[go: up one dir, main page]

WO2025051578A1 - Method and system for predicting an adverse technical and/or medical event - Google Patents

Method and system for predicting an adverse technical and/or medical event Download PDF

Info

Publication number
WO2025051578A1
WO2025051578A1 PCT/EP2024/073868 EP2024073868W WO2025051578A1 WO 2025051578 A1 WO2025051578 A1 WO 2025051578A1 EP 2024073868 W EP2024073868 W EP 2024073868W WO 2025051578 A1 WO2025051578 A1 WO 2025051578A1
Authority
WO
WIPO (PCT)
Prior art keywords
technical
data
medical device
implantable medical
event
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
PCT/EP2024/073868
Other languages
French (fr)
Inventor
Anna BEREZHKOVA
Franziska HENKE
Kirill PRAKHOV
Ferdinand Maximilian HESSELMANN
Benjamin PIETROWICZ
Stephan Bauer
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.)
Biotronik SE and Co KG
Original Assignee
Biotronik SE and Co KG
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 Biotronik SE and Co KG filed Critical Biotronik SE and Co KG
Publication of WO2025051578A1 publication Critical patent/WO2025051578A1/en
Anticipated expiration legal-status Critical
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment

Definitions

  • the invention relates to a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the invention relates to a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the invention relates to a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the object is solved by a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 1. Furthermore, the object is solved by a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 14.
  • the object is solved by a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 15.
  • the present invention provides a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the method comprises providing a first dataset comprising medical and/or technical data acquired by the implantable medical device, applying a machine learning algorithm such as XGBoost or logistic lasso regression to the first dataset for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device, and outputting a second dataset comprising a classification and/or a probability indication of the predicted and/or detected adverse technical event of the implantable medical device.
  • a machine learning algorithm such as XGBoost or logistic lasso regression
  • the present invention provides a computer implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the method comprises providing a first training dataset comprising medical and/or technical data acquired by the implantable medical device. Furthermore, the method comprises providing a second training dataset comprising a classification of a predicted and/or detected adverse technical event of the implantable medical device. The method moreover comprises training the machine learning algorithm by an optimization algorithm which calculates an extreme value of a loss function.
  • the present invention provides a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
  • the system comprises an implantable medical device (e.g. an IPG, ICD or ICM) configured to provide a first dataset comprising medical and/or technical data. Furthermore, the system comprises a training computation unit configured to apply a machine learning algorithm to the first dataset to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device.
  • an implantable medical device e.g. an IPG, ICD or ICM
  • a training computation unit configured to apply a machine learning algorithm to the first dataset to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device.
  • the system moreover comprises a data output unit configured to output a second dataset comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device.
  • An idea of the present invention is to enable an automated and regular analysis of home monitoring data from IPG/ICD/ICM devices to identify signals that could indicate an upcoming or acute technical event. There are thus no further manual efforts e.g. for the clinical staff. Due to the complexity of the home monitoring data, a powerful analysis approach using a machine learning algorithm is provided.
  • a “medical event” may also be considered.
  • the term “technical event” could be replaced by the term “medical event” or the term “technical/medical event” at one, several or any point in this description.
  • Machine learning algorithms are based on using statistical techniques to train a data processing system to perform a specific task without being explicitly programmed to do so.
  • the goal of machine learning is to construct algorithms that can learn from data and make predictions. These algorithms create mathematical models that can be used, for example, to classify data or to solve regression type problems.
  • the first dataset comprises a device status, a lead status, a patient status, recorded episodes and/or IPGS/SECGs (e.g. ECG pictures, machine- readable ECG data), therapy statistics and/or a device programming of the implantable medical device.
  • IPGS/SECGs e.g. ECG pictures, machine- readable ECG data
  • therapy statistics e.g. a device programming of the implantable medical device.
  • the first dataset comprises medical and/or technical data acquired by the implantable medical device during a first predetermined time period, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device, the first dataset comprises medical and/or technical data acquired by the implantable medical device during a second predetermined time period, preferably during the past 120 days.
  • the technical and/or medical data may be analyzed 90 days before a possible event and 30 days after.
  • a ratio of a number of data transmissions comprising medical and/or technical data sent by the implantable medical device during the first and/or second predetermined time period to a number of expected data transmissions comprising medical and/or technical data sent by the implantable medical device is equal to or greater than a predetermined threshold value, the data transmissions are stored, and wherein if said ratio is less than the predetermined threshold value, the data transmissions are deleted.
  • the medical and/or technical parameters comprised by the first dataset which contain no data, in particular are null, or contain no variance during the first and/or second predetermined time period, in particular are static, are deleted.
  • irrelevant data is deleted such that the analysis is performed only on useful data.
  • the medical and/or technical parameters comprised by the first dataset are pre-processed, wherein invalid data classified by predetermined escape values are set to null, wherein if several data transmissions are received in a day, respective parameter values are averaged, in particular for numerical parameters, or randomly picked, in particular for non-numeric parameters, and wherein all parameters whose completeness is less than a predetermined threshold value, in particular less than 60%, are deleted.
  • Said pre-processing thus provides a complete dataset free of invalid data.
  • any remaining data gaps of the medical and/or technical parameters are eliminated using linear interpolation between a previous and a next value, adoption of a previous value and/or taking an average value of a predetermined time period.
  • the first dataset comprises first data comprising, preferably pre-processed, raw medical and/or technical data acquired by the implantable medical device.
  • Said raw sensor data provides the most accurate information regarding medical and/or technical issues since no data processing has been performed on this data.
