WO2019229686A1 - Système de détection de dysfonctionnements dans des dispositifs de distribution d'insuline - Google Patents

Système de détection de dysfonctionnements dans des dispositifs de distribution d'insuline Download PDF

Info

Publication number
WO2019229686A1
WO2019229686A1 PCT/IB2019/054470 IB2019054470W WO2019229686A1 WO 2019229686 A1 WO2019229686 A1 WO 2019229686A1 IB 2019054470 W IB2019054470 W IB 2019054470W WO 2019229686 A1 WO2019229686 A1 WO 2019229686A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
processing unit
data processing
insulin
variable
Prior art date
Application number
PCT/IB2019/054470
Other languages
English (en)
Inventor
Simone Del Favero
Lorenzo MENEGHETTI
Matteo TERZI
Gian Antonio SUSTO
Claudio Cobelli
Original Assignee
Universita' Degli Studi Di Padova
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 Universita' Degli Studi Di Padova filed Critical Universita' Degli Studi Di Padova
Publication of WO2019229686A1 publication Critical patent/WO2019229686A1/fr

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
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/16831—Monitoring, detecting, signalling or eliminating infusion flow anomalies
    • 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
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • G16H20/17—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients delivered via infusion or injection
    • 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
    • G16H40/40—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 management of medical equipment or devices, e.g. scheduling maintenance or upgrades
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/142—Pressure infusion, e.g. using pumps
    • A61M2005/14208—Pressure infusion, e.g. using pumps with a programmable infusion control system, characterised by the infusion program
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/172—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body electrical or electronic
    • A61M5/1723—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body electrical or electronic using feedback of body parameters, e.g. blood-sugar, pressure

