Detection of a risk of cardiac decompensation in a patient suffering from heart failure

Abstract

Title of the invention: Detection of risk of cardiac decompensation in a patient with heart failure. The invention relates to a method for the early detection of risk of cardiac decompensation in a patient with heart failure, wherein several measurements of the patient's weight are taken and then transmitted to processor-based processing means where they are processed using a predefined mathematical model to derive a risk score for cardiac decompensation of said patient. A cardiac decompensation alert is triggered when the risk score exceeds a predetermined alert threshold. Abstract figure: Fig. 1

Description

Title of the invention: Detection of a risk of cardiac decompensation in a patient suffering from heart failure

[0001] The invention relates to a method and a system for detecting a risk of cardiac decompensation in a patient suffering from heart failure.

[0002] Chronic heart failure (CHF) is a serious illness in which the heart becomes short of breath and can no longer perform its role as a blood pump, sending blood throughout the body, including the organs. It can develop into complications such as acute heart failure, acute pulmonary or peripheral edema, or even cardiac decompensation.

[0003] In the case of cardiac decompensation (CD), the body can no longer compensate for the negative effects of HF and other organs may then be affected. For example, a left-sided CD impacts the lungs more (dyspnea), while a right-sided CD impacts the liver and a global CD, i.e. of the whole heart, causes palpitations, discomfort, drops in blood pressure, edema of the lower limbs, etc.

[0004] In all cases, a heart attack requires rapid emergency intervention, generally involving hospitalization of the patient. People suffering from heart failure, i.e. patients, must therefore be monitored in order to quickly detect any risk of cardiac decompensation.

[0005] However, such monitoring is not easy to carry out because patients suffering from IC are generally outside the hospital, typically at home, when a DC occurs and it is easy to understand that it is not possible to permanently place a nurse behind each patient.

[0006] Furthermore, it is more efficient to be able to intervene and treat the patient before the DC occurs rather than after it has been triggered, for obvious patient safety reasons.

[0007] In this context, the problem is to propose a method, i.e. a process, and a system for remotely monitoring one or more patients suffering from heart failure in order to be able to anticipate the occurrence of a heart attack in this patient and to warn the healthcare staff early to enable them to intervene upon detection and thus avoid complications in the patient and / or their hospitalization, or even a fatal risk.

[0008] The invention relates to a method, i.e. a process, for early detection of a risk of cardiac decompensation in a patient suffering from heart failure comprising the steps of:

[0009] a. Take several measurements of the patient's weight, b. transmit the weight measurements to processor processing means, c. Processing the weight measurements within the processor processing means using a predefined mathematical model to deduce a cardiac decompensation risk score for said patient, and d. Trigger a cardiac decompensation alert when the risk score exceeds a pre-set alert threshold.

[0010] Depending on the embodiment considered, the method of the invention may comprise one or more of the following characteristics:

[0011] - the patient's weight measurements obtained in step a), the given alert threshold and the mathematical model are stored in storage means. - the storage means comprise at least one computer memory, for example a hard disk, an SSD disk, an S3 storage or other. - the microprocessor processing means include (at least) a computer server. - the computer server(s) is a local server, i.e. arranged on site, an embedded (or edge) or remote (or cloud) server. - the weight measurements are transmitted electronically in step b). - the patient's weight measurements are carried out using a scale, by example a connected scale including means of remote transmission of measurements. - the patient's weight measurements are taken over several days, preferably at least 3 days, more preferably at least 5 days, typically at least 7 days. - in step c), the mathematical model includes sub-models each obtained (i.e. based on) from historical weight data and decompensation events from several patient samples from the same group of patients. - historical weight data and decompensation events of a group of patients are stored, preferably stored in a correlated manner. - in step c), the weight measurements are processed within each sub-model to determine a risk score per sub-model between 0 and 1 and to deduce therefrom an average risk score equal to the average of the risk scores per sub-model having been determined. - in step d), the cardiac decompensation alert is triggered when the average risk score exceeds the given alert threshold. Preferably, the appropriate alert threshold can be set by a healthcare professional to ensure a good compromise between false alerts and non-detection of risk, taking into account the occurrence of events. - decompensation events include unplanned medical consultations and hospitalizations for heart failure of patients. - in step d), triggering a cardiac decompensation alert comprises a display on display means of at least the cardiac decompensation risk score of patients. - each sub-model is configured to determine a risk score between 0 and 1 from historical weight data and decompensation events from a patient sample, in particular by:

[0012] i. Extraction and prioritization of several statistical characteristics obtained from historical weight data and decompensation events, and their frequency of occurrence, ii. Determination of at least one decision threshold level for each of the statistical characteristics to create at least one condition associated with each exceedance of the decision threshold, iii. Analysis of the conditions following a hierarchical order defined according to the prioritization of the statistical characteristics of step i), iv. Defining a risk score based on the frequencies of occurrence of decompensation events for said statistical characteristics and said decision threshold levels, and v. Development of a decision rule based on the satisfaction of several conditions representative of a risk of decompensation from the risk score defined in step iv).