  • the first dataset comprises second data comprising medical and/or technical parameters having a variability exceeding a predetermined threshold value, wherein for every medical and/or technical parameter of said second data, a plurality of features are generated using a library for automated feature generation.
  • Said data having a high variability can potentially assist in identifying patterns and/or changes in medical and/or technical parameters that could be caused by technical issues.
  • the first dataset comprises third data comprising medical and/or technical parameters having a variability that is less than a predetermined threshold value, in particular static parameters such as programming settings, wherein for every medical and/or technical parameter of said third data, a plurality of features are generated using the library for automated feature generation. Said data having a low variability can potentially assist in identifying parameters that deviate from an expected value due to potential technical issues.
  • the first dataset comprises fourth data comprising medical and/or technical parameters, wherein the fourth data is a composition of the first data, the second data and the third data. Said fourth data thus combines the above- mentioned advantages of each of the other datasets.
  • the second dataset outputted by the machine learning algorithm comprises at least one class of an adverse technical event and/or an adverse medical event, a probability for said class and a feature importance of at least one medical and/or technical parameter associated with said class.
  • the upcoming adverse technical event and/or the acute adverse technical event of the implantable medical device is a malfunction or failure of a component and/or function of the implantable medical device, in particular a lead breakage, an inappropriate shock delivery of an ICD and/or a timely/promptly necessary (or an inadvertent) reprogramming of the implantable medical device. Said technical issue can thus be effectively predicted or detected by the method according to the present invention.
  • the herein described features of the computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device are also disclosed for the system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device and vice versa.
  • the input data for the model are the daily transmitted Home Monitoring messages.
  • the device status comprises parameters which describe the latest device status for IPG, ICD or ICM devices, e.g. battery fuel gauge and/or date of last device interrogation.
  • the lead status comprises parameters which describe the latest lead status for RA/RV/LV leads (e.g. LV sensing amplitude daily mean, LV sensing amplitude daily mean and/or LV pulse amplitude) or for ICM (e.g. sensing and/or pacing related measurements and/or statistics).
  • the patient status comprises parameters which describes the patient status including statistics to detected arrhythmias, e.g. mean atrial heart rate, mean ventricular heart rate, mean PVC/h, atrial burden, fast non-sustained VT per day and/or high atrial rate episodes per day.
  • Further parameter groups may be ‘recorded episodes’, ‘therapy statistics’ and/or “device programming’.
  • the recorded episodes comprise parameters to the detailed description of recorded and transmitted episodes, e.g. episode type, episode duration, episode start timestamp, episode end timestamp.
  • the therapy statistics comprise parameters which describe the therapy status and statistics to delivered therapies, e.g. atrial pacing % , BiV pacing %, ATP in VT zones started, ATP in VT zones successful, ATP one shot started and/or ATP one shot successful.
  • the device programming e.g. zone programming or therapy programming
  • the actual analysis time window depends on the type of event analysis.
  • the model distinguishes between event “predictions” and event “detections”. For prediction, the last 90 days before a possible event day are analyzed retrospectively, where day “-90D” corresponds to 90 days before the event and "0D” represents the day of the event. For event detection, the model uses an analysis window of 120 days, divided into 90 days before a possible event and 30 days after the possible event.
  • TSI Transmission Success Index; Number of received HM messages / Number of expected HM messages during transmission period. Datasets that do not meet this condition will be deleted. Afterwards the completeness & quality of the individual parameters will be evaluated. Parameters which include only NULL (empty) or STATIC (no variance during observational period) will be deleted.
  • Invalid data classified by certain escape values e g. INVALID, GREATER THAN MAX, FAILED
  • escape values e g. INVALID, GREATER THAN MAX, FAILED
  • the first data input set is the raw data. This means that the processed home monitoring data (from pre-processing) is imported directly into the model. There is no further processing of statistical characteristics or other advanced features.
  • An example of a home monitoring input parameter is "Thoracic Impedance Daily Mean".
  • the second data group is the creation of features for home monitoring parameters with high variability ("many unique"). This mainly concerns measured values.
  • An example feature is e.g. “the variance of the HM parameter "Thoracic Impedance Daily Mean”.
  • the third data group is the creation of features for home monitoring parameters with low variability ("few unique"). This mainly contains static parameters such as programming settings. The following features are calculated per HM parameter with low variability: [nunique] Number of unique values in the last Tookback_period' days.
  • the fourth data group is a composition of all raw Home Monitoring values, the “many unique” features and “few unique” features.
  • the model then uses the respective input data to detect signals in the data that are associated with the occurrence of a particular event.
  • the models which may/will be used are: Logistic Lasso Regression and/or XGBoost.
  • the output of the model is a probability that a certain event will occur on a certain day. The higher the value, the more likely that the event will occur.
  • the model outputs the feature importance (feature importance measure is “gain”: average gain across all splits where feature was used).
  • the harmonic mean of the area under the receiver-operating-characteristic and precision-recall curve is then calculated and the model quality is categorized as follows: High Signal > 0.85, Medium Signal > 0.70, Low Signal > 0.5.
  • Fig. 1 shows a flowchart of a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to a preferred embodiment of the invention
  • Fig. 2 shows a flowchart of a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to the preferred embodiment of the invention
  • Fig. 3 shows a diagram of a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to the preferred embodiment of the invention.
  • the computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 shown in Fig. 1 comprises providing SI a first dataset DS1 comprising medical and/or technical data acquired by the implantable medical device 10.
  • the method comprises applying S2 a machine learning algorithm A to the first dataset DS1 for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10, and outputting S3 a second dataset DS2 comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device 10.
  • the first dataset DS1 moreover comprises a device status, a lead status, a patient status, recorded episodes (e.g. IPGS/SECGs, ECG pictures, machine-readable ECG data), therapy statistics and/or a device programming of the implantable medical device 10.