Definitions

  • the present invention relates to the field of systems for detecting malfunctions in medical devices.
  • the invention relates to a system for detecting malfunctions in devices for administering insulin, such as for example devices comprising an insulin pump and a glucose sensor.
  • the invention also refers to a device of this type incorporating such a detection system.
  • Type I diabetes is a disease that makes a patient no longer able to produce insulin and therefore insulin needs to be artificially administered using a special medical device, such as an insulin pump. Often this insulin pump is integrated with a glucose sensor and in some cases the delivery is automated by a control algorithm (artificial pancreas).
  • a control algorithm artificial pancreas
  • the artificial pancreas is a more advanced system that operates in closed-loop.
  • the artificial pancreas also includes a sensor that detects the patient's glycaemia level.
  • the sensor periodically measures, for example every five minutes, the patient's glycaemia and sends the information to a control unit that controls the pump and decides how much insulin to administer in order to keep the patient's glycaemia levels under control.
  • a problem with this type of medical device is the possible occurrence of failures or malfunctions that are potentially very dangerous for the patient's health. Both the sensor and the pump could break or not work properly, with the risk of administering the wrong amount of insulin. However, even in the absence of pump or sensor malfunctions there may be other problems that affect the correct operation of the medical device, for example the catheter that connects the pump to the needle may come off and consequently insulin not be administered.
  • model-free systems based on the analysis of historical data. Basically, these systems carry out an analysis of the data collected and learn from them what are the "normal" modes of functioning and the "anomalous" behaviors, according to a classification criterion that depends on the method. For example, one of the simplest methods assumes that if a value is very far from the others observed it is probably an anomaly (outlier). These methods can be divided into supervised (supervised learning) and unsupervised (unsupervised learning) depending on whether or not they require a preliminary procedure for labeling historical data such as "failure” / "not failure".
  • An object of the present invention is to solve the problems of medical devices for administering insulin, and in particular of systems for detecting malfunctions in this type of device.
  • a system for detecting malfunctions that uses an unsupervised model-free algorithm.
  • the system includes: a data processing unit capable of receiving measurements of the glucose concentration from the said glucose monitoring sensor and further capable of receiving measurements of the amount of insulin infused by the said infusion pump;
  • a data input unit operatively connected to the data processing unit and capable of receiving estimates from the user of the quantity of carbohydrates consumed
  • a storage unit operatively connected to the data processing unit and capable of storing the glucose concentration measurements, the measurements of the amount of insulin infused and the estimates of the amount of carbohydrates taken.
  • the data processing unit calculates, at a given instant of time, an estimate of the residual insulin present in the blood (IOB), and an estimate of the residual carbohydrates present in the blood (COB).
  • IOB an estimate of the residual insulin present in the blood
  • COB an estimate of the residual carbohydrates present in the blood
  • the data processing unit calculates the value of a variable DCOB(t) which represents the variation of the current glucose weighted by the residual amount of carbohydrates ingested and estimated to still be present in the blood:
  • g(t) is the variable that describes the derivative of glucose concentration measurements over time
  • COB (t) is the variable that describes, over time, the estimate of residual carbohydrates present in the blood
  • a and b are two positive constants.
  • the DCOB(t) variable will be indicated in the following "weighted glucose variation” or “glucose variation weighed on residual carbohydrates”.
  • the data processing unit is further capable of calculating an ICOB variable (t) which represents an estimate of the residual insulin weighed for the residual of ingested and estimated carbohydrates still present in the blood:
  • IOB (t) is the variable that describes, over time, the estimate of residual insulin present in the blood
  • g and d are two positive constants.
  • this variable will be called “weighted residual insulin” or “residual insulin weighed on residual carbohydrates”.
  • the data processing unit is capable of executing at least one anomaly detection algorithm of the unsupervised type configured for assigning an anomaly score to data in a patient data set, based on at least one criterion selected among:
  • each patient data in said set, at a given time comprises at least a triad of values, said triad of values being constituted by a value of the variable DCOB(t) at a given time, a value of the variable ICOB(t) in said given time, and a measurement of glucose concentration by said sensor at said given time
  • Said data processing unit is able to generate an alert signal if said algorithm detects at least one anomaly within the patient data set, said data processing unit detecting at least one anomaly within the patient data set if said anomaly detection algorithm generates an anomaly score out of a predefined range of threshold values.
  • the proposed triple constitutes a particular non-linear transformation of the signals acquired by the data processing unit, and substitutes (or integrates) such signals due to the fact that it considerably increases the possibility of highlighting anomalies.
  • the anomaly detection algorithm is the Local Outlier Factor (LOF), the Connectivity-based Outlier Factor (COF), the k-nearest-neighbors (KNN), the Isolation Forest (iForest), the Histogram-based Outlier Score (HBOS) and/ or the One-Class Support Vector Machine (OCSVM).
