the raw output of such a physiological algorithm, the raw parameter data, shall also be understood as raw physiological data in context of the invention.

[0030] For example, ECG waveform signals are the result of processing raw sensor data, i.e. the sensed physical voltage signals from multiple electrodes. The ECG waveforms (e.g. 1-lead, 3-lead, 12- lead ECG waveforms) resulting from this basic processing would be raw physiological data in the definition of the invention, the raw waveform data. The ECG waveforms are then typically processed further to determine other physiological parameters, e.g. the Heart Rate. The output of such a Heart Rate algorithm must be understood in context of the invention as a separate example of raw physiological data, namely raw parameter data in the form of a raw Heart Rate in this case.

[0031] Further caregiver dependent processing of such Heart Rate data, like e.g. averaging based on a caregiver setting, or processing to achieve data reduction before the Heart Rate is displayed or provided as output of the patient monitoring system would according to the invention be the modified physiological data with “physiological uncertainty” due to the introduced variability caused by the caregiver settings.

[0032] Another example would be pulse oximetry, wherein the photoplethysmogram (PPG) waveforms are raw physiological signals as the result of processing the raw sensor data, i.e., the sensed opto-electrical signals from the sensor. Such PPG waveforms are then typically processed further with a pulse oximeter algorithm to determine other physiological parameters, e.g. the SpO2 value or the Pulse Rate. The output of such a pulse oximeter algorithm must be understood in context of the invention as another example of raw physiological data, namely the raw SpO2 value or the raw Pulse Rate.

[0033] A data stream, in general, contains data (in this case, raw physiological data) to convey information.

[0034] In the context of the invention, a clinical support algorithm is a trained algorithm that uses a subject’s health information (i.e., the raw physiological data) to assist in making a diagnosis, treatment decisions, monitoring of the subject, deriving further physiological data or for similar purposes. For example, the clinical support algorithm may be a clinical decision support algorithm.

[0035] It will be appreciated that when the method is used with raw physiological data obtained from the same device (or similar devices), it can be assumed that the raw physiological data will have the same representation.

[0036] The method may comprise irreversibly modifying the raw physiological data using the caregiver settings to generate the second data stream.

[0037] Irreversibly modifying the raw physiological data means modifying the raw physiological data such that the raw physiological data cannot be derived from modified physiological data without additional information.

[0038] Modifying the raw physiological data may comprise processing the raw physiological data with one or more mathematical calculations to generate the modified physiological data.

[0039] Irreversibly modifying the raw physiological data using caregiver settings may comprise one or more of filtering the raw physiological data, averaging the raw physiological data over time, offsetting the raw physiological data, normalizing the raw physiological data, and calculating a subjectspecific indexed representation of the raw physiological data. All of these adjustments irreversibly change the raw physiological data. Of course, other modifications will be known which irreversibly change the raw physiological data.

[0040] Inputting the raw physiological data into a clinical support algorithm may comprise adapting physiological data (e.g„ sensor data or waveform data) using standardized processing prior to deriving the raw physiological data to be input into the clinical support algorithm, wherein the prior raw physiological data on which the clinical support algorithm is trained may also be physiological data adapted using the standardized processing.

[0041] In some cases, the representation of the raw physiological data may not be the preferred representation of the clinical support algorithm. In these cases, it is preferable to process (in the first data stream) physiological data using standardized settings to adapt the representation to the preferred representation of the raw physiological data. In other words, the same standardized processing is used on the inputs to the clinical support algorithm as well as the data used to train the clinical support algorithm.

[0042] The standardized processing is independent from caregiver settings and typically different to the modifications due to the caregiver settings.

[0043] The standardized processing may be an initial processing / pre-processing of physiological data (e.g., sensor data or waveform data). For example, the standardized processing may include removing outlier datapoints from the raw physiological data.

[0044] The raw physiological data may comprise raw hemodynamic data of the subject. Often, hemodynamic data is adapted by the caregiver in such a manner that does not allow it to be mapped back to the raw hemodynamic data. For example, such data is often offset, normalized, corrected, filtered, averaged etc.

[0045] The method may further comprise storing and / or displaying the output of the clinical support algorithm and / or storing and / or displaying the modified physiological data.

[0046] The first data stream and the second data stream may be output to the clinician.