  • recorded episodes e.g. IPGS/SECGs, ECG pictures, machine-readable ECG data
  • therapy statistics e.g. a device programming of the implantable medical device 10.
  • the device status are parameters which describe the latest device status for IPG, ICD or ICM devices such as Battery Fuel Gauge and Date of Last Device Interrogation.
  • the lead status are parameters which describe the latest lead status for RA/RV/LV leads or for ICM e.g., sensing and pacing related measurements and statistics such as LV Sensing Amplitude (Daily Mean), LV Sensing Amplitude (Daily Min) and LV Pulse Amplitude.
  • the patient status are parameters which describe the patient status including statistics to detected arrhythmias such as Mean Atrial Heart Rate, Mean Ventricular Heart Rate, Mean PVC/h, Atrial Burden, Fast non-sustained VT per Day and High Atrial Rate Episodes per Day.
  • Recorded episodes are parameters for the detailed description of recorded and transmitted episodes such as Episode Type, Episode Duration, Episode Start Timestamp and Episode End Timestamp.
  • Therapy statistics are parameters which describe the therapy status and statistics to delivered therapies such as Atrial Pacing Percentage, BiV Pacing Percentage, ATP in VT zones started, ATP in VT zones successful, ATP One Shot started, ATP One Shot successful, Device Programming parameters (including device programming such as zone programming, therapy programming etc.), AT/AF ATP type, Mode Switching Mode and AV Delay.
  • the first dataset DS1 comprises medical and/or technical data acquired by the implantable medical device 10 during a first predetermined time period Tl, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device 10, the first dataset DS1 comprises medical and/or technical data acquired by the implantable medical device 10 during a second predetermined time period T2, preferably during the past 120 days.
  • a ratio of a number of data transmissions comprising medical and/or technical data sent by the implantable medical device 10 during the first and/or second predetermined time period Tl, T2 to a number of expected data transmissions comprising medical and/or technical data sent by the implantable medical device 10 is equal to or greater than a predetermined threshold value, the data transmissions are stored, and wherein if said ratio is less than the predetermined threshold value, the data transmissions are deleted.
  • the medical and/or technical parameters comprised by the first dataset DS1 which contain no data, in particular are null, or contain no variance during the first and/or second predetermined time period Tl, T2, in particular are static, are deleted.
  • the medical and/or technical parameters comprised by the first dataset DS1 are pre- processed, wherein invalid data classified by predetermined escape values are set to null, wherein if several data transmissions are received in a day, respective parameter values are averaged, in particular for numerical parameters, or randomly picked, in particular for nonnumeric parameters, and wherein all parameters whose completeness is less than a predetermined threshold value, in particular less than 60%, are deleted.
  • the first dataset DS1 furthermore comprises first data DI comprising, preferably pre-processed, raw medical and/or technical data acquired by the implantable medical device 10.
  • the first dataset DS1 comprises second data D2 comprising medical and/or technical parameters having a variability exceeding a predetermined threshold value, wherein for every medical and/or technical parameter of said second data D2, a plurality of features are generated using a library for automated feature generation.
  • An example feature is e.g., the variance of the HM parameter "Thoracic Impedance Daily Mean".
  • the first dataset DS1 comprises third data D3 comprising medical and/or technical parameters having a variability that is less than a predetermined threshold value, in particular static parameters such as programming settings, wherein for every medical and/or technical parameter of said third data D3, a plurality of features are generated using the library for automated feature generation.
  • the following features are e.g., calculated per parameter with low variability.
  • the first dataset DS1 comprises fourth data D4 comprising medical and/or technical parameters, wherein the fourth data D4 is a composition of the first data, the second data D2 and the third data D3.
  • Said pre-processing is performed in the implantable medical device 10.
  • said pre-processing may be performed in a further device communicatively coupled to the implantable medical device 10 such as a patient communication device and/or a backend server located in an IT infrastructure of an IT service provider or medical service provider.
  • the second dataset DS2 outputted by the machine learning algorithm A comprises at least one class of an adverse technical event and/or an adverse medical event, a probability for said class and a feature importance of at least one medical and/or technical parameter associated with said class.
  • the upcoming adverse technical event and/or the acute adverse technical event of the implantable medical device 10 may be a malfunction or failure of a component and/or function of the implantable medical device 10, in particular a lead breakage, an inappropriate shock delivery of an ICD and/or an inadvertent reprogramming of the implantable medical device 10.
  • the harmonic mean of the area under the receiver-operating-characteristic and precision-recall curve is then calculated and the model quality is categorized e.g., as follows: high signal > 0.85, medium signal > 0.70 and low signal > 0.5.
  • Fig. 2 shows a flowchart of a computer-implemented method for providing a machine learning algorithm A for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 according to the preferred embodiment of the invention.
  • the method comprises providing SI’ a first training dataset TD1 comprising medical and/or technical data acquired by the implantable medical device 10, providing S2’ a second training dataset TD2 comprising a classification of a predicted and/or detected adverse technical event of the implantable medical device 10, and training S3’ the machine learning algorithm A by an optimization algorithm which calculates an extreme value of a loss function.
  • the first training dataset TD1 comprises medical and/or technical data acquired by the implantable medical device 10 during a first predetermined time period Tl, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device 10, the first training dataset TD1 comprises medical and/or technical data acquired by the implantable medical device 10 during a second predetermined time period, preferably during the past 120 days divided into 90 days before a possible event and 30 days after the possible event.
  • Fig.3 shows a diagram of a system 1 for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 according to the preferred embodiment of the invention.
  • the system 1 comprises an implantable medical device 10 configured to provide a first dataset DS1 comprising medical and/or technical data.
  • the system 1 comprises a training computation unit 12 configured to apply a machine learning algorithm A to the first dataset DS1 to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device 10, and a data output unit 14 configured to output a second dataset DS2 comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device 10.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Business, Economics & Management (AREA)
  • General Business, Economics & Management (AREA)
  • Electrotherapy Devices (AREA)