  • LEF Local Outlier Factor
  • COF Connectivity-based Outlier Factor
  • KNN k-nearest-neighbors
  • IForest Isolation Forest
  • HBOS Histogram-based Outlier Score
  • OCSVM One-Class Support Vector Machine
  • the data processing unit is capable of using at least one unsupervised anomaly detection algorithm, selected among Local Outlier Factor (LOF), the Connectivity-based Outlier Factor (COF), the k-nearest-neighbors (KNN), the Isolation Forest (iForest), the Histogram-based Outlier Score (HBOS) and the One-Class Support Vector Machine (OCSVM) to the data of the weighted glucose variation (DCOB), to the data of the weighted residual insulin (ICOB) and to the data of the measurements of the concentration of glucose.
  • LEF Local Outlier Factor
  • COF Connectivity-based Outlier Factor
  • KNN k-nearest-neighbors
  • IForest Isolation Forest
  • HBOS Histogram-based Outlier Score
  • OCSVM One-Class Support Vector Machine
  • the system operates using one or more algorithms on a set of three data defined by the weighted glucose variation (DCOB), the weighted residual insulin (ICOB) and the glucose concentration measurement.
  • DCOB weighted glucose variation
  • ICOB weighted residual insulin
  • the data processing unit is adapted to employ at least one unsupervised anomaly detection algorithm selected from Local Outlier Factor (LOF), the Connectivity-based Outlier Factor (COF), the k-nearest- neighbors (KNN), the Isolation Forest (iForest), the Histogram-based Outlier Score (HBOS) and the One-Class Support Vector Machine (OCSVM) to a specific working portion of the weighted glucose variation data (DCOB), the weighted residual insulin (ICOB) and the glucose concentration measurement data.
  • the working portion is defined by excluding a predefined number of data in earlier time instants and a predefined number of data at later time instants with respect to a predetermined time instant.
  • an insulin delivery device which comprises an insulin pump, a glucose sensor and a malfunction detection system incorporating the features of the appended claims, which form an integral part of the present invention.
  • FIG. 1 is a schematic view of an insulin delivery device incorporating the system for detecting malfunctions in insulin delivery devices, in accordance with the present invention.
  • a preferred embodiment will be described, as illustrated in Figure 1, of the system 11 for detecting malfunctions in devices for administering insulin of the type comprising at least one glucose monitoring sensor that measures a person's glycaemia values and an infusion pump of insulin.
  • a preferred embodiment will be described, as illustrated in Figure 1, of an artificial pancreas 1 including a glucose monitoring sensor 21 that monitors the glucose in the blood of a person and an insulin infusion pump 31 and integrating the system 11 to detect malfunctions.
  • the system for detecting malfunctions in devices for administering insulin could be implemented by means of a centralized architecture designed to detect malfunctions in one or more devices for administering insulin connected to the aforementioned centralized architecture.
  • This architecture preferably of the cloud type, could allow the detection of remote malfunctions.
  • the insulin infusion pump 31 is a device that allows continuous infusion 24 hours a day of insulin into the subcutaneous tissue of a patient, favoring the achievement of the best possible glycemic control.
  • the infusion pump 31 is able to administer insulin into the subcutaneous tissue by continuous basal infusion, to maintain normal glycaemia levels during the fasting period, and by infusion of a fast bolus, to regulate glycaemia values in relation to the intake of a meal or too high glycaemia levels.
  • a peculiar characteristic of the infusion pumps is to infuse insulin continuously with the possibility of varying the rate of infusion during the day according to the needs of the individual, thus reproducing in a more precise way the physiological presence of the hormone in the body that varies in the body during the day.
  • continuous infusion means, in the present invention, that the infusion pump 31, or the related control device, is capable of measuring the amount of insulin infused to the patient in predetermined time instants.
  • the continuous glucose monitoring sensor 21 is also inserted into the subcutaneous tissue of the patient to measure the concentration of glucose at predefined time instants, in place of finger pricking measurements.
  • the artificial pancreas 1 can be provided with both the infusion pump 31 and the monitoring sensor 21, or operatively connected to the latter. Furthermore, the artificial pancreas 1 according to the present invention is further provided with the system 11 to detect malfunctions in devices for the administration of insulin according to the present invention or can be operatively coupled to the latter.
  • the system 11 for detecting malfunctions allows, therefore, to detect the malfunctions that can affect the infusion pump 31 regarding mechanical defects, occlusions or detachments of the pump 31 itself from the catheter.
  • the aforementioned defects can in fact lead to a reduced amount of insulin infusion that can lead the subject to hyperglycaemia, or to high glycaemia values.
  • the aforementioned system 11 includes a data processing unit 111, a data input unit 311 and a storage unit 211.
  • the data processing unit 111 includes one or more microprocessors programmed to carry out a plurality of operations as defined in the operation described below.
  • the data processing unit 111 according to the present invention is operatively connected to the aforementioned glucose monitoring sensor 21 and to the aforementioned insulin infusion pump 31.
  • the data processing unit 111 is capable of receiving measurements of a patient's glucose concentration from the continuous glucose monitoring sensor 21, called "g(t) M .