[0047] The invention also provides a computer implemented method for training the clinical support algorithm with data indicative of the health of a subject, the method comprising: obtaining raw physiological data of a subject; and training a clinical support algorithm with the prior raw physiological data of the subject.

[0048] As with using clinical support algorithms, it is also important that the data on which they are trained have a consistent representation.

[0049] Training a clinical support algorithm with the prior raw physiological data may comprise adapting the physiological data applying standardized processing prior to training the clinical support algorithm with the raw physiological data.

[0050] The raw physiological data may comprise raw hemodynamic data of the subject.

[0051] The invention also provides a computer program carrier comprising computer program code which, when executed on a processor, causes the processor to perform all of the steps of any of the aforementioned methods.

[0052] A computer program carrier may be, for example, a long-term storage product (e.g., nonvolatile memory such a hard drive or a solid-state drive) or a bitstream.

[0053] The invention also provides a system for collecting data indicative of the health of a subject, the system comprising a processor configured to: obtain raw physiological data of a subject; input the raw physiological data into a clinical support algorithm trained on prior physiological data of one or more subjects to generate a first data stream containing the output of the clinical support algorithm; and modify the raw physiological data using caregiver settings to generate a second data stream containing the modified physiological data.

[0054] The raw physiological data may comprise raw hemodynamic data of the subject.

[0055] The processor may be configured to modify the raw physiological data based on caregiver settings by filtering the raw physiological data, averaging the raw physiological data overtime, offsetting the raw physiological data, normalizing the raw physiological data and / or calculating an index using the raw physiological data.

[0056] The processor may be configured to irreversibly modify the raw physiological data using the caregiver settings or by applying adjustments for user interface purposes to generate the second data stream.

[0057] The processor may be configured to input the raw physiological data into a clinical support algorithm by adapting the raw physiological data applying standardized processing prior to inputting the raw physiological data into the clinical support algorithm, wherein the prior raw physiological data on which the clinical support algorithm is trained is also adapted applying the standardized processing.

[0058] The system may further comprise one or more physiological measurement devices configured to obtain the raw physiological data.

[0059] The system may further comprise a display configured to display the output of the clinical support algorithm and / or configured to display the modified physiological data.

[0060] The system may also comprise a data storage device configured to store the output of the clinical support algorithm and / or configured to store the modified physiological data.

[0061] The invention also provides a computer implemented method for collecting data indicative of the health of a subject, the method comprising: obtaining raw physiological data of a subject; adapting the raw physiological data using standardized settings to generate standardized physiological data; inputting the standardized physiological data into a clinical support algorithm trained on prior standardized physiological data of one or more subjects to generate a first data stream containing the output of the clinical support algorithm; and modifying the raw physiological data using caregiver settings to generate a second data stream containing the modified physiological data.

[0062] All of the optional features described herein can also be used with the afore-mentioned method using standardized settings.

[0063] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] For a beter understanding of the invention, and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0065] Fig. 1 shows a patient monitoring system inputing raw physiological data into a clinical support algorithm;

[0066] Fig. 2 shows the patient monitoring system inputing standardized physiological data into a clinical support algorithm;

[0067] Fig. 3 shows a patient monitoring system obtaining ataining dataset of raw physiological data;

[0068] Fig. 4 shows the patient monitoring system obtaining ataining dataset of standardized physiological data;

[0069] Fig. 5 shows a method for collecting data indicative of the health of a subject; and

[0070] Fig. 6 shows a system for collecting data indicative of the health of a subject.

[0071] DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] The invention will be described with reference to the Figures.

[0073] It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the apparatus, systems and methods, are intended for purposes of illustration only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems and methods of the present invention will become beter understood from the following description, appended claims, and accompanying drawings. It should be understood that the Figures are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the Figures to indicate the same or similar parts.

[0074] The invention provides a method for collecting data indicative of the health of a subject. The method comprises obtaining raw physiological data of a subject and inputing the raw physiological data into a clinical support algorithm, where the clinical support algorithm is trained on prior physiological data of one or more subjects. A first data stream containing the output of the clinical support algorithm is generated. The raw physiological data is also separately modified using caregiver setings to generate a second data stream containing the modified physiological data.

