System for detecting and alerting of a risk of non-adherence of a patient treated with NIV

Abstract

Title of the invention System for detecting and alerting of a risk of non-adherence of a patient treated by NIV The invention relates to a system (1) for detecting and alerting of a risk of non-adherence of a patient (P) to treatment by non-invasive ventilation (NIV) comprising a non-invasive ventilation device (2) comprising measuring means (3) for determining values ​​of several respiratory parameters (p1, p2…, pi) of the patient in question; computer data processing means (4); and a display device for displaying a predictive score of risk of non-adherence for the patient in question determined by the computer means from the respiratory parameters. Abstract figure: Fig. 1

Description

Title of the invention: System for detecting and alerting of a risk of non-adherence of a patient treated by NIV

[0001] The invention relates to a system for detecting and alerting of a risk of non-adherence of a patient suffering from a pathology or respiratory insufficiency, who is treated by non-invasive ventilation (NIV).

[0002] Non-invasive ventilation (NIV) is assisted ventilation of a patient using a ventilatory assistance device, commonly called a medical ventilator or respirator, supplying a breathing mask or the like with pressurized air, which air is administered to the patient in need in order to help them breathe.

[0003] For example, NIV can be used to help patients suffering from chronic obstructive pulmonary disease (COPD) to re-oxygenate their blood which is too loaded with CO2, which makes it possible to significantly reduce the re-hospitalization of these patients and their mortality rate in the 12 months following the initiation of NIV treatment.

[0004] In practice, it is observed that some patients do not adhere to their NIV treatment and give up their treatment, which increases their probability of re-hospitalization and increases their mortality rate.

[0005] However, detecting such non-adherence is not easy because these patients are generally treated at home and are therefore not constantly monitored by healthcare staff.

[0006] The problem is therefore to be able to effectively detect and immediately alert a patient on NIV, particularly one treated / treated at home, to a risk of non-adherence to treatment.

[0007] The solution of the invention then concerns a system for detecting and alerting of a risk of non-adherence of a patient to a treatment, i.e. to his treatment, by non-invasive ventilation (NIV) comprising:

[0008] - a non-invasive ventilation (NIV) device comprising measuring means to determine values ​​of several respiratory parameters (pb p2..., pû of the patient considered,

[0009] - computer means for processing data configured to:

[0010] . process the values ​​of said respiratory parameters (pb p2...) and deduce characteristics representative characteristics (cbc2.. .,c;) of a risk of non-adherence of the patient to his NIV treatment, and

[0011] . compare the representative characteristics (cb c2.. .,c;) to threshold values ​​(sb s2 ,...) memorized to deduce a predictive score of risk of non-adherence for the patient considered, and

[0012] - a display device for displaying the predictive score of non-adherence risk for the patient in question.

[0013] Depending on the embodiment considered, the system of the invention may comprise one or more of the following characteristics: - the NIV device is a medical ventilator. - the non-invasive ventilation device is configured to deliver pressurized air. - the NIV device includes a turbine supplying pressurized air. - the NIV device is fluidically connected to a respiratory mask, in especially a nasal or face mask, to which it supplies air under pressure. - it further comprises alert means which cooperate with the display device to display an alert when the predictive score is higher than a given risk value. - the alert means are configured to trigger a visual and / or audible alert. - the measuring means are configured to determine values ​​of several respiratory parameters (pb p2..., pû of the patient considered including measurements of flow rate and / or pressure of a respiratory gas supplied to the patient. - the computer data processing means are further configured to process the flow rate and / or pressure measurements in order to determine or estimate a pressure, a treatment duration, an apnea / hypopnea index (AHI), a leak rate, a spontaneous breathing / assisted breathing ratio, an inspiration time / expiration time ratio, a respiratory rate, a duration of spontaneous ventilation of the patient and / or a tidal volume of air exchanged. - it further comprises data transmission means configured to remotely transmit the values ​​of respiratory parameters (pb p2..., p;) to a computer server communicating remotely with the measuring device. - the computer server includes means of storing threshold values ​​(sb s2, etc.). - the predictive score of non-adherence risk is between 0 and 100%. - the computer data processing means (x) are configured to process the values ​​of the respiratory parameters (pb p2...) and deduce the representative characteristics (ci,c2...,C;) of a risk of non-adherence based on a given treatment duration of less than or equal to 90 days from the treatment start date. - the comparison of the representative characteristics (cb c2.. .,c;) with the stored threshold values ​​(sb s2,...) is carried out according to the given processing duration. - the alert means are configured to trigger a visual alert displayed on the display device, for example a color code (green, orange, red) which is a function of the patient's risk level, a percentage (%) of risk, a graph, such as a progress bar or other... - the alert means cooperate with loudspeaker means to trigger an audible, i.e. sound, alert. - the alert means are configured to trigger an alert at a healthcare provider operating remote patient monitoring (PSAD) and possibly at the patient himself. - the computer data processing means are further configured to determine a risk profile (prb pr2, pr3...) of non-adherence for the patient considered from the predictive score of risk of non-adherence and cooperate with the display device to display the risk profile (prb pr2, pr3...) of non-adherence for the patient considered. - the representative characteristics (cb c2.. ,,Ci) are prioritized according to the risk of non-adherence. - it further includes data transmission means for remotely transmitting the values ​​of respiratory parameters (pb p2..., p;) to a computer server (e.g. computer, cloud, edge...) communicating remotely with the measurement means. - the computer means of data processing include one or more processors, in particular one or more microprocessors or microcontrollers. - the computer means of data processing include a computer server. - the data transmission means are configured to transmit data or other information, via a communication protocol, such as wifi, GSM (3G / 4G / 5G), Bluetooth®, the internet, an intranet.... - the computer server includes means of storing threshold values ​​(sb s2, etc.).