[0013] - statistical characteristics include deviation (in %), standard deviation, variance, trend, deviation between minimum and maximum weight value over several days (typically 2 to 5 days), absolute mean deviation, kurtosis coefficient, symmetry, activity, mobility and complexity. - in step d), triggering a cardiac decompensation alert comprises a display on display means of at least the patient's cardiac decompensation risk score and at least one of said statistical characteristics, preferably a difference between the minimum and maximum weight values, and a trend. - each sub-model is configured to determine the risk score based on at least one patient symptom (dyspnea, orthopnea, cough, edema, fatigue) and / or additional data (age, sex, place of residence, medical history, etc.) provided by the patient using a questionnaire or other means. - processing of weight measurements includes filtering of outlier weight values ​​and / or missing values. - the patient is an adult or an adolescent. - the processor processing means comprise one or more microprocessors. - the display means comprise at least one display screen, for example a computer screen, digital tablet or multifunction telephone (smartphone). - the display means are located in a remote monitoring center, i.e. tele-monitoring, for example at a home health provider (PSAD). - the transmission of weight measurements is carried out via means of communication or remote transmission, i.e. remote communication or transmission via a telecommunications protocol allowing data exchange, for example via Bluetooth® or via a mobile telephone network, such as the GSM network (eg 4G / 5G). - the means of remote transmission include and / or implement a multifunction telephone, a digital tablet, a computer or other. - one or more additional measurements of one or more other vital signals of the patient are also carried out using suitable devices, such as blood pressure using a tensiometer, blood oxygen saturation using an oximeter or other. - additional measurements of one or more other vital signals of the patient are also transmitted remotely and can be used in the mathematical model to increase its accuracy.

[0014] The invention further relates to a system, such as an installation, for early detection of the risk of cardiac decompensation in a patient suffering from heart failure comprising:

[0015] - means for measuring the patient's weight comprising a weight measuring scale, - transmission means for transmitting weight measurements from the weight measuring means, - computer processing means with a processor configured to receive the weight measurements transmitted by the transmission means and to process said weight measurements by means of a predefined mathematical model implemented by the processor to deduce therefrom a risk score of cardiac decompensation of said patient, - alert means to trigger a cardiac decompensation alert when the processor determines that the risk score exceeds an alert threshold

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024] given prefixed, and - display means for displaying at least said risk score. Depending on the embodiment considered, the system of the invention may comprise one or more of the following characteristics: - the patient's weight measurement means include a connected scale. - the computer processing means include at least one computer server. - the display means comprise at least one display screen, such as a computer screen. - it further includes data storage means, for example a hard disk, an SSD disk, S3 storage or the like. - transmission means are or include means of communication or remote transmission configured to operate remote communication or transmission via a telecommunications protocol allowing data exchange, for example via Bluetooth® or via a mobile telephone network, such as the GSM network (eg 4G / 5G). The early detection system for the risk of cardiac decompensation according to the invention can be implemented on site, i.e. at the patient's home, or remotely, for example at a home health provider (PSAD) or other. The invention will now be better understood thanks to the following detailed description, given for illustrative but non-limiting purposes, with reference to the appended figures among which: [Fig.l] shows a diagram of a first embodiment of the invention. [Fig.2] shows a diagram of a second embodiment of the invention. [Fig.l] shows a diagram of a first embodiment of the invention for detecting a risk of cardiac decompensation in a patient suffering from heart failure. First, the patient's weight measurements are taken using a scale 1 equipped with a weight sensor 1a, a memory 1b and a current source 1c, for example a connected scale. These weight measurements can be recorded in the memory 1b of the scale 1. It also includes a display 1d and possibly one or more other elements, such as a loudspeaker for example. In this embodiment, most of the components are integrated into the scale 1, which allows weight measurements, retrieval of optional information or data, calculations, triggering of alerts in the event of detection of a risk of cardiac decompensation in a patient, etc. More precisely, the weight measurements are carried out by the weight sensor on

[0026]