  • the data processing unit 111 is capable of receiving measurements of the amount of insulin infused to the patient by the infusion pump 31, named "i(t) M .
  • This last value i(t) includes both the administration of continuous basal infusion and the administration of fast bolus.
  • the data entry unit 311 is, for example, made up of a data input interface, or user interface, provided directly on the insulin delivery device 1, e.g. an artificial pancreas, or by an external system operatively connected to it.
  • an external system operatively connected to it.
  • such system can comprise a wireless device, such as smartphone or related device.
  • the data input unit 311 is operatively connected to the data processing unit and it is adapted to receive from the user the estimation of the amount of carbohydrates taken in one or more of the instant times named "m(t)", which are known to be associated to the increment of postprandial glucose.
  • the storage unit 211 is preferably of the non-volatile type, for example a storage unit 211 arranged in the insulin delivery device 1 or in an external system operationally similar to what previously described for the data input unit 311.
  • the storage unit 211 is operatively connected to the data processing unit 111 and apt to store the measurement of the glucose concentration g(t), the quantity of insulin infused i(t) and the estimate of the quantity of carbohydrates m(t) taken in the said moments of time.
  • This storage therefore allows the data relating to the trend of the aforementioned signals to be kept for the calculation of the subsequent variables, as described below.
  • the user can be modelled as a dynamic system whose output to be controlled, at each instant of time, is the measurement of the glucose concentration g(t). This output is influenced by the measurement of the amount of infused insulin i(t) and by the estimate of the amount of carbohydrates m(t) taken at the same time instants.
  • a set of potential variables describing the state of the system at each time instant is the following: g(t), g (t), i(t) and m(t), where g(t) is the variable that describes the derivative of glucose concentration measurements in time g(t).
  • the insulin infused at a given time does not have a direct impact on the behavior of the system at the same time instant but on the behavior of the same system in the following 1-8 hours. This is due to the slow absorption dynamics and to the persistence of the past insulin, whose concentration decreases exponentially in a time range equal to 4-8 hours after the insulin infusion.
  • the data processing unit 111 is apt to calculate, at a given instant of time, the estimate of the residual insulin present in the blood (IOB) as a function of the measurement of the amount of insulin taken at each time instant i(t).
  • the IOB value at a given time instant is preferably calculated as a convolution of the measure i(t) with an exponential decay function, as described and incorporated by reference in C. Ellingsen, E. Dassau, H. Zisser, B Grosman, MW Percival, L. JovanovTc, and FJ Doyle III, " Safety constraints in an artificial pancreatic cell: an implementation of model predictive control with insulin on board," Journal of diabetes science and technology, vol. 3, no. 3, pp. 536-544, 2009.
  • the data processing unit 111 is apt to calculate, at a given time instant, the estimate of the residual carbohydrates being in the blood (COB) as a function of the estimate of the amount of carbohydrates taken at each time instant m(t).
  • the COB value at a given time instant is preferably calculated as a convolution of the measure m(t) with an exponential decay function, as described and incorporated by reference in M. Schiavon, C.
  • a set of potential variables that better describes the state of the system at each time instant is the following: g(t), g(t), IOB(t) and COB(t).
  • the system 11 determines possible anomalies on the basis of a set of modified and improved variables defined by a non-linear transformation of the acquired signals (g(t), m(t) and i(t)) and / or processed (g(t), IOB(t), COB(t)) from the data processing unit 111, and stored in the storage unit 211, as described in detail below.
  • the data processing unit 111 is apt to calculate, at a given time instant, a variable called “weighted glucose variation” or “weighted glucose change on residual carbohydrates” (DCOB) as a function of the first derivative of the measurement of the weighted concentration of glucose in the blood for the estimation of the carbohydrates ingested and still present in the COB blood.
  • DCOB variable glucose weighted variation
  • g(t) is the variable that describes the derivative before measuring the concentration of glucose in the blood
  • COB(t ) is the variable that describes, over time, the estimate of residual carbohydrates being in the blood (COB), and a e b are two constants, preferably positive and preferably predefined.
  • COB(t ) in non-anomalous conditions, a rise in glycaemia values is expected in response to a meal. Therefore, high COB values are, even in proper functioning conditions, associated with high glycaemia variations. However, in this condition the DCOB ratio is small.
  • the glycaemia increases persistently even not in response to a meal and therefore also in correspondence with low or zero COB values.
  • COB(t) is small or equal to zero (i.e. distant from meals)
  • DCOB(t) is strongly influenced by the change in glucose g(t).
  • large positive DCOB(t) values occur only when glucose disproportionately increases not at meals and are symptoms of failure of the infusion pump 31.
  • the constants a e b are predefined, preferably positive, constants that allow to increase the impact of one signal on the other.
  • the data processing unit 111 is apt to calculate, at each time instants, an ICOB variable called "weighted residual insulin” or "weighed residual insulin on residual carbohydrates”.
  • an ICOB variable called "weighted residual insulin” or "weighed residual insulin on residual carbohydrates”.
  • the weighted residual insulin variable (ICOB) is described by the following formula
  • IOB(t ) is the variable that describes, over time, the estimate of residual insulin present in the blood (IOB),