[0075] The invention can be used in an improved patient monitoring system to provide an additional independent data path, for the physiological data, that is not impacted by the caregiver modifications. Such caregiver modifications are performed to achieve a commonly accepted way to present physiological data to caregivers and / or modifications that an individual caregiver might want to make regarding how the physiological data is filtered, averaged, indexed, or presented based on a selected reference method (i.e., using caregiver setings to modify the physiological data). This improved patient monitoring system can be used for collecting the clinical data for algorithm development purposes (i.e., training) as well as when such an algorithm is used.

[0076] Thus, a caregiver can be provided with improved outputs from the clinical support algorithm (e.g., clinical user gets improved Clinical Decision Support Algorithms and Analytics) without constraining the clinical practice.

[0077] Fig. 1 shows a patient monitoring system 104 inputting raw physiological data 106,108 into a clinical support algorithm 114 and at the same time, based on the same raw physiological data 106,108, provides a second data stream of physiological data 118 that includes caregiver modifications.

[0078] To overcome the above-described systematic deficiency of using physiological data with “physiological uncertainty” when using clinical support algorithms, a new approach is necessary. It is therefore proposed that, instead of using existing physiological data 110,112 that is optimized for use by caregivers (e.g., for presenting the data or for alarming), additional data paths are provided directly between raw physiological data 106,108 (as explained below) and the clinical support algorithm 114. As such, the input data for the clinical support algorithm 114 is not impacted by caregiver modifications in contrary to the monitoring output 118 that is available in existing patient monitoring systems.

[0079] "Raw physiological data" as used in this application, means data representing a physiology of a patient that has not been subjected to modifications based on customer selections or modification that reduces the signal content due to usability considerations or standard requirements for patient monitoring and alarming. The raw physiological data may comprise raw sensor data, i.e. the output signal from a physiological sensor, e.g. the output of a temperature sensor. However, for many sensor types, this raw sensor data is a physical signal that is only indicative of a physiological signal and that may require some pre-processing applied to the raw sensor data before it is considered to be raw physiological data, typically raw waveform data. For example, this pre-processing may involve converting the output signal from a physical sensor, e.g., voltage or current, into e.g., blood pressure level, photoplethysmogram (PPG) , etc. Furthermore, some physiological data may not yet represent physiology in a meaningful way and therefore goes through some standardized processing. For example, this standardized processing may also involve offset calibration, and / or zero calibration, and / or artefact removal. For example, invasive blood pressure measurements may be calibrated to obtain the raw physiological data.

[0080] In many cases the raw waveform data is not the only raw physiological data that is derived from a sensor. Often physiological parameters are derived from such raw waveform data by applying a physiological algorithm. E.g., an ECG waveform or PPG waveform are inputs to determine physiological parameters, e.g. Heart Rate or SpO2 value. Also this raw parameter data, which has not been subjected to modifications based on customer selections or modification due to usability considerations or standard requirements for patient monitoring and alarming, shall be understood as raw physiological data in context of the invention. The standardized processing reduces the variability of the data representation of the patient, so that it has a standard form across all patients. The standardized processing is independent of the caregiver, i.e. it is the same for all users of the system.

[0081] In the context of this invention, the "raw physiological data" may comprise physiological data that has been processed by this standardized processing.

[0082] Thus, in all cases, the "raw physiological data" is data before any caregiver selected processing has been applied or any user interface (e.g., display) adjustments.

[0083] In Fig. 1, the raw sensor data from the subject 102 comprises signals for two separate physiological measurements, measurement 106 and measurement 108. In this case, there exists a data path directly between the measurements 106,108 (i.e. the raw physiological data) and the clinical support algorithm 114. A first data stream 116, containing the output of the clinical support algorithm 114, is then generated. Of course, it will be appreciated that the raw sensor data from the subject 102 may comprise one or more measurements.

[0084] In Fig. 2, physiological data 106,108 has been processed by standardized processing to derive the raw physiological data. Thus, standardized physiological data 202,203 is input into the clinical support algorithm 114. The standardized processing may apply standardized settings used for adjusting the physiological data 106,108. In this case, the best performance can be achieved in case the clinical support algorithm has also been trained with training data, which has been adjusted based on the standardized processing.