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033] several days, namely at least 3 days, typically at least 7 days, in order to have enough measurements to obtain a representative result, as explained below. The computer processing means 6 integrated in the scale 1 are configured to process the weight measurements within their (micro) processor(s) 7 using (at least) a predefined mathematical model, i.e. recorded, to deduce therefrom a cardiac decompensation risk score for the patient in question, i.e. the patient for whom the weight measurements have been transmitted, and to trigger a cardiac decompensation alert 8 when the risk score exceeds a given pre-set alert threshold. More precisely, the weight measurements received by the computer processing means 6 can undergo, within the processor 7, pre-processing (in 7.1), such as filtering: - to eliminate outliers by anomaly detection techniques, for example by detecting weights outside acceptable value ranges, and / or - to fill in missing values ​​using moving average replacement techniques, value interpolation or value propagation. Then, the mathematical model includes sub-models each obtained from historical weight data and decompensation events from several patient samples from the same patient group. These historical weight data and the associated decompensation events are information stored in a pre-registered database that has been compiled from data / information collected in a total patient group comprising several hundred patients, for example at least 400 patients. All of this information, i.e. weight history and decompensation events, constitutes a reference learning database. Of course, all inadequate, unrepresentative or erroneous information has been removed from this database or corrected, including: - Data from patients who had less than 15 days of follow-up between the first day and the last day of valid measurements. - Patients with fewer than 7 valid measurements over 15 days of follow-up. - Missing weight measurements for one day are replaced by the previous measurements. - Missing measurements of symptoms or other possible vital signs are replaced by values ​​from linear interpolations. In order to constitute the sub-models, the processor 7 uses different samples of several dozen patients taken from the total group of patients.

[0035] More precisely, each sub-model is configured to determine a risk score (between 0 and 1) by:

[0036] - extraction and prioritization of several statistical characteristics obtained from historical weight data and decompensation events, and their frequency of occurrence, - determination of at least one alert threshold level for each of the statistical characteristics, and - definition of a risk score based on the frequencies of occurrence of decompensation events for the statistical characteristics considered and the alert threshold levels having been determined.

[0037] The statistical characteristics obtained from historical weight data and decompensation events are typically: deviation (in %), standard deviation, variance, trend, difference between minimum and maximum weight value over several days (2 to 5 days), absolute mean deviation, kurtosis coefficient, symmetry, activity, mobility and complexity.

[0038] More precisely, the statistical characteristics are such that:

[0039] - the deviation (in %) in weight corresponds to:

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] Where: P is the deviation (in %) in weight, x(t) is the weight measured on day t, j is the number of days following day t for the calculation of the deviation. - the standard deviation is calculated as follows: ~ - ^2 1 V t Where: Sigma is the standard deviation, x(t) is the weight measured on day t, mu is the average of the weights x(t) over N days. - the variance corresponds to: \W(£) — cr(t}~ Where: Var is the variance, sigma is the standard deviation. - the trend corresponds to: to, b — argmin — (at + b))2 t Where: â is the estimated trend, bA is the estimated origin, x(t) is the weight measured on day t. - the difference between the minimum and maximum weight values ​​over several days (2 to 5 days) corresponds to:

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073] Af(Z) — max(æ(£)) — min(æ(t)) Where: M is the difference between the minimum and maximum value, x(t) is the weight measured on day t. - the absolute mean deviation corresponds to: - z4 t Where: A is the absolute mean deviation, N is the number of days, x(t) is the weight measured on day t, mu is the average of the weights x(t) over N days. - the flattening coefficient (kurtosis) corresponds to: Where: G is the flattening coefficient, mu4 is the 4th order statistical moment, mu2 is the 2nd order statistical moment. - symmetry corresponds to: b ~ — has' J Where: S is the asymmetry, mu3 is the statistical moment of order 3, sigma is the standard deviation. - the activity corresponds to: Act — var(;r(t)) Where: Act is the activity, var is the variance, x(t) is the weight measured on day t. - mobility corresponds to: Mob - a Where: Mob is the mobility, var is the variance, x(t) is the weight measured on day t, dx(t) is the derivative of weight x on day t. - complexity corresponds to: AT ï / S t I Gom = , r , z ? , MobtÆ(G) Where: Com is the complexity, Mob is the mobility, var is the variance, x(t) is the weight measured on day t, dx(t) is the derivative of weight x on day t. Once the different sub-models have been obtained, the weight measurements are processed (in 7.2) within each sub-model to determine a risk score per sub-model, which score is between 0 and 1, then to deduce an average risk score (in 7.3) equal to the average of the risk scores per sub-model. Processor 7 triggers a cardiac decompensation alert (at 8) when the average risk score exceeds a given alert threshold, typically stored.

[0074] The triggering of the alert (at 8) notably comprises the emission 2 of a cardiac decompensation alert and a display of this alert on display means 3, such as a computer screen or the like, including a display of the cardiac decompensation risk score determined for the patient in question and possibly other useful information.

[0075] Optionally, an external database 4 is also provided which can contain additional data such as age, sex, place of residence, medical history or other information related to the patient which can also be transmitted 2 and displayed on the display means 3.