  • COB(t ) is the variable that describes, over time, the estimate of residual carbohydrates being in the blood (COB), and g e d are two constants, preferably positive and preferably predefined.
  • the new ingested carbohydrates are compensated by an extra administration of insulin, or fast insulin bolus. So, after a meal, IOB will tend to grow a lot even in case of proper functioning, but in this condition the ICOB value remains small because at the same time COB also grows.
  • the pump 31 attempts to lower the glucose g(t) by increasing the insulin infusion. However, the attempt is unsuccessful, since insulin i(t) is not actually administered due to the fault. In this case the IOB value increases significantly compared to the COB value. For example, if the failure occurs far from the meal, IOB increases while COB remains small and ICOB is very high.
  • the constants g and d similarly to a and b, are predefined, preferably positive, constants that allow increasing the impact of one signal on the other.
  • the data processing unit 111 is apt to perform at least an algorithm for detecting anomalous values, also called anomaly detection algorithm, within a set of patient data, in which each of the patient data comprises at least one set of values, preferably constituted by a value of the weighted glucose variation (DCOB) variable at a given instant of time, a value of the weighted residual insulin variable (ICOB) in said time instant, and a measure of the glucose concentration carried out by the sensor 21 in said time instant.
  • DCOB weighted glucose variation
  • ICOB weighted residual insulin variable
  • the data processing unit 111 is apt to generate an alert signal if the algorithm detects at least one anomalous value within the patient data set.
  • the data processing unit 111 generates an alert signal if the algorithm detects an external result at a predefined threshold value range.
  • the set of preferred characteristics which describe the state of the system at a given time instant is as follows: g(t), ICOB(t) and DCOB(t).
  • the constants a, b, g and d are all equal to 10. Different values can still be used.
  • Other embodiments of the invention can expand the triple by comprising successive derivative of the signals g(t), IOB(t), COB(t), ICOB(t) e DCOB(t) and combinations thereof, possibly non-linear.
  • the anomaly detection algorithm used by the processing unit according to the present invention is preferably an unsupervised and model-free anomaly detection algorithm.
  • the algorithms for the detection of anomalous values based on unsupervised learning do not require training data provided with examples of failures and correct functioning, previously labeled as such (labeled dataset), but are based on the analysis of the set of acquired historical data, i.e. the set of patient data in the problem object of the invention, and are designed to identify and classify as patient anomalies those patient data which appear to possess one or more characteristics that deviate from the remaining data observed in that patient, based on a predetermined criterion.
  • a generic unsupervised anomaly detection algorithm assigns a so-called “anomaly score” to each datum acquired, for example a value that is greater the more the data is considered anomalous.
  • the result generated by the algorithm i.e. the anomaly score, can be used to rank the data based on this score and then distinguish "normal" data from “anomalous” data.
  • the main difference between each unsupervised anomaly detection algorithm is the criterion that is used to examine the data and assign the anomaly scores.
  • LEF Local Outlier Factor
  • COF Connectivity-based Outlier Factor
  • Isolation Forest iF
  • KNN k- nearest-neighbors
  • Isolation Forest iForest
  • HBOS Histogram-based Outlier Score
  • OCSVM One-Class Support Vector Machine
  • the system 11 preferably operates using one or more algorithms, in particular selected among LOF, COF, iF, KNN, HBOS and OCVSM on a set of three data defined by the weighted variation of glucose variable DCOB(t), from the weighted residual insulin variable ICOB(t) and by measuring the concentration of glucose g(t).
  • algorithms in particular selected among LOF, COF, iF, KNN, HBOS and OCVSM on a set of three data defined by the weighted variation of glucose variable DCOB(t), from the weighted residual insulin variable ICOB(t) and by measuring the concentration of glucose g(t).
  • the LOF, COF and KNN algorithms belong to the family of density-based methods. These approaches are based on the study of neighborhood values: an observation in a dense region is considered an inlier, or non-anomalous value, while a given point in a low-density region is considered as an outlier, or anomalous value.
  • these algorithms are also indicated as methods based on a distance criterion, or "distance-based” as well as density-based, since a datum surrounded by very close points (and therefore lies in a high-density region) is considered non-anomalous, while a datum very distant from the others (and therefore lies in a low-density region) is considered anomalous.
  • the basic idea of the LOF algorithm is to compare the density of a point with respect to its neighbor points instead of considering all the points in the data set. In this way, it is possible to identify regions with similar densities and points with a significantly lower density than its neighbors. The latter are identified as outliers.
  • the data processing unit 111 is apt to use at least the aforesaid unsupervised anomaly detection algorithm LOF on the entire feature set which describe the state of the system, that is at least to the three variables "variation of weighted glucose” DCOB(t), "weighted residual insulin” ICOB(t) and "glucose concentration measurement" g(t).
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value within the patient data set, i.e. a result outside a predefined threshold values range.
  • the COF algorithm is a modified version of LOF, more effective in the case of complex and low-dimensional data structures.