[0014] 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:

[0015] [Fig-1] schematizes an embodiment of the system of the invention.

[0016] [Fig.2] represents compliance over a period of 120 days.

[0017] [Fig.l] represents an embodiment of a system 1 according to the invention for treating a patient P by NIV. The patient P is supplied with pressurized respiratory gas, such as air, provided by a gas supply apparatus 2, i.e. a medical ventilator, as part of NIV treatment. The pressurized gas is conveyed by a flexible conduit from the medical ventilator 2 to a respiratory mask delivering the gas to the airways of the patient P, such as a nasal or facial mask.

[0018] The system 1 of the invention for detecting and alerting of a risk of non-adherence of the patient P to his treatment by NIV comprises measuring means 3 arranged in the medical ventilator 2, typically measuring means configured to determine the pressure and / or the flow rate of the air supplied to the patient P, which measurements then make it possible to determine or estimate values ​​of several respiratory parameters (pi, p2..., Pi) relating to the patient P, in particular a pressure, a treatment duration, an apnea / hypopnea index (AHI), a leak rate, a spontaneous breathing / assisted breathing ratio, an inspiration time / expiration time ratio, a respiratory rate, a duration of spontaneous ventilation of the patient and / or a tidal volume of air exchanged. Calculating such respiratory parameters is known per se, in particular from EP-A-2542287.

[0019] Computer data processing means 4 are also provided, such as a processor, making it possible to process the values ​​of the respiratory parameters (pb p2...) measured by the measuring means 3 arranged in the medical ventilator 2 and to deduce therefrom representative characteristics (ci,c2...,c;) of a risk of non-adherence of the patient P, which are then compared to threshold values ​​(sb s2,...) stored in storage means, such as a memory (hard disk, flash memory, cloud storage...) or other, in order to deduce therefrom a predictive score of risk of non-adherence for the patient P between 0 and 100%.

[0020] The data processing computer means 4 are electrically powered by means of supplying electrical current, such as the electrical network (110 / 220V) or one or more storage batteries.

[0021] The data processing computer means 4 first recover and pre-process (in 4.1) the respiratory parameters (pb p2..., p;) relating to the patient P, such as one or more pressures, treatment duration, apnea / hypopnea index (AHI=) or leak rate.

[0022] For example, the preprocessing (4.1) may comprise steps of: - Scaling of respiratory parameter values ​​(pb p2..., pû according to the CPAP machine (i.e. according to the machine manufacturer) used and its intrinsic characteristics.