Abstract

The invention relates to a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10), comprising the steps of providing a first dataset (DS1) comprising medical and/or technical data acquired by the implantable medical device (10), applying a machine learning algorithm (A) to the first dataset (DS1) for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10) and outputting a second dataset (DS2) comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device (10). In addition, the invention relates to a computer-implemented method for providing a machine learning algorithm (A) and to a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10).

Description

Method and system for predicting an adverse technical and/or medical event
The invention relates to a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
Furthermore, the invention relates to a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
Moreover, the invention relates to a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
Many diseases of the heart are accompanied by changes in the data acquired by an implantable medical device. Said changes in the data could be for medical or technical reasons, e.g. due to possibly malfunctioning components of the implantable medical device. If a functionality of the implantable medical device is impaired this could in turn lead to an adverse medical event.
It is therefore an object of the present invention to provide a method and system for predicting an adverse technical and/or medical event, in particular to predict an adverse technical event and/or detecting an acute adverse technical event such that an adverse medical event can be prevented.
The object is solved by a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 1. Furthermore, the object is solved by a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 14.
Moreover, the object is solved by a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device having the features of claim 15.
Further developments and advantageous embodiments are defined in the dependent claims.
The present invention provides a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
The method comprises providing a first dataset comprising medical and/or technical data acquired by the implantable medical device, applying a machine learning algorithm such as XGBoost or logistic lasso regression to the first dataset for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device, and outputting a second dataset comprising a classification and/or a probability indication of the predicted and/or detected adverse technical event of the implantable medical device.
Furthermore, the present invention provides a computer implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
The method comprises providing a first training dataset comprising medical and/or technical data acquired by the implantable medical device. Furthermore, the method comprises providing a second training dataset comprising a classification of a predicted and/or detected adverse technical event of the implantable medical device. The method moreover comprises training the machine learning algorithm by an optimization algorithm which calculates an extreme value of a loss function. In addition, the present invention provides a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device.
The system comprises an implantable medical device (e.g. an IPG, ICD or ICM) configured to provide a first dataset comprising medical and/or technical data. Furthermore, the system comprises a training computation unit configured to apply a machine learning algorithm to the first dataset to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device.
The system moreover comprises a data output unit configured to output a second dataset comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device.
An idea of the present invention is to enable an automated and regular analysis of home monitoring data from IPG/ICD/ICM devices to identify signals that could indicate an upcoming or acute technical event. There are thus no further manual efforts e.g. for the clinical staff. Due to the complexity of the home monitoring data, a powerful analysis approach using a machine learning algorithm is provided.
Instead of or in addition to a "technical event", a "medical event" may also be considered. Thus, the term "technical event" could be replaced by the term "medical event" or the term "technical/medical event" at one, several or any point in this description.
Machine learning algorithms are based on using statistical techniques to train a data processing system to perform a specific task without being explicitly programmed to do so. The goal of machine learning is to construct algorithms that can learn from data and make predictions. These algorithms create mathematical models that can be used, for example, to classify data or to solve regression type problems.
According to an aspect of the invention, the first dataset comprises a device status, a lead status, a patient status, recorded episodes and/or IPGS/SECGs (e.g. ECG pictures, machine- readable ECG data), therapy statistics and/or a device programming of the implantable medical device. By using a wide range of parameters, an upcoming adverse technical event can be more likely predicted and/or an acute adverse technical event effectively detected.
According to a further aspect of the invention, for predicting an upcoming adverse technical event of the implantable medical device, the first dataset comprises medical and/or technical data acquired by the implantable medical device during a first predetermined time period, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device, the first dataset comprises medical and/or technical data acquired by the implantable medical device during a second predetermined time period, preferably during the past 120 days. The technical and/or medical data may be analyzed 90 days before a possible event and 30 days after. By analyzing not only a single dataset acquired at a particular point in time, e.g. at present, evolving patterns and/or changes in specific parameters can be effectively tracked.
According to a further aspect of the invention, if a ratio of a number of data transmissions comprising medical and/or technical data sent by the implantable medical device during the first and/or second predetermined time period to a number of expected data transmissions comprising medical and/or technical data sent by the implantable medical device is equal to or greater than a predetermined threshold value, the data transmissions are stored, and wherein if said ratio is less than the predetermined threshold value, the data transmissions are deleted. This way the dataset used as a basis for analysis fulfills the defined criteria for completeness.
According to a further aspect of the invention, the medical and/or technical parameters comprised by the first dataset which contain no data, in particular are null, or contain no variance during the first and/or second predetermined time period, in particular are static, are deleted. Hence, irrelevant data is deleted such that the analysis is performed only on useful data.
According to a further aspect of the invention, the medical and/or technical parameters comprised by the first dataset are pre-processed, wherein invalid data classified by predetermined escape values are set to null, wherein if several data transmissions are received in a day, respective parameter values are averaged, in particular for numerical parameters, or randomly picked, in particular for non-numeric parameters, and wherein all parameters whose completeness is less than a predetermined threshold value, in particular less than 60%, are deleted. Said pre-processing thus provides a complete dataset free of invalid data.
According to a further aspect of the invention, any remaining data gaps of the medical and/or technical parameters are eliminated using linear interpolation between a previous and a next value, adoption of a previous value and/or taking an average value of a predetermined time period. By eliminating said data gaps, the analysis is more accurate since invalid and/or missing data does not influence said analysis.
According to a further aspect of the invention, the first dataset comprises first data comprising, preferably pre-processed, raw medical and/or technical data acquired by the implantable medical device. Said raw sensor data provides the most accurate information regarding medical and/or technical issues since no data processing has been performed on this data.
According to a further aspect of the invention, the first dataset comprises second data comprising medical and/or technical parameters having a variability exceeding a predetermined threshold value, wherein for every medical and/or technical parameter of said second data, a plurality of features are generated using a library for automated feature generation. Said data having a high variability can potentially assist in identifying patterns and/or changes in medical and/or technical parameters that could be caused by technical issues.