  • the data processing unit 111 is able to use at least the aforesaid unsupervised anomaly detection algorithm COF on the entire feature set which describe the state of the system, that is the weighted variation of glucose DCOB(t) variable, weighted residual insulin ICOB(t) variable and glucose concentration measurement.
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value within the patient data set, i.e. a result outside a predefined threshold values range.
  • KNN k-nearest-neighbors algorithm
  • the data processing unit 111 is adapted to use at least the aforementioned KNN algorithm to the entire set of preferred characteristics which describe the state of the system, or to the data of the variable of the weighted glucose variation DCOB(t), of the residual weighted insulin ICOB(t) and the measurement of glucose concentration.
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value between the patient data, or a result outside a predefined threshold values range.
  • the iF algorithm is based on the hypothesis that anomalous points must be few and isolated.
  • the main idea of this method is based on the concept of space partitioning: an isolated point, i.e. an anomaly, requires on average a smaller number of iterations of space partitions to be isolated through partitioning with respect to a non-anomalous value.
  • an isolated point - that is an anomalous datum - requires on average a smaller number of random partitions of the space - for example, a Cartesian plane in which the data are reported - to be completely isolated from the rest of the data.
  • a "normal" datum requires on average a greater number of partitions to be isolated, being very similar to the remaining data.
  • the data processing unit 111 is apt to use at least the aforesaid unsupervised anomaly detection algorithm iF on the entire feature set which describe the state of the system, that is the weighted variation of glucose DCOB(t) variable, weighted residual insulin ICOB(t) and glucose concentration.
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value within the patient data set, i.e. a result outside a predefined threshold values range.
  • OCSVM One Class Support Vector Machine
  • the data processing unit 111 is able to use at least the aforementioned OCSVM algorithm for the whole set of preferred characteristics which describe the state of the system, or the variable of weighted glucose variation DCOB(t), the weighted residual insulin ICOB(t) and glucose concentration.
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value within the patient data set, or a result outside a predefined threshold values range.
  • the Histogram-based Outlier Score (HBOS) algorithm belongs to methods based on the so-called “statistics-based” criterion. According to this criterion, an anomalous datum is identified using one or more statistical criteria such as, distance from the average, possibly normalized for the variance of the data.
  • the HBOS algorithm assigns the anomaly score to a datum with a value inversely proportional to the frequency with which data with analogous values are observed in the rest of the data set.
  • the data processing unit 111 is adapted to use at least the aforementioned HBOS algorithm to the whole set of preferred characteristics which describe the state of the system, or the variable variation of DCOB- weighted glucose (t), the residual insulin ICOB weighing (t) and glucose concentration.
  • the data processing unit 111 is capable of generating an alert signal if the algorithm detects an anomalous value within the patient data set, i.e. a result outside a predefined threshold values range.
  • the aforementioned algorithms are methods that effectively capture isolated points.
  • the points are strongly temporally correlated and never completely isolated, because the patient, a few moments before the current instant, is always in a condition not too dissimilar from the current condition. This causes the anomalies to be generally more difficult to detect than in other problems in which the KNN, LOF, COF, iF, OCSVM and HBOS methods are used as known art.
  • the data set that is the work portion of the aforementioned algorithms, excluding a predefined number of data in earlier time instants and a predefined number of data in later time instants with respect to a predetermined time instant time.
  • the normal data points will be even more embedded in a cluster region, while the anomalous points will be more isolated and possibly far from the defined cluster region.
  • this procedure allows the algorithm not to consider a set of strongly correlated data and therefore further decrease the density around anomalous data, thus increasing performance.
  • This selection is, in particular, possible thanks to the storage of the data in the storage unit 211 during their measurement.
  • the selection of the number of past and future time instants to be excluded is part of the design choices relating to the use of the relative algorithms and can be carried out in a tuning phase of system 11.
  • the anomaly detection system 11, and the relative artificial pancreas 1, according to the present invention is therefore efficient and adaptable to the individual patient.
  • the system 11 for the detection of malfunctions is model- free and unsupervised, therefore it is possible to determine the anomalies on the patient's data on which the medical device is applied and without needing to label the "failure / non-fault" to be supplied to the device and without needing to determine a mathematical model describing the patient's normal response.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Biomedical Technology (AREA)
  • Epidemiology (AREA)
  • Medical Informatics (AREA)
  • Business, Economics & Management (AREA)
  • Medicinal Chemistry (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Chemical & Material Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Anesthesiology (AREA)
  • Vascular Medicine (AREA)
  • Hematology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Medicines That Contain Protein Lipid Enzymes And Other Medicines (AREA)
  • Peptides Or Proteins (AREA)