[0085] The standardized processing is known processing that is performed by a patient monitoring system 104. In addition, further processing 110, 112 is performed before the monitoring output 118 of the patient monitoring system 104 is generated. It is noted that this monitoring output is often referred to as the "raw" data in known systems, because it is currently the only data stream output by a patient monitoring system. This further processing is discussed below. It may modify the raw physiological data (which may be standardized physiological data as explained above) based on customer selections or it may apply modification that may reduce the signal content due to usability considerations or standard requirements for patient monitoring and alarming. The purpose of this further processing 110,112 is to prepare the data to be in a suitable form for the user interface or other analysis. For example, it may adapt the data according to the measurement units to be displayed, provide a normalization of the data (e.g. to a body surface area), perform filtering, or perform a time averaging function so that the data does not change too rapidly for a user. These adaptations may be the result of individual caregiver preferences, or they may be the result of hospital policies (i.e. a caregiver organization) about the way patient data is represented. The term "caregiver" should be understood accordingly. This processing, based on caregiver settings, is mainly for the purposes of visualization of real-time physiological data and corresponding alarming of a patient monitoring system.

[0086] The patient monitoring system 104 thus comprises at least one measurement function that is intended to provide physiological data to a caregiver. Such a patient monitoring system 104 also processes the physiological data (by the "further processing" explained above) to achieve a commonly accepted way to present physiological data to caregivers and / or even allows the caregiver to adjust how the physiological data is processed to achieve a desired behavior when the physiological data is presented on the user interface or forwarded to another system components, e.g., for alarming or for remote viewing.

[0087] In Fig. 1 and Fig. 2, the further processing is in a separate data path (separate to the data path to the clinical support algorithm) between the measurements 106 and 108 and an element 120 that is using a standard patient monitor output 118 (e.g. display, data interface port, alarming). The measurements 106 and 108 are modified using the "further processing" explained above. This further processing comprises caregiver modifications 110 and 112 to achieve a commonly accepted way to present physiological data to caregivers and / or are modified based on caregiver setting (e.g., caregiver preferences). The caregiver modifications 110 and 112 can indicate caregiver settings to be used to modify the measurements 106 and 108 or the standardized physiological data 202 and 203, respectively. A second data stream 118 is then generated containing the modified measurements. The second data stream can then provide the modified measurements to a display (e.g., at a user interface) and / or may be otherwise used by the caregiver (e.g., for alarming or remote viewing purposes). More generally, the raw physiological data 106,108 is modified to achieve a target behavior for user interface purposes or it is modified based on caregiver settings. The modified physiological data is used by the caregiver.

[0088] Fig. 3 shows a patient monitoring system 304 obtaining a training dataset 316. Raw physiological data 306, 308 is obtaining from a plurality of subjects 302. In this case, the raw physiological data comprises two separate physiological measurements from the plurality of subjects, measurements 306 and 308. Like Fig. 1, the raw sensor data is used. A data path is provided directly between the measurements 306 and 308 and a data collection module 314. The raw physiological data 302 is then provided directly to the training dataset 316 without any caregiver modifications. A separate data path is provided between the measurements 306 and 308 and a user interface 320 or data interface. The caregiver modifications 310 and 312 can then be applied to the raw physiological data 306,308 without affecting the training data in the training dataset 316 being collected.

[0089] A clinical support algorithm can then be trained with the collected training data in the training dataset 316.

[0090] Fig. 4 shows the patient monitoring system 304 obtaining a standardized training dataset 404. Thus, it uses the standardized processing discussed for Fig. 2. The standardized training dataset 404 is used to train a clinical support algorithm when the clinical support algorithm is intended to be input with standardized physiological data. As such, the raw physiological data 306, 308 is standardized and the standardized physiological data 402, 403 is collected by the data collection module 314 and provided to the standardized training dataset 404 for use in training the clinical support algorithm.

[0091] Table 1 below shows examples of physiological data and corresponding settings that caregivers adjust which irreversibly impact the representation of the data (i.e., the physiological data is irreversibly modified) without knowledge of all of the inputs used to modify the physiological data. In other words, the caregiver settings shown in Table 1 cause modifications to raw physiological data which are irreversible per se and can only be reversed with knowledge of the inputs used to modify the raw physiological data. It is noted that said inputs are usually not stored, effectively making the modifications irreversible.

[0092] The caregiver settings may include settings set by the caregiver for the purpose of displaying the modified data at a user interface, setting alarms, or for remote viewing purposes. The caregiver settings may also, or alternatively, include settings set by an institution (e.g., a hospital) for the purpose of safety and / or performance standards.