According to a further aspect of the invention, the first dataset comprises third data comprising medical and/or technical parameters having a variability that is less than a predetermined threshold value, in particular static parameters such as programming settings, wherein for every medical and/or technical parameter of said third data, a plurality of features are generated using the library for automated feature generation. Said data having a low variability can potentially assist in identifying parameters that deviate from an expected value due to potential technical issues. According to a further aspect of the invention, the first dataset comprises fourth data comprising medical and/or technical parameters, wherein the fourth data is a composition of the first data, the second data and the third data. Said fourth data thus combines the above- mentioned advantages of each of the other datasets.
According to a further aspect of the invention, the second dataset outputted by the machine learning algorithm comprises at least one class of an adverse technical event and/or an adverse medical event, a probability for said class and a feature importance of at least one medical and/or technical parameter associated with said class.
Thus, not only a probability for a particular class is outputted but also a feature importance, i.e. which parameter(s) of the input data contributed most to the prediction or detection result.
According to a further aspect of the invention, the upcoming adverse technical event and/or the acute adverse technical event of the implantable medical device is a malfunction or failure of a component and/or function of the implantable medical device, in particular a lead breakage, an inappropriate shock delivery of an ICD and/or a timely/promptly necessary (or an inadvertent) reprogramming of the implantable medical device. Said technical issue can thus be effectively predicted or detected by the method according to the present invention.
The herein described features of the computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device are also disclosed for the system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device and vice versa.
Additional information on system & patient status:
The input data for the model are the daily transmitted Home Monitoring messages. For the model we are focusing on the parameter groups ‘device status’, ‘lead status’ and ‘patient status’. The device status comprises parameters which describe the latest device status for IPG, ICD or ICM devices, e.g. battery fuel gauge and/or date of last device interrogation. The lead status comprises parameters which describe the latest lead status for RA/RV/LV leads (e.g. LV sensing amplitude daily mean, LV sensing amplitude daily mean and/or LV pulse amplitude) or for ICM (e.g. sensing and/or pacing related measurements and/or statistics). The patient status comprises parameters which describes the patient status including statistics to detected arrhythmias, e.g. mean atrial heart rate, mean ventricular heart rate, mean PVC/h, atrial burden, fast non-sustained VT per day and/or high atrial rate episodes per day.
Further parameter groups may be ‘recorded episodes’, ‘therapy statistics’ and/or “device programming’. The recorded episodes comprise parameters to the detailed description of recorded and transmitted episodes, e.g. episode type, episode duration, episode start timestamp, episode end timestamp. The therapy statistics comprise parameters which describe the therapy status and statistics to delivered therapies, e.g. atrial pacing % , BiV pacing %, ATP in VT zones started, ATP in VT zones successful, ATP one shot started and/or ATP one shot successful. The device programming (e.g. zone programming or therapy programming) comprises device programming parameters AT/AF ATP type, mode switching mode and/or AV Delay 1.
Additional data pre-processing details:
The actual analysis time window depends on the type of event analysis. The model distinguishes between event "predictions" and event "detections". For prediction, the last 90 days before a possible event day are analyzed retrospectively, where day "-90D" corresponds to 90 days before the event and "0D" represents the day of the event. For event detection, the model uses an analysis window of 120 days, divided into 90 days before a possible event and 30 days after the possible event.
Then, the general home monitoring performance is analyzed in these analysis time windows. Both analysis tasks (prediction & detection) require a general overall Home Monitoring performance of equal or higher than 80% TSI during the observational period (TSI = Transmission Success Index; Number of received HM messages / Number of expected HM messages during transmission period). Datasets that do not meet this condition will be deleted. Afterwards the completeness & quality of the individual parameters will be evaluated. Parameters which include only NULL (empty) or STATIC (no variance during observational period) will be deleted.
Afterwards parameters with be pre-processed. Invalid data classified by certain escape values (e g. INVALID, GREATER THAN MAX, FAILED) are set to NULL. If several messages have arrived on one day, the respective parameter values are averaged (for numerical) or randomly picked (for non-numeric parameters). All parameters whose completeness is less than 60% are removed.
Any remaining data gaps will be finally eliminated using different strategies (linear interpolation between the previous and next value / adoption of the previous value / average value of the last 7 days).
Additional information on feature engineering process & model input parameters, wherein the tool builds 4 different input datasets for the model:
The first data input set is the raw data. This means that the processed home monitoring data (from pre-processing) is imported directly into the model. There is no further processing of statistical characteristics or other advanced features. An example of a home monitoring input parameter is "Thoracic Impedance Daily Mean".
The second data group is the creation of features for home monitoring parameters with high variability ("many unique"). This mainly concerns measured values. We use a standard library for automated feature generation to calculate the features. An example feature is e.g. “the variance of the HM parameter "Thoracic Impedance Daily Mean".
The third data group is the creation of features for home monitoring parameters with low variability ("few unique"). This mainly contains static parameters such as programming settings. The following features are calculated per HM parameter with low variability: [nunique] Number of unique values in the last Tookback_period' days.
[curr freq] Frequency of current value in the last Tookback_period' days. [days since change] Number of days since last change. Note: Feature is zero up until the first change.
[var] Variance over the last 'lookback_period' days.
[diff] First order difference. Note: Feature is zero on the first day.
[diff avg] Average of first order differences of the last 'lookback_period' days.
[mode dev] Deviation from all-time mode.
The fourth data group is a composition of all raw Home Monitoring values, the “many unique” features and “few unique” features.
Additional information on model output parameters & model accuracy assessment:
The model then uses the respective input data to detect signals in the data that are associated with the occurrence of a particular event. The models which may/will be used are: Logistic Lasso Regression and/or XGBoost.
The output of the model is a probability that a certain event will occur on a certain day. The higher the value, the more likely that the event will occur. In addition, the model outputs the feature importance (feature importance measure is “gain”: average gain across all splits where feature was used).
For further evaluation of the overall performance of the models, the harmonic mean of the area under the receiver-operating-characteristic and precision-recall curve is then calculated and the model quality is categorized as follows: High Signal > 0.85, Medium Signal > 0.70, Low Signal > 0.5.
For a more complete understanding of the present invention and advantages thereof, reference is now made to the following description taken in conjunction with the accompanying drawings. The invention is explained in more detail below using exemplary embodiments, which are specified in the schematic figures of the drawings, in which:
Fig. 1 shows a flowchart of a computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to a preferred embodiment of the invention;
Fig. 2 shows a flowchart of a computer-implemented method for providing a machine learning algorithm for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to the preferred embodiment of the invention; and
Fig. 3 shows a diagram of a system for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device according to the preferred embodiment of the invention.
The computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 shown in Fig. 1 comprises providing SI a first dataset DS1 comprising medical and/or technical data acquired by the implantable medical device 10.
