Remote monitoring system for diabetic patients

Abstract

Title of the invention Remote monitoring system for diabetic patients The invention relates to a monitoring system (1), preferably remotely, for a diabetic patient (P) comprising a continuous blood glucose measurement device (2) or CGM configured to carry out blood glucose measurements reflecting the blood glucose of the patient (P) over a long period of time (T) of several days, and microprocessor data processing means (3) for processing the blood glucose measurements carried out and then controlling display means (4) for carrying out a display of the variations in the blood glucose of the patient and of one or more zones (Z1, Z2,… Zi) preceding phases of hypoglycemia or hyperglycemia identified and corresponding to a cause of hypoglycemia or hyperglycemia. Abstract figure: Fig. 3

Description

Title of the invention: Remote monitoring system for diabetic patients

[0001] The invention relates to a system for (remote)monitoring a diabetic patient, preferably for remote monitoring, comprising display means for displaying not only the variations in the patient's blood sugar level over time but also the periods of time corresponding to phases of hypoglycemia and / or hyperglycemia, as well as the causes of these hypo- or hyperglycemias, and preferably a recommendation for action to be taken for better management of the patient by the healthcare staff, i.e. doctor or the like.

[0002] Diabetic patients are commonly treated by injecting insulin, which is a drug used to promote the entry of glucose into cells and therefore its metabolism by the body.

[0003] Insulin therefore plays a role in regulating blood sugar, particularly during and after a meal by the diabetic patient. However, it is also essential that the dose of insulin administered corresponds to the patient's needs, i.e. neither too high nor too low so that it does not induce either hypo- or hyperglycemia.

[0004] However, it is important for the practitioner to be able to determine or estimate the causes leading to hypo- or hyperglycemia, knowing that blood sugar peaks can also be caused by other causes, in particular significant stress.

[0005] The problem is therefore to propose a monitoring system, preferably remotely, for diabetic patient(s) allowing a practitioner to know when a patient has been in hypo- and / or hyperglycemia during a given period of time, typically several days, and to be able to determine the cause of this or these hypo- and / or hyperglycemias in order in particular to be able to check whether the treatment applied to this patient is suitable or needs to be readjusted and / or whether their way of eating is correct / compatible or not with their diabetes.

[0006] To respond to this, the invention proposes a monitoring system making it possible to automatically detect the phases of hypo- and hyperglycemia in a patient but also the probable causes of these phases of hypo- and hyperglycemia from the data of this patient's blood sugar and to provide this information to a practitioner, such as a doctor or other, so that the latter can become aware of it and in particular check whether the treatment applied to the latter is suitable or needs to be readjusted.

[0007] More specifically, the solution of the invention relates to a system for monitoring a diabetic patient, preferably for remote monitoring or telemonitoring, comprising: • a continuous blood glucose measurement device or CGM configured to carry out blood glucose measurements reflecting the patient's blood glucose over a given long period of time (T), • storage means to store a history of blood sugar measurements, meals consumed and insulin doses obtained from several patients, • means of display and • microprocessor data processing means configured to:

[0008] a) process the blood glucose measurements to determine short time periods (dt) corresponding to phases of hypoglycemia or hyperglycemia during the long time period (T),

[0009] b) using the blood glucose measurements and stored data histories to determine the causes of hypoglycemia or hyperglycemia via a mathematical model derived from machine learning, and

[0010] c) controlling the display means (4) to operate a display of a graphic representation (RG) comprising: - a curve (Cg) illustrating the variations in the patient's blood sugar level over the long period of time (T), - short time periods (dt) identified within said curve (Cg), corresponding to one or more phases of hypoglycemia and / or hyperglycemia, and - one or more zones (Zl, Z2,... Zi) identified on said curve (Cg), each zone (Zi) preceding an identified phase of hypoglycemia or hyperglycemia and corresponding to a cause of hypoglycemia or hyperglycemia chosen from insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake

[0011] In the context of the invention:

[0012] - by “diabetes” we mean type 1 or type 2 diabetes,

[0013] - by "insulin overcompensation" is meant a quantity of insulin administered too high in relation to the quantity of glucose to be assimilated.

[0014] - by "insulin undercompensation" is meant an amount of insulin ad insufficiently administered in relation to the quantity of glucose to be assimilated.

[0015] - by "insufficient meal intake" we mean an ingestion of food in quantity caloric intake lower than a person's physiological needs.