[0076] The display of the alert and possibly other information related to the patient is done on a display 3 which can be located at the patient's home or, remotely, in a remote monitoring center, i.e. tele-monitoring, for example at a home health provider (PSAD) for example, via the internet, GSM or other network.

[0077] The display 3 can also be the same as the display Id of the scale or another display, such as that of a multifunction telephone (i.e. smartphone), a digital tablet, a portable or fixed computer, or any other device with a display.

[0078] Thus, a PSDA operator carrying out remote monitoring of the patient can immediately and simply detect, by consulting a display screen, a risk of cardiac decompensation in the patient considered to be suffering from heart failure and therefore take appropriate measures, such as sending a nurse to the site, etc., to intervene without delay in order to avoid hospitalization of the patient.

[0079] Optionally, one or more additional measurements of other vital signals of the patient can also be carried out by means of suitable device(s) 10, such as blood pressure by means of a tensiometer, blood oxygen saturation by means of an oximeter or other. These additional measurements of one or more other vital signals of the patient can also be transmitted remotely (11) and used in the mathematical model implemented by the processor 7 or used differently to refine or improve the detection of cardiac decompensation.

[0080] [Fig.2] schematizes a second embodiment of the invention for detecting a risk of cardiac decompensation in a patient suffering from heart failure, which is similar and operates in the same way as the first embodiment of [Fig.l], except that most of the components used for recovering measurements made by the scale 1 and possibly other information or optional data, the calculations, the triggering of an alert in the event of detection of a risk of cardiac decompensation in a patient, etc. are no longer integrated into the scale 1 but located on a remote server.

[0081] In this case, the weight measurements coming from the scale 1, and possibly from other devices, such as a tablet 12 or the like used to answer health questionnaires or other measuring devices 10, are first transmitted 2 via remote transmission means 3 implementing a telecommunications protocol allowing an exchange of data, for example in Bluetooth® or via a mobile telephone network, such as the GSM network (eg 4G / 5G).

[0082] The weight measurements can also be (remotely)transmitted 2 to an external database 4 to be stored there, i.e. memorized, and to a remote computer server 9 comprising computer processing means with (micro)processor 7 to process the data and measurements, as previously, for example a 'cloud' server, or other.

[0083] Here again, the weight measurements are carried out over several days, namely at least 3 days, typically at least 7 days, in order to have sufficient measurements to obtain a representative result, as explained below.

[0084] The computer processing means with microprocessor 7 are, here too, configured to process the weight measurements using the predefined mathematical model, i.e. recorded, as explained above to deduce therefrom a cardiac decompensation risk score for the patient in question, i.e. the patient for whom the weight measurements have been transmitted and to trigger a cardiac decompensation alert 8 when the risk score exceeds a given pre-set alert threshold.

[0085] The display of the alert is done, here again, on a display 3 located remotely, typically in a remote monitoring center, i.e. tele-monitoring, at a home health provider (PSAD) for example, via the communication network 3.1, such as the internet, GSM or other network.

Claims

Claims

1. System for early detection of risk of cardiac decompensation in a patient suffering from heart failure comprising: - measuring means (1) of the patient's weight comprising a weight measuring scale, - transmission means for transmitting the weight measurements from the weight measuring means (1), - computer processing means with a processor (7) configured to receive the weight measurements transmitted by the transmission means and process said weight measurements by means of a predefined mathematical model implemented by the processor (7) to deduce therefrom a risk score of cardiac decompensation of said patient, - alert means for triggering a cardiac decompensation alert when the processor (7) determines that the risk score exceeds a given pre-set alert threshold, and - display means (3) for displaying at least said risk score,and wherein the processor-based computer processing means (7) are configured to: a) implement a mathematical model comprising sub-models each obtained from historical weight data and decompensation events from several samples of patients from the same group of patients, and b) process the weight measurements within each sub-model to determine a risk score per sub-model between 0 and 1 from the historical weight data and decompensation events from a sample of patients, and deduce therefrom an average risk score equal to the average of the risk scores per sub-model.,

2. System according to claim 1, characterized in that the weight measuring means (1) are configured to carry out several weight measurements of the patient over several days, preferably at least 3 days.

3. System according to claim 1, characterized in that it comprises storage means for storing the patient's weight measurements, the given alert threshold and the mathematical model.

4. System according to claim 1, characterized in that the microprocessor processing means comprise a computer server.

5. System according to claim 1, characterized in that the patient's weight measurements are carried out using a connected scale.

6. System according to claim 1, characterized in that the decompensation events include unplanned medical consultations and hospitalizations for heart failure of the patients.

7. System according to claim 1, characterized in that the transmission means comprise communication or remote transmission means configured to operate a remote communication or transmission via a telecommunications protocol allowing an exchange of data.

8. System according to claim 1, characterized in that the display means (3) comprise at least one display screen.

Images