[0023] Thus, for a respiratory parameter p_i and a given machine m, we have:

[0024] q_i(m) = p_i(m) - average(p_i, m) / std(p_i, m)

[0025] where: . average(p_i, m) is the empirical average of the respiratory parameter pi for the machine m (i.e. according to the manufacturer). . std(p_i, m) is the standard deviation of the respiratory parameter p_i for machine m (i.e. according to the manufacturer). . q_i is the scaled respiratory parameter value. - Replacement of missing values ​​in respiratory parameters (pi, p 2..., Pi). For a respiratory parameter p_i whose value is missing, several methods can be applied, in particular:

[0026] . Fixed value imputation: p_i(t) = 0

[0027] . Imputation by the mean: p_i(t) = mean(pi)

[0028] . Imputation by the previous value: p_i(t) = p_i(tl) - Encoding of patient-related information (i.e. gender) by grouping them by category and calculating an average. For gender information, encoding is done by the average of the respiratory parameter p_i for each modality of the category (i.e. female or male): - Encoding(gender=female) = mean(p_i, gender=female) - Encoding(gender=male) = mean(p_i, gender=male)

[0029] The identification of the period (4.2) makes it possible to select the operating mode of the algorithm. If the processing duration relative to the installation date is less than 90 days, mode 1 is activated. When the processing duration relative to the installation date is greater than 90 days, mode 2 is activated.

[0030] However, in both modes, the respiratory parameters (pb p2..., p;) are processed and the representative characteristics (cb c2.. .,c;) are deduced from them, which are finally compared (in 4.4) and (4.7) to the threshold values ​​(sb s2,...) stored to determine (in 4.5) and (4.8) the predictive score of risk of non-adherence.

[0031] In mode 1, the processing (4.3) comprises steps of calculating statistical indicators on the respiratory parameters (AHI, compliance, leaks, tidal volume, Tidal volume, Spontaneous breathing / assisted breathing ratio, Inspiration time / expiration time ratio, Respiratory frequency, Daily duration of spontaneous ventilation), for example determination of average, trend..., as illustrated in [Tab. 1] below.

[0032] [Tab. 1] Indicator Description Gender Male, Female Age Patient's age Duration Duration of treatment since installation Leaks Mean Variance Trend Eccentricity (asymmetry) Compliance Mean Variance Trend Eccentricity (asymmetry) AHI Mean Variance Trend Eccentricity (asymmetry) Tidal volume Mean Variance Trend Eccentricity (asymmetry) Spontaneous breathing / assisted breathing ratio Mean Variance Trend Eccentricity (asymmetry) Inspiration time / expiration time ratio Mean Variance Trend Eccentricity (asymmetry) Respiratory rate Mean Variance Trend Eccentricity (asymmetry) Daily duration of spontaneous ventilation Mean Variance Trend Eccentricity (asymmetry)

[0033] These statistical indicators are calculated for example via the following formulas where N is the number of days and Xt the respiratory parameter recorded on day t.

[0034] The average is calculated as follows: p = (1 / N).

[0035] The standard deviation is calculated as follows: o = (1 / N). 2

[0036] The variance corresponds to: V= o2

[0037] The trend corresponds to: â, B = argmin + ^))2

[0038] The eccentricity (asymmetry) corresponds to: S =p3 / o3

[0039] The predictive score obtained is used to trigger a non-adherence alert (in 4.4) and furthermore to calculate a risk profile (in 4.5). The non-adherence alert (4.4) and the risk profile (4.5) can be transmitted to a display device 6, such as a computer screen or a multifunction telephone (smartphone), which allows it to be displayed. The transmission of the score can be done via a protocol and / or communication network 8, such as web, GSM or other. The triggering of a non-adherence alert (in 4.4) is preferably done via a Machine Learning process, for example

[0040] The risk score of non-adherence of a patient P is carried out over a period of N days. The risk score is based on a logistic regression model which makes it possible to determine the weighting coefficients via statistical learning on historical data of the patients' respiratory parameters.

[0041] The calculation of a risk score by the weighted sum of the different statistical indicators from the respiratory parameters according to the following function:

[0042] where: - w is the vector of weighting coefficients. - y_i is the non-adherence risk score of a patient i. - X_i is the set of statistical indicators (i.e. mean, variance, asymmetry ...) resulting from respiratory parameters.

[0043] This is an adjustment constant.

[0044] Furthermore, in mode 2, the processing (4.6) includes steps of calculating statistical indicators on the respiratory parameters (compliance, AHI, etc.). for example, the determination of percentiles, as given in [Tab 2] below.