[0016] - by "excessive meal intake" is meant an ingestion of food in quantity caloric intake greater than a person's physiological needs.

[0017] Depending on the embodiment considered, the system of the invention may comprise one or more of the following characteristics: - the data processing means are further configured to control the display means to operate a display of at least one recommendation relating to the patient's meal consumption or insulin intake. The recommendation includes a change in insulin dosage and / or calorie intake during meals. The causes of hypoglycemia are chosen from insulin overcompensation and insufficient meal intake. The causes of hyperglycemia are chosen from insulin undercompensation and excessive meal intake. the display means are configured to display on the graphic representation (RG), a first threshold (Si) corresponds to a hypoglycemia threshold and a second threshold (S2) corresponds to a hyperglycemia threshold. The zones (Zb Z2, Z3, ... Zi) corresponding to the causes of hypoglycemia or hyperglycemia are represented in different colors, each color corresponding to a given cause. The curves illustrating the variations in blood glucose and the zones of interest can be represented by colors to facilitate reading by medical personnel. For example, the curve (Cg) illustrating the variations in the patient's blood glucose can be represented in blue, the short periods of time (dt) can be represented in black for a hypoglycemia phase or in red for a hyperglycemia phase and the zones (Zl, Z2, ... Zi) can be represented in orange. Of course, other colors can be used. the recommendation is represented by a text message corresponding to the cause considered, for example a text message aimed at encouraging the patient to consume less or more food (meals) or to inject less or more insulin. the first threshold (Si) corresponds to a hypoglycemia threshold between 65 mg / dL and 75 mg / dL, preferably equal to approximately 70 mg / dL (i.e. 0.70 g / L) or approximately 3.9 mmol / dL. the second threshold (S2) corresponds to a hyperglycemia threshold between 170 mg / dL and 200 mg / dL, preferably equal to approximately 180 mg / dL (i.e. 1.80 g / L) or approximately 10 mmol / dL. the display means include a fixed or portable computer screen, a digital tablet or a multifunction telephone. the hypoglycemia threshold is displayed as a first horizontal line. the hyperglycemia threshold is displayed as a second straight line ho- rizontal. - the first line and the second line are parallel. - the first right and the second right are displayed on the means display. - the cause of hyper / hypoglycemia is displayed by a yellow zone.

[0018] Depending on the embodiment considered, the system of the invention may further comprise one or more of the following characteristics: - the continuous glucose monitoring or CGM device is configured to take blood glucose measurements every 2 to 10 minutes, for example every 5 minutes. - the continuous glucose monitoring (CGM) device is configured to perform measurements of the rate (i.e. concentration or level) of interstitial glucose. - the continuous glucose monitoring (CGM) device is configured to be implanted on the patient's body, for example stuck to their skin. - the long time period (T) is at least 5 days, preferably at least 7 days. The longer the duration T, the more the measurements taken reflect the patient's eating habits and physiology, i.e. the postprandial response (i.e. glycemic diffusion). - the data processing means comprise one or more microprocessors. - the data processing means include a computer server, in particular a remote server, i.e. one located several hundred meters or several km from the place where the measurements are taken (e.g. at home, in hospital, in an EHPAD or in any other establishment receiving one or more diabetic patients). - the data processing means are configured to carry out identification or detection of hypoglycemia and / or hyperglycemia phases by implementing artificial learning or machine learning, i.e. machine learning in English. - the continuous glucose monitoring (CGM) device is configured to transmit or provide the glucose measurements to the data processing means, preferably to transmit these measurements remotely. - the continuous blood glucose monitoring (CGM) device includes (tele)communication means for (re)transmitting the measurements. - the means of communication are configured to transmit the measurements via GSM (3G / 4G / 5G), wifi, Bluetooth™, internet or other. - the display means are configured to simultaneously display the first line and the second line, the graphic representation showing the variations in blood sugar, the phases of hypoglycemia and / or hyperglycemia and the zones (Zb Z2, Z3, ... Z) corresponding to the causes of hypoglycemia or hyperglycemia. - the display means include a screen, preferably in color. - the screen is a fixed or portable computer screen, a digital tablet, multifunction phone (i.e. smartphone in English) or other. - the server comprises the storage means configured to store the artificial learning model and the first and second thresholds. - the storage means include at least one computer memory. - the means of (tele)communication are configured to (remote)transmit the measurements continuously, periodically or punctually, preferably punctually in one go at the end of the long period of time T considered.