[0093] Table 1

[0094] Adjusting raw physiological data based on the settings mentioned above can cause an irreversible change to the representation of the raw physiological data if additional information is not provided which would enable the reversal of said modifications. Currently, when obtaining training data for a clinical support algorithm, the additional information may not be available. Similarly, the additional information may not be available when using the modified physiological data as an input for a clinical support algorithm.

[0095] Table 2 below shows examples of physiological data and corresponding caregiver settings that caregivers adjust which irreversibly impact the raw physiological data. In these cases, the modified physiological data cannot be mapped back to a unified behavior even when additional information is provided.

[0096] Table 2 Using any of the caregiver settings defined in Tables 1 and 2 above can often aid the caregiver when interpreting the physiological data, especially when compared with interpreting raw physiological data. However, clinical support algorithms may work best with raw physiological data or specific standardized representations of the input data. As such, it is proposed to use the raw physiological data as an input to the clinical support algorithm whilst providing the raw physiological data for modification based on the user interface needs and caregiver settings separately. The raw physiological data can be adapted based on standardized processing prior to being input into the clinical support algorithm, where the clinical support algorithm may be trained with raw physiological data adapted with the same standardized processing. The standardized processing is independent from the caregiver settings and may be different to the caregiver related adjustments.

[0097] In a first example, the standardized processing for non-invasive blood pressure (NiBP) measurements may define that the NiBP values are calibrated for invasive radial artery reference independently from the customer configuration (which could also be invasive femoral reference or auscultatory reference).

[0098] In a second example, the standardized processing for hemodynamic measurements may define the use of a standardized BSA method to index the hemodynamic parameters.

[0099] In a third example, the standardized processing for the raw physiological data may define a standardized unit for the measurement independent for the selected unit for presentation (e.g., kPa, mmHg, cm02 etc.).

[0100] The approach described herein can also be applied in distributed patient monitoring systems where the measurement function, the data storage, the modifications to the raw physiological data, the use of the data as well as the provision of the output of the clinical support algorithm can be distributed across different devices. For example, a telemetry transmitter and a central station may form part of the patient monitoring system. Additionally, applications of the approach in context of electronic medical records or cloud-based analytics can also benefit from the approach described herein.

[0101] Fig. 5 shows a method for collecting data indicative of the health of a subject. In step 502, raw physiological data of a subject is obtained. As explained above, raw physiological data may comprise raw waveform data or raw parameter data. Standardized processing is optionally applied in step 504 to waveform data or parameter data in order to derive the raw physiological data. For example, the raw physiological data may be measured using a physiological measurement device. In step 506, the raw physiological data or standardized physiological data is input into a clinical support algorithm. The clinical support algorithm has been trained on prior physiological data of one or more subjects. A first data stream containing the output of the clinical support algorithm is thus generated. Optionally in step 508, the output of the clinical support algorithm is displayed and / or stored.

[0102] In step 510, the raw physiological data or standardized physiological data is modified using caregiver settings or by applying adjustments for user interface purposes to generate a second data stream containing the modified physiological data. Optionally, the modified physiological data can be displayed and / or stored in step 512.

[0103] The standardized processing in general is different from the possible modifications to the raw physiological based on the caregiver settings. It is in particular caregiver-independent. But it can also occur that a specific combination of caregiver settings results in the same behavior as the standardized processing.

[0104] Fig. 6 shows a system for collecting data indicative of the health of a subject. The system comprises one or more physiological measurement devices 602 configured to obtain raw physiological data, a processor 604 capable of performing all of the steps shown in Fig. 5 and a display 606 for displaying the output of the clinical support algorithm and / or the modified physiological data. The system may also comprise a non-volatile memory for storing the output of the clinical support algorithm and / or the modified physiological data.

[0105] The skilled person would be readily capable of developing a processor for carrying out any herein described method. Thus, each step of a flow chart may represent a different action performed by a processor and may be performed by a respective module of the processing processor.

[0106] As discussed above, the system makes use of a processor to perform the data processing. The processor can be implemented in numerous ways, with software and / or hardware, to perform the various functions required. The processor typically employs one or more microprocessors that may be programmed using software (e.g., microcode) to perform the required functions. The processor may be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0107] Examples of circuitry that may be employed in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0108] In various implementations, the processor may be associated with one or more storage media such as volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when executed on one or more processors and / or controllers, perform the required functions. Various storage media may be fixed within a processor or controller or may be transportable, such that the one or more programs stored thereon can be loaded into a processor.