Abstract

La présente invention concerne un système (11) permettant de détecter des dysfonctionnements dans des dispositifs (1) destinés à l'administration d'insuline comprenant une unité de traitement de données (111) adaptée à recevoir des mesures de la concentration de glucose à partir d'un capteur (21) de régulation de glucose et des mesures de la quantité d'insuline perfusée par une pompe à perfusion (31), une unité d'entrée de données (311) fonctionnellement connectée à l'unité de traitement de données (111) et adaptée à recevoir des estimations d'hydrates de carbone supposées à partir d'un utilisateur, une unité de stockage (211) fonctionnellement connectée à l'unité de traitement de données (111) et adaptée à stocker les mesures de concentration de glucose, les mesures de la quantité d'insuline perfusée, les mesures de la quantité d'hydrates de carbone prise ; dans lequel l'unité de traitement de données (111) est capable de calculer la valeur d'une variable DCOB (t) appelée "variation de glucose pondérée" : dans laquelle l'unité de traitement de données (111) est en outre adaptée à calculer une variable ICOB (t) appelée "insuline résiduelle pondérée" : dans lequel IOB(t) est la variable qui décrit l'estimation de l'insuline résiduelle dans le sang dans le temps, γ et δ sont deux constantes positives, et dans lequel l'unité de traitement de données (111) est capable d'exécuter au moins un algorithme de détection d'anomalie du type non supervisé configuré pour assigner un score d'anomalie à des données dans un ensemble de données de patient, sur la base d'au moins un critère sélectionné parmi : - "basé sur la densité", - "basé sur l'isolation", - 'basé sur des statistiques" et - "basé sur l'hypersurface", dans lequel chaque donnée de patient dans ledit ensemble, à un moment donné, comprend au moins une triade de valeurs, ladite triade de valeurs étant constituée d'une valeur de la variable DCOB(t) à un moment donné, d'une valeur de la variable ICOB(t) audit moment donné et d'une mesure de concentration de glucose par ledit capteur (21) audit moment donné, et dans lequel ladite unité de traitement de données (111) est capable de générer un signal d'alerte si ledit algorithme détecte au moins une anomalie à l'intérieur dudit ensemble de données de patient si ledit algorithme de détection d'anomalie génère un score d'anomalie parmi une plage prédéfinie de valeurs seuil.
PCT/IB2019/054470 2018-05-31 2019-05-30 Système de détection de dysfonctionnements dans des dispositifs de distribution d'insuline WO2019229686A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
IT102018000005918A IT201800005918A1 (it) 2018-05-31 2018-05-31 Sistema per la rilevazione di malfunzionamenti in dispositivi per la somministrazione di insulina
IT102018000005918 2018-05-31

Publications (1)