Furthermore, the method comprises applying S2 a machine learning algorithm A to the first dataset DS1 for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10, and outputting S3 a second dataset DS2 comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device 10.
The first dataset DS1 moreover comprises a device status, a lead status, a patient status, recorded episodes (e.g. IPGS/SECGs, ECG pictures, machine-readable ECG data), therapy statistics and/or a device programming of the implantable medical device 10.
The device status are parameters which describe the latest device status for IPG, ICD or ICM devices such as Battery Fuel Gauge and Date of Last Device Interrogation. The lead status are parameters which describe the latest lead status for RA/RV/LV leads or for ICM e.g., sensing and pacing related measurements and statistics such as LV Sensing Amplitude (Daily Mean), LV Sensing Amplitude (Daily Min) and LV Pulse Amplitude. The patient status are parameters which describe the patient status including statistics to detected arrhythmias such as Mean Atrial Heart Rate, Mean Ventricular Heart Rate, Mean PVC/h, Atrial Burden, Fast non-sustained VT per Day and High Atrial Rate Episodes per Day. Recorded episodes are parameters for the detailed description of recorded and transmitted episodes such as Episode Type, Episode Duration, Episode Start Timestamp and Episode End Timestamp.
Therapy statistics are parameters which describe the therapy status and statistics to delivered therapies such as Atrial Pacing Percentage, BiV Pacing Percentage, ATP in VT zones started, ATP in VT zones successful, ATP One Shot started, ATP One Shot successful, Device Programming parameters (including device programming such as zone programming, therapy programming etc.), AT/AF ATP type, Mode Switching Mode and AV Delay.
For predicting an upcoming adverse technical event of the implantable medical device 10, the first dataset DS1 comprises medical and/or technical data acquired by the implantable medical device 10 during a first predetermined time period Tl, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device 10, the first dataset DS1 comprises medical and/or technical data acquired by the implantable medical device 10 during a second predetermined time period T2, preferably during the past 120 days.
If a ratio of a number of data transmissions comprising medical and/or technical data sent by the implantable medical device 10 during the first and/or second predetermined time period Tl, T2 to a number of expected data transmissions comprising medical and/or technical data sent by the implantable medical device 10 is equal to or greater than a predetermined threshold value, the data transmissions are stored, and wherein if said ratio is less than the predetermined threshold value, the data transmissions are deleted.
Furthermore, the medical and/or technical parameters comprised by the first dataset DS1 which contain no data, in particular are null, or contain no variance during the first and/or second predetermined time period Tl, T2, in particular are static, are deleted.
The medical and/or technical parameters comprised by the first dataset DS1 are pre- processed, wherein invalid data classified by predetermined escape values are set to null, wherein if several data transmissions are received in a day, respective parameter values are averaged, in particular for numerical parameters, or randomly picked, in particular for nonnumeric parameters, and wherein all parameters whose completeness is less than a predetermined threshold value, in particular less than 60%, are deleted.
Any remaining data gaps of the medical and/or technical parameters are eliminated using linear interpolation between a previous and a next value, adoption of a previous value and/or taking an average value of a predetermined time period. The first dataset DS1 furthermore comprises first data DI comprising, preferably pre-processed, raw medical and/or technical data acquired by the implantable medical device 10.
The first dataset DS1 comprises second data D2 comprising medical and/or technical parameters having a variability exceeding a predetermined threshold value, wherein for every medical and/or technical parameter of said second data D2, a plurality of features are generated using a library for automated feature generation. An example feature is e.g., the variance of the HM parameter "Thoracic Impedance Daily Mean".
The first dataset DS1 comprises third data D3 comprising medical and/or technical parameters having a variability that is less than a predetermined threshold value, in particular static parameters such as programming settings, wherein for every medical and/or technical parameter of said third data D3, a plurality of features are generated using the library for automated feature generation.
The following features are e.g., calculated per parameter with low variability. A number of unique values in the last Tookback_period' days, a frequency of current value in the last Tookback_period' days, a number of days since a last change, wherein a feature is zero up until the first change, a variance over the last Tookback_period' days, a first order difference, wherein a feature is zero on the first day, an average of first order differences of the last Tookback_period' days and a deviation from all-time mode.
Moreover, the first dataset DS1 comprises fourth data D4 comprising medical and/or technical parameters, wherein the fourth data D4 is a composition of the first data, the second data D2 and the third data D3. Said pre-processing is performed in the implantable medical device 10. Alternatively or in addition, said pre-processing may be performed in a further device communicatively coupled to the implantable medical device 10 such as a patient communication device and/or a backend server located in an IT infrastructure of an IT service provider or medical service provider.
In addition, the second dataset DS2 outputted by the machine learning algorithm A comprises at least one class of an adverse technical event and/or an adverse medical event, a probability for said class and a feature importance of at least one medical and/or technical parameter associated with said class. Further, the upcoming adverse technical event and/or the acute adverse technical event of the implantable medical device 10 may be a malfunction or failure of a component and/or function of the implantable medical device 10, in particular a lead breakage, an inappropriate shock delivery of an ICD and/or an inadvertent reprogramming of the implantable medical device 10.
For further evaluation of the overall performance of the machine learning algorithm A, the harmonic mean of the area under the receiver-operating-characteristic and precision-recall curve is then calculated and the model quality is categorized e.g., as follows: high signal > 0.85, medium signal > 0.70 and low signal > 0.5.
Fig. 2 shows a flowchart of a computer-implemented method for providing a machine learning algorithm A for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 according to the preferred embodiment of the invention.
The method comprises providing SI’ a first training dataset TD1 comprising medical and/or technical data acquired by the implantable medical device 10, providing S2’ a second training dataset TD2 comprising a classification of a predicted and/or detected adverse technical event of the implantable medical device 10, and training S3’ the machine learning algorithm A by an optimization algorithm which calculates an extreme value of a loss function. In particular, the first training dataset TD1 comprises medical and/or technical data acquired by the implantable medical device 10 during a first predetermined time period Tl, preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device 10, the first training dataset TD1 comprises medical and/or technical data acquired by the implantable medical device 10 during a second predetermined time period, preferably during the past 120 days divided into 90 days before a possible event and 30 days after the possible event.
Fig.3 shows a diagram of a system 1 for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device 10 according to the preferred embodiment of the invention.
The system 1 comprises an implantable medical device 10 configured to provide a first dataset DS1 comprising medical and/or technical data.
Furthermore, the system 1 comprises a training computation unit 12 configured to apply a machine learning algorithm A to the first dataset DS1 to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device 10, and a data output unit 14 configured to output a second dataset DS2 comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device 10.
Reference Signs
1 system
10 implantable medical device
12 training computation unit
14 data output unit
A machine learning algorithm
DI first data
D2 second data
D3 third data
D4 fourth data
DS1 first dataset
DS2 second dataset
TD1 first training dataset
TD2 second training dataset
SI -S3 method steps
S 1’ -S3 ’ training method steps
T1 first predetermined time period
T2 second predetermined time period