[0019] 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:

[0020] [Fig-1] shows a diagram of an embodiment of a patient remote monitoring system diabetic according to the invention;

[0021] [Fig.2] illustrates an example of the measurement processing chain within the server of [Fig.l]; and

[0022] [Fig.3] illustrates an example of display on the display means of the remote monitoring system of [Fig.l].

[0023] [Fig.l] schematizes an embodiment of a remote monitoring system 1 of a diabetic patient P (type 1 or 2 diabetes) according to the invention comprising a continuous blood glucose measurement device, or CGM device 2 (for Continuous Glucose Monitoring in English), making it possible to carry out blood glucose measurements reflecting the blood glucose level of the patient P, that is to say his interstitial glucose level, over a given long period of time T of several days, typically at least 2 to 5 days, preferably at least 7 days or more.

[0024] In [Fig.3], the abscissa axis gives the time (here in minutes), typically a long period of time T, and the ordinate axis gives the glucose concentration (here in mg / dL).

[0025] In general, a CGM device 2 is implanted on the patient's body, for example stuck to his skin, so as to carry out the measurements almost continuously over the entire long period of time T considered, for example every 2 to 10 minutes, typically every 5 minutes.

[0026] Once the measurements have been made, the CGM device 2 transmits them remotely to data processing means 3 with microprocessor(s), such as a remote server located several hundred meters, generally several kilometers or tens of kilometers, or even at a longer distance, from the place where the measurements are taken, namely at the patient's home or at the hospital, in an EHPAD or in any other establishment receiving one or more diabetic patients.

[0027] To do this, the CGM device 2 is equipped with telecommunication means 2-1 allowing the measurements to be transmitted remotely via a telecommunication protocol of the GSM (3G / 4G / 5G), wifi, Bluetooth™, internet or other type. Such telecommunication means are conventional and are not detailed here; they may for example include a modem, an antenna, a transmitter / receiver module, etc.

[0028] Depending on the embodiment, the remote transmission of measurements can be done: - either continuously, that is to say as soon as a measurement is taken, - either periodically, that is to say repetitively but spaced out over time, for example 1 or 2 times a day, - either punctually, that is to say in a unique manner, for example once at the end of the long period of time T considered, typically after 7 days.

[0029] According to the invention, the data processing means 3 with microprocessor(s), such as a remote server, are configured, for example programmed, to receive the blood glucose measurements, then process them to determine, during the long period of time (T), all the short periods of time which precede periods or phases of hypoglycemia and / or hyperglycemia and which correspond to causes of hypoglycemia or hyperglycemia chosen from insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake.

[0030] These short periods of time which precede periods or phases of hypoglycemia and / or hyperglycemia and which correspond to causes of hypoglycemia or hyperglycemia, can then be displayed on a graphic representation in the form of one or more zones (Zb Z2,... Zj), preceding the phases of hypoglycemia and / or hyperglycemia which are visible on the graphic representation.

[0031] In other words, once collected and processed by a learning model, as explained below in relation to [Fig. 2], the blood glucose measurements are displayed on display means 4, such as a display screen of a fixed or portable computer, digital tablet, multifunction telephone or other, in the form of a blood glucose curve (Cg) over time, on which the zones (Zb Z2,... Zi) representing the causes of hypoglycemia or hyperglycemia are also represented so as to facilitate understanding of the curve by the healthcare personnel.

[0032] More precisely, as illustrated in [Fig.3], the data processing means 3, here a remote computer server, controls the display means 4 to operate a display, preferably in color, on an information display screen used by a healthcare personnel, such as a doctor M or similar, of: - a graphical representation RG, such as a curve, showing the variations in the blood sugar level of patient P, during the long period of time T, - several (i) zones Zb Z2, Z3, ... Z representing causes of hypoglycemia or hyperglycemia chosen from insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake. - a first threshold Si or hypoglycemia threshold, namely here a first threshold Si equal to approximately 70 mg / dL (i.e. 3.9 mmol / dL); and - a second threshold S2 or hyperglycemia threshold, namely here a second threshold S2 equal to approximately 180 mg / dL (i.e. 10 mmol / dL).

[0033] In [Fig.3], the first threshold S1 and the second threshold S2 are displayed by the display means 4 in the form of parallel horizontal lines.

[0034] Furthermore, the server 3 may comprise storage means for storing the artificial learning model used and the first and second thresholds S1, S2 or other information, data or other.