[0109] Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality.

[0110] A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0111] A computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0112] If the term "adapted to" is used in the claims or description, it is noted the term "adapted to" is intended to be equivalent to the term "configured to".

[0113] Any reference signs in the claims should not be construed as limiting the scope.

Claims

CLAIMS:

1. A computer implemented method for collecting data indicative of the health of a subject, the method comprising: obtaining (502) raw physiological data of a subject, comprising raw sensor data, raw waveform data, raw parameter data and / or physiological data adapted using standardized caregiverindependent processing to derive standardized physiological data; inputting (506) the raw physiological data into a clinical support algorithm trained on prior physiological data of one or more subjects to generate a first data stream containing the output of the clinical support algorithm; modifying (510) the raw physiological data using caregiver settings or by applying adjustments for user interface purposes to generate a second data stream containing the modified physiological data, and displaying physiological information based on the second data stream, wherein the obtaining and the modifying are performed by a patient monitoring system, and the second data stream comprises the monitoring output of the patient monitoring system.

2. The method of claim 1, comprising irreversibly modifying the raw physiological data using the caregiver settings or by applying adjustments for user interface purposes to generate the second data stream.

3. The method of claim 2, wherein irreversibly modifying the raw physiological data using caregiver settings or by applying adjustments for user interface purposes comprises one or more of: filtering the raw physiological data; averaging the raw physiological data over time; offsetting the raw physiological data; normalizing the raw physiological data; and calculating a subject-specific indexed representation of the raw physiological data.

4. The method of any of claims 1 to 3, wherein the prior physiological data on which the clinical support algorithm is trained comprises said raw physiological data.

5. The method of any of claims 1 to 4, wherein the raw physiological data comprises hemodynamic data of the subject.

6. The method of any of claims 1 to 5, further comprising storing and / or displaying (508) the output of the clinical support algorithm.

7. A computer implemented method for training the clinical support algorithm of any of claims 1 to 6 with data indicative of the health of a subject, the method comprising: obtaining raw physiological data of a subject comprising raw sensor data, raw waveform data, raw parameter data and / or physiological data adapted using standardized caregiver-independent processing to derive standardized physiological data; and training a clinical support algorithm with the prior raw physiological data of the subject.

8. The method of claim 7, wherein the raw physiological data comprises hemodynamic data of the subject.

9. A computer program carrier comprising computer program code which, when executed on a processor, causes the processor to perform all of the steps of the method according to any of the preceding claims.

10. A system for collecting data indicative of the health of a subject, the system comprising a patient monitoring system and a clinical support algorithm, wherein the system comprises: a processor (604) configured to: obtain raw physiological data of a subject (102) comprising raw sensor data, raw waveform data, raw parameter data and / or physiological data adapted using standardized caregiverindependent processing to derive standardized physiological data; input the raw physiological data into the clinical support algorithm (114) trained on prior physiological data of one or more subjects to generate a first data stream (116) containing the output of the clinical support algorithm; and modify the raw physiological data using caregiver settings or by applying adjustments for user interface purposes to generate a second data stream (118) containing the modified physiological data; a display configured to display physiological information based on the second data stream. wherein the obtaining and the modifying are performed by the patient monitoring system, and the second data stream comprises the monitoring output of the patient monitoring system.

11. The system of claim 10, wherein the processor is configured to irreversibly modify the raw physiological data using the caregiver settings or by applying adjustments for user interface purposes to generate the second data stream.

12. The system of any of claims 10 to 11, wherein the patient monitoring system comprises one or more physiological measurement devices (602) configured to obtain the raw sensor data, the raw waveform data or the raw parameter data.

13. The system of claims 10 to 12, wherein the display (606) is further configured to display the output of the clinical support algorithm.

Citation Information

Patent Citations

  • Method and system to integrate data, analyze and develop improved care plan for a patient at home

    US11587682B2

  • Systems and Methods for Personalized Medication Therapy Management

    US20210027891A1

  • Systems and methods for automatically notifying a caregiver that a patient requires medical intervention

    US20220084662A1

  • System and methods of monitoring a patient and documenting treatment

    US20220301666A1