Claims

Claims
1. Computer-implemented method for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10), comprising the steps of: providing (SI) a first dataset (DS1) comprising medical and/or technical data acquired by the implantable medical device (10); applying (S2) a machine learning algorithm (A) to the first dataset (DS1) for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10); and outputting (S3) a second dataset (DS2) comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device (10).
2. Computer-implemented method of claim 1, wherein the first dataset (DS1) comprises a device status, a lead status, a patient status, recorded episodes, therapy statistics and/or a device programming of the implantable medical device (10).
3. Computer-implemented method of claim 1 or 2, wherein for predicting an upcoming adverse technical event of the implantable medical device (10), the first dataset (DS1) comprises medical and/or technical data acquired by the implantable medical device (10) during a first predetermined time period (Tl), preferably during the past 90 days, and wherein for detecting an acute adverse technical event of the implantable medical device (10), the first dataset (DS1) comprises medical and/or technical data acquired by the implantable medical device (10) during a second predetermined time period (T2), preferably during the past 120 days.
4. Computer-implemented method of claim 3, wherein if a ratio of a number of data transmissions comprising medical and/or technical data sent by the implantable medical device (10) during the first and/or second predetermined time period (Tl, T2) to a number of expected data transmissions comprising medical and/or technical data sent by the implantable medical device (10) is equal to or greater than a predetermined threshold value, the data transmissions are stored, and wherein if said ratio is less than the predetermined threshold value, the data transmissions are deleted.
5. Computer-implemented method of claim 3 or 4, wherein the medical and/or technical parameters comprised by the first dataset (DS1) which contain no data, in particular are null, or contain no variance during the first and/or second predetermined time period (Tl, T2), in particular are static, are deleted.
6. Computer-implemented method of any one of claims 3 to 5, wherein the medical and/or technical parameters comprised by the first dataset (DS1) are pre-processed, wherein invalid data classified by predetermined escape values are set to null, wherein if several data transmissions are received in a day, respective parameter values are averaged, in particular for numerical parameters, or randomly picked, in particular for non-numeric parameters, and wherein all parameters whose completeness is less than a predetermined threshold value, in particular less than 60%, are deleted.
7. Computer-implemented method of claim 6, wherein any remaining data gaps of the medical and/or technical parameters are eliminated using linear interpolation between a previous and a next value, adoption of a previous value and/or taking an average value of a predetermined time period.
8. Computer-implemented method of any one of the preceding claims, wherein the first dataset (DS1) comprises first data (DI) comprising, preferably pre-processed, raw medical and/or technical data acquired by the implantable medical device (10).
9. Computer-implemented method of any one of the preceding claims, wherein the first dataset (DS1) comprises second data (D2) comprising medical and/or technical parameters having a variability exceeding a predetermined threshold value, wherein for every medical and/or technical parameter of said second data (D2), a plurality of features are generated using a library for automated feature generation.
10. Computer-implemented method of claim 9, wherein the first dataset (DS1) comprises third data (D3) comprising medical and/or technical parameters having a variability that is less than a predetermined threshold value, in particular static parameters such as programming settings, wherein for every medical and/or technical parameter of said third data (D3), a plurality of features are generated using the library for automated feature generation.
11. Computer-implemented method of any one of the preceding claims, wherein the first dataset (DS1) comprises fourth data (D4) comprising medical and/or technical parameters, wherein the fourth data (D4) is a composition of the first data, the second data (D2) and the third data (D3).
12. Computer-implemented method of any one of the preceding claims, wherein the second dataset (DS2) outputted by the machine learning algorithm (A) comprises at least one class of an adverse technical event and/or an adverse medical event, a probability for said class and a feature importance of at least one medical and/or technical parameter associated with said class.
13. Computer-implemented method of any one of the preceding claims, wherein the upcoming adverse technical event and/or the acute adverse technical event of the implantable medical device (10) is a malfunction or failure of a component and/or function of the implantable medical device (10), in particular a lead breakage, an inappropriate shock delivery of an ICD and/or an inadvertent reprogramming of the implantable medical device (10).
14. Computer-implemented method for providing a machine learning algorithm (A) for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10), comprising the steps of: providing (ST) a first training dataset (TD1) comprising medical and/or technical data acquired by the implantable medical device (10); providing (S2’) a second training dataset (TD2) comprising a classification of a predicted and/or detected adverse technical event of the implantable medical device (io); training (S3’) the machine learning algorithm (A) by an optimization algorithm which calculates an extreme value of a loss function.
15. System (1) for predicting an upcoming adverse technical event and/or for detecting an acute adverse technical event of the implantable medical device (10), comprising: an implantable medical device (10) configured to provide a first dataset (DS1) comprising medical and/or technical data; a training computation unit (12) configured to apply a machine learning algorithm (A) to the first dataset (DS1) to predict an upcoming adverse technical event and/or to detect an acute adverse technical event of the implantable medical device (10); and a data output unit (14) configured to output a second dataset (DS2) comprising a classification of the predicted and/or detected adverse technical event of the implantable medical device (10).
PCT/EP2024/073868 2023-09-05 2024-08-27 Method and system for predicting an adverse technical and/or medical event Pending WO2025051578A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP23195275.5 2023-09-05
EP23195275 2023-09-05