[0035] Preferably, the different zones Zb Z2, Z3, ... Zi representing the causes of hypoglycemia or hyperglycemia are displayed in different colors, depending on the type of cause itself, namely insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake.

[0036] Generally speaking, the system 1 of the invention makes it possible to detect and display the causes of hypoglycemia or hyperglycemia from the blood sugar data of this patient P and then to provide this information to a practitioner, such as a doctor M or other, in the form of a visual representation that is easy to read and interpret so that the latter can take note of it and check whether the treatment applied is suitable or needs to be readjusted.

[0037] In other words, thanks to the system 1 of the invention, the doctor M or similar has the history of the blood sugar level of a patient P over a long period T of several days and an easy and rapid visualization not only of the meals taken and the insulin injections, i.e. their effects, on the blood sugar level, in particular when this results in one or more hypo or hyperglycemias.

[0038] All this information is of major importance because it makes it possible to confirm or adjust the treatment of the diabetic patient P and therefore to achieve better monitoring of the patient by improving the balance between his eating habits and his treatment by insulin injection or other.

[0039] [Fig.2] illustrates an example of the measurement processing chain within the server 3 of [Fig. 1],

[0040] As can be seen, the processing chain for the blood glucose measurements taken in a given patient P firstly comprises, i.e. as input, processing of the time series 20 of blood glucose measurements provided by the CGM device 2, called series G.

[0041] This series G comprises a sequence of indexes (G_l, G_2,..., G_N) of length N corresponding to the N measurements taken by the CGM device 2 during a given period of time T, for example 7 days, each index G_i being associated with a time marker (date and time) and a blood glucose value.

[0042] An example of a G series is given in the following [Table 1].

[0043] [Tables 1] Index G_i Time Marker (Date - Time) Blood Glucose Value (G) G_1 10 / 01 / 2021 19:20 120 G_2 10 / 01 / 2021 19:27 123 G_3 10 / 01 / 2021 19:32 125

[0044] During the first processing step, the server is configured to perform a series of pre-processing operations 21 on the time series G.

[0045] A first pre-processing may consist of removing any aberrant value from the series G. For example, a value G_i is considered aberrant if G_i < 30 or G_i > 400 or if a variation I G_i+1 - G_i I > 50.

[0046] A second preprocessing can consist of resampling the time series G and replacing the missing values. Resampling makes it possible to uniformly obtain measurements spaced several minutes apart, for example 5 minutes. To do this, a linear interpolation is performed between neighboring measurements. For example, the linear interpolation between G_1 (t_l = 19:20) and G_2 (t_2 = 19:27) makes it possible to approximate a value G_2p (t_2p = 19:25) according to the following formula: G_2p = G_1 + (G_2 - G_l) / (t_2 - t_l)

[0047] When there are fewer than 5 consecutive missing values, i.e. a period of less than 30 minutes, the values ​​are also estimated by linear interpolation.

[0048] Then, a second step 22 of the processing chain can comprise a segmentation of the time series preprocessed during the first step, that is to say a preparation of segments.

[0049] Segmentation consists of selecting time periods of several hours data, for example 3-hour periods, within the time series Gp. This step is repeated in 1-hour steps. Stacking these segments produces a two-dimensional matrix X where each row represents a 3-hour segment, as shown in [Table 2].

[0050] [Tables2] Time Marker (3h Segment) Blood Glucose Value 10 / 01 / 2021 19:25 -> 10 / 01 / 2021 22:25 123, 125, ... 10 / 01 / 2021 22:30 -> 11 / 01 / 2021 01:30 102, 98, ...

[0051] The third step, called 'inference' 23, includes a calculation of the probability that a 3-hour segment corresponds to a meal-taking phase.

[0052] For this, an automatic classification model is previously trained of the LSTM or “short and long term memory” type. The LSTM model is described by the following document which can be referred to for more details:

[0053] Long Short Term Memory; S. Hochreiter & J. Schmidhuber; Neural Computation; Volume 9; Issue 8; November 15, 1997; p. 1735-1780

[0054] In summary, the LSTM model includes a set of coefficients W which make it possible to calculate (in 24) a score representative of the probability (between 0 and 1, 1 corresponds to a meal) that a 3-hour segment X_i corresponds to a cause y_i.

[0055] This translates into: LSTM(X_i, W) = y_i

[0056] Beforehand, for the LSTM model to function correctly, a step of estimating the optimal W coefficients is necessary. This step, called model training, is a process capable of extracting correlations in a history of blood glucose measurement data from several patients associated with mealtime events, data previously stored within the server or similar.