[0045] [Tab 2] Indicator Description Compliance Baseline compliance [t-120, t-30] Current compliance [t-30, t] AHI Baseline AHI [t-120, t-30] Current AHI [t-30, t] Leaks Baseline leaks [t-120, t-30] Current leaks [t-30, t] Tidal volume Baseline tidal volume [t-120, t-30] Current tidal volume [t-30, t] Spontaneous / assisted breathing ratio Baseline spontaneous / assisted breathing ratio [t-120, t-30] Current spontaneous / assisted breathing ratio [t-30, t] Inspiration time / expiration time ratio Baseline compliance [t-120, t-30] Current compliance [t-30, t] Respiratory rate Baseline respiratory rate [t-120, t-30] Current respiratory rate [t-30, t] Time daily spontaneous ventilation Reference daily spontaneous ventilation duration [t-120, t-30] Current daily spontaneous ventilation duration [t-30, t]

[0046] These statistical indicators are calculated for example via a comparison between the current compliance at t, t-30 and the reference compliance between t-120, t-30. The following formulas where N is the number of days and Xt the respiratory parameter recorded on day t.

[0047] Thus, [Fig.2] represents compliance over a period of 120 days.

[0048] It includes a so-called “reference” period (PR) ranging from 0 to 90 days and a so-called “current” period (PA) ranging from 90 to 120 days. As can be seen, when the 25th percentile (P25) of the reference period [0, 90] is greater than the median (M) of the current period [90, 120], the alert is triggered.

[0049] We have: - Reference compliance [0, 90] = P25(Compliance[0, 90]) ) - Current compliance [90, 120] = M(Compliance[90, 120]) )

[0050] Indeed, the fact that the 25th percentile (P25) of the reference period [0, 90] is higher than the median (M) of the current period (i.e. P25 >M) means that the patient is no longer observing or is incorrectly observing his treatment, i.e. he is no longer adhering.

[0051] An alert must therefore be triggered to warn the healthcare staff of this non-adherence of the patient to his treatment. The non-adherence alert may be visual and / or audible. To this end, the system of the invention may comprise visual display means and / or audible means, such as a computer display, digital tablet or other device making it possible to display the alert and thus warn the healthcare staff of this non-adherence of the patient.

[0052] More generally, the system 1 of the invention makes it possible to detect and alert of a risk of non-adherence to a treatment of a patient P under NIV with administration of pressurized air, delivered by a respiratory assistance device, i.e. a medical ventilator.

Claims

Claims

1. System (1) for detecting and alerting of a risk of non-adherence of a patient (P) to a treatment by non-invasive ventilation (NIV) comprising: - a non-invasive ventilation device (2) comprising measuring means (3) for determining values of several respiratory parameters (pb p2..., Pi) of the patient in question, - computer data processing means (4) configured to: . process the values of said respiratory parameters (pb p2...) and deduce therefrom representative characteristics (ci,c2...,c;) of a risk of non-adherence of the patient to his treatment by NIV, and . comparing the representative characteristics (cb c2.. .,c;) with stored threshold values (si, s2,...) to deduce a predictive score of risk of non-adherence for the patient considered, and - a display device (5) for displaying the predictive score of risk of non-adherence for the patient considered.

2. System according to claim 1, characterized in that it further comprises alert means which cooperate with the display device (5) to display an alert when the predictive score is greater than a given risk value.

3. System according to claim 1, characterized in that the measuring means (3) are configured to determine values of several respiratory parameters (pb p2..., pû of the patient considered including measurements of flow rate and / or pressure of a respiratory gas supplied to the patient.

4. System according to claim 3, characterized in that the data processing computer means (4) are further configured to process the flow rate and / or pressure measurements in order to determine or estimate a pressure, a treatment duration, an apnea / hypopnea index (AHI), a leak rate, a spontaneous breathing / assisted breathing ratio, an inspiration time / expiration time ratio, a respiratory rate, a duration of assisted ventilation of the patient and / or a tidal volume of air exchanged.

5. System according to claim 1, characterized in that it further comprises data transmission means configured to remotely transmit the values of respiratory parameters (pb p2..., p;

6.

7.

8.

9. ) to a computer server communicating remotely with the measuring device. System according to claim 7, characterized in that the computer server comprises means for storing the threshold values (si, s2,...). System according to claim 1, characterized in that the predictive score of risk of non-adherence is between 0 and 100%. System according to claim 1, characterized in that the computer data processing means (4) are configured to process the values of the respiratory parameters (pb p2...) and deduce therefrom the representative characteristics (ci,c2...,Cj) of a risk of non-adherence as a function of a given treatment duration less than or equal to 90 days from the treatment start date. System according to claims 1 and 8, characterized in that the comparison of the representative characteristics (cb c2...,C;) with the stored threshold values (si, s2,...) is carried out as a function of the given processing duration.

Images