Publications (1)

Publication Number Publication Date
WO2025051578A1 true WO2025051578A1 (en) 2025-03-13

Family

ID=87933789

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2024/073868 Pending WO2025051578A1 (en) 2023-09-05 2024-08-27 Method and system for predicting an adverse technical and/or medical event

Country Status (1)

Country Link
WO (1) WO2025051578A1 (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110313691A1 (en) * 2010-06-22 2011-12-22 WorldHeart, Inc. Direct sequence spread spectrum predictive cable and component failure technology
JP5346318B2 (en) * 2002-12-17 2013-11-20 カーディアック ペースメイカーズ， インコーポレイテッド Communication repeater device for implantable medical devices
US20230009416A1 (en) * 2019-12-11 2023-01-12 Glaucia Pereira Novel nanotechnology-driven prototypes for ai-enriched biocompatible prosthetics following either risk of organ failure or moderate to severe impairment

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5346318B2 (en) * 2002-12-17 2013-11-20 カーディアック ペースメイカーズ， インコーポレイテッド Communication repeater device for implantable medical devices
US20110313691A1 (en) * 2010-06-22 2011-12-22 WorldHeart, Inc. Direct sequence spread spectrum predictive cable and component failure technology
US20230009416A1 (en) * 2019-12-11 2023-01-12 Glaucia Pereira Novel nanotechnology-driven prototypes for ai-enriched biocompatible prosthetics following either risk of organ failure or moderate to severe impairment

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ELSAID OSSAMA ET AL: "Total laser cycles-a measure of transvenous lead extraction difficulty", JOURNAL OF INTERVENTIONAL CARDIAC ELECTROPHYSIOLOGY, SPRINGER NEW YORK LLC, US, vol. 53, no. 3, 16 August 2018 (2018-08-16), pages 383 - 389, XP036658989, ISSN: 1383-875X, [retrieved on 20180816], DOI: 10.1007/S10840-018-0422-3 *
YING XU ET AL: "Prognostics and Health Management of Electromedical Equipment Lithium Battery", ARTIFICIAL INTELLIGENCE AND COMPUTER SCIENCE, ACM, 2 PENN PLAZA, SUITE 701NEW YORKNY10121-0701USA, 12 July 2019 (2019-07-12), pages 354 - 357, XP058441664, ISBN: 978-1-4503-7150-6, DOI: 10.1145/3349341.3349430 *

Similar Documents

Publication Publication Date Title
US7289857B2 (en) Data analysis system and method
CN117594253A (en) Personnel health early warning method and system based on intelligent monitoring equipment
CN120298098B (en) AI-based Big Data Risk Assessment System
CN115238828A (en) Chromatograph fault monitoring method and device
CN116153505B (en) Intelligent critical patient sign identification method and system based on medical pressure sensor
CN109567747A (en) Sleep monitor method, apparatus, computer equipment and storage medium
CN118228130B (en) Monitoring method, system and storage medium based on equipment health state
CN120388733A (en) A method and system for early detection of chronic diseases based on a multimodal large model
CN120370205B (en) DC charging pile insulation monitoring method and system
EP3502979A1 (en) A probability-based detector and controller apparatus, method, computer program
CN117373604A (en) Method, system and medium for quickly looking up health report
CN113112043A (en) Method, device and equipment for determining abnormal resource transfer condition
WO2025051578A1 (en) Method and system for predicting an adverse technical and/or medical event
US9183352B2 (en) Method and arrangement for predicting at least one system event, corresponding computer program, and corresponding computer-readable storage medium
CN119626493A (en) Hemodialysis fistula puncture needle consumables management system and method
US11789439B2 (en) Failure sign diagnosis device and method therefor
WO2025051577A1 (en) Method and system for predicting an adverse patient clinical event
US20210397992A1 (en) Inference apparatus, information processing apparatus, inference method, program and recording medium
CN116662904B (en) Data type variation detection method, device, computer equipment and medium
CN112168188A (en) Processing method and device for pressure detection data
US11379801B2 (en) Maintenance support system and maintenance support method
CN119714389B (en) Automated sensor performance monitoring and early warning system
CN116189868B (en) Medical equipment management method and system based on big data
CN120950895B (en) Pulse signal intelligent switching control method and system based on neural network
US12414735B2 (en) Method for detecting epileptic and psychogenic seizures

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 24761634

Country of ref document: EP

Kind code of ref document: A1