[0057] Finally, the last step (in 25) consists of determining whether a segment X_i corresponds to one of the 4 phases, namely: insulin undercompensation, insulin overcompensation, excessive meal intake and insufficient meal intake.

[0058] For this, 4 scores representative of the probability y_i are compared to a threshold sp (for example sp=0.5).

[0059] A cause is identified when y_i >= sp. Alternatively, when y_i < sp, the segment does not constitute a cause.

[0060] The LTSM model is thus capable of determining the zones corresponding to meal intake by generating scores on each segment of the blood glucose curve and by detection of exceeding the sp threshold. The optimal sp threshold is previously set from experiments by finding the best compromise between false detections and non-detections.

[0061] Hypoglycemia and hyperglycemia are then detected each time a meal intake zone Zi exceeds the first or second threshold (SI, S2) respectively.

[0062] Thanks to the display means making it possible to display the graphic representation (Cg curve) of the variations in blood sugar, the zones (Zl, Z2, Z3, ... Zi) preceding a phase of hypoglycemia or hyperglycemia identified and corresponding to a cause of hypoglycemia or hyperglycemia chosen from insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake, and preferably the first threshold (SI) corresponding to a hypoglycemia threshold and the second threshold (S2) corresponding to a hyperglycemia threshold. The health personnel can then make the best decisions and actions for effective treatment and monitoring of the diabetic patient.

Claims

1. Claims System for monitoring (1), preferably remotely, a diabetic patient (P) comprising: • a continuous blood glucose monitoring device (2) configured to perform blood glucose measurements reflecting the patient's blood glucose (P) over a given long period of time (T), • storage means to store a history of blood sugar measurements, meals consumed and insulin doses obtained from several patients. • display means (4) and • microprocessor data processing means (3) configured for: a) processing the blood glucose measurements to determine short time periods (dt) corresponding to phases of hypoglycemia or hyperglycemia during the long time period (T), b) using the blood glucose measurements and the stored data histories to determine the causes of hypoglycemia or hyperglycemia via a mathematical model derived from machine learning, and c) controlling the display means (4) to operate a display of a graphic representation (RG) comprising: - a curve (Cg) illustrating the variations in the patient's blood sugar level over the long period of time (T), - short time periods (dt) identified within said curve (Cg), corresponding to one or more phases of hypoglycemia and / or hyperglycemia, and - one or more zones (Zl, Z2,... Zi) identified on said curve (Cg), each zone (Zi) preceding an identified phase of hypoglycemia or hyperglycemia and corresponding to a cause of hypoglycemia or hyperglycemia chosen from insulin overcompensation, insulin undercompensation, insufficient meal intake and excessive meal intake.

2. Monitoring system according to claim 1, characterized in that the data processing means (3) are further configured to control the display means (4) to operate a display of at least one recommendation relating to the consumption of meals or the taking of insulin by the patient.

3. Monitoring system according to claim 2, characterized in that the recommendation comprises a modification of the insulin dosage and / or the caloric intake during meals.

4. Monitoring system according to claim 1, characterized in that the causes of hypoglycemia are chosen from insulin overcompensation and insufficient meal intake, and / or the causes of hyperglycemia are chosen from insulin undercompensation and excessive meal intake.

5. Monitoring system according to claim 1, characterized in that the display means (4) are configured to display on the graphic representation (RG), a first threshold (Si) corresponds to a hypoglycemia threshold and a second threshold (S2) corresponds to a hyperglycemia threshold.

6. Monitoring system according to claim 1, characterized in that the zones (Zi, Z2, Z3, ... Z) corresponding to the causes of hypoglycemia or hyperglycemia are represented in different colors, each color corresponding to a given cause.

7. Tracking system according to claim 2, characterized in that the recommendation is represented by a text message corresponding to the cause considered.

8. Monitoring system according to claim 5, characterized in that the first threshold (Si) corresponds to a hypoglycemia threshold between 65 mg / dL and 75 mg / dL and the second threshold (S2) corresponds to a hyperglycemia threshold between 170 mg / dL and 200 mg / dL.

9. Tracking system according to claim 1, characterized in that the display means (4) comprise a screen of a fixed or portable computer, digital tablet or multifunction telephone.

10. Tracking system according to claim 1, characterized in that the data processing means comprise a computer server, in particular a remote server.

Images