Publication Number Publication Date
WO2019229686A1 true WO2019229686A1 (fr) 2019-12-05

Family

ID=63762598

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2019/054470 WO2019229686A1 (fr) 2018-05-31 2019-05-30 Système de détection de dysfonctionnements dans des dispositifs de distribution d'insuline

Country Status (2)

Country Link
IT (1) IT201800005918A1 (fr)
WO (1) WO2019229686A1 (fr)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11317944B2 (en) 2011-03-14 2022-05-03 Unomedical A/S Inserter system with transport protection
US11458292B2 (en) 2019-05-20 2022-10-04 Unomedical A/S Rotatable infusion device and methods thereof
US11617827B2 (en) 2005-09-12 2023-04-04 Unomedical A/S Invisible needle
CN117612692A (zh) * 2024-01-19 2024-02-27 太原理工大学 一种基于连续血糖监测的胰岛素泵故障诊断系统及方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030060692A1 (en) * 2001-08-03 2003-03-27 Timothy L. Ruchti Intelligent system for detecting errors and determining failure modes in noninvasive measurement of blood and tissue analytes
US6558351B1 (en) * 1999-06-03 2003-05-06 Medtronic Minimed, Inc. Closed loop system for controlling insulin infusion
US20160193409A1 (en) * 2012-03-05 2016-07-07 Dexcom, Inc. Method to improve safety monitoring in type-1 diabetic patients by detecting in real-time failures of the glucose

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6558351B1 (en) * 1999-06-03 2003-05-06 Medtronic Minimed, Inc. Closed loop system for controlling insulin infusion
US20030060692A1 (en) * 2001-08-03 2003-03-27 Timothy L. Ruchti Intelligent system for detecting errors and determining failure modes in noninvasive measurement of blood and tissue analytes
US20160193409A1 (en) * 2012-03-05 2016-07-07 Dexcom, Inc. Method to improve safety monitoring in type-1 diabetic patients by detecting in real-time failures of the glucose

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
RUBEN ROJAS ET AL: "Multivariate statistical analysis to detect insulin infusion set failure", 2015 AMERICAN CONTROL CONFERENCE (ACC), 1 June 2011 (2011-06-01), pages 1952 - 1957, XP055547476, ISSN: 0743-1619, DOI: 10.1109/ACC.2011.5990951 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11617827B2 (en) 2005-09-12 2023-04-04 Unomedical A/S Invisible needle
US11317944B2 (en) 2011-03-14 2022-05-03 Unomedical A/S Inserter system with transport protection
US11458292B2 (en) 2019-05-20 2022-10-04 Unomedical A/S Rotatable infusion device and methods thereof
US11944775B2 (en) 2019-05-20 2024-04-02 Unomedical A/S Rotatable infusion device and methods thereof
CN117612692A (zh) * 2024-01-19 2024-02-27 太原理工大学 一种基于连续血糖监测的胰岛素泵故障诊断系统及方法
CN117612692B (zh) * 2024-01-19 2024-04-05 太原理工大学 一种基于连续血糖监测的胰岛素泵故障诊断系统及方法

Also Published As

Publication number Publication date
IT201800005918A1 (it) 2019-12-01

Similar Documents

Publication Publication Date Title
WO2019229686A1 (fr) Système de détection de dysfonctionnements dans des dispositifs de distribution d'insuline
US11628254B2 (en) System for monitoring and delivering medication to a patient and method of using the same to minimize the risks associated with automated therapy
EP2516671B1 (fr) Systèmes de gestion de dispositifs d'administration de médicaments
US20210299354A1 (en) Insulin delivery methods, systems and devices
CN109891510B (zh) 用于检测和应对低血糖响应不足的方法和装置
Eren-Oruklu et al. Adaptive system identification for estimating future glucose concentrations and hypoglycemia alarms
US20170332952A1 (en) Insulin delivery system and methods with risk based set points
CN109069003B (zh) 基于概率的控制器增益
Meneghetti et al. Model-based detection and classification of insulin pump faults and missed meal announcements in artificial pancreas systems for type 1 diabetes therapy
US20160193409A1 (en) Method to improve safety monitoring in type-1 diabetic patients by detecting in real-time failures of the glucose
CN116762136A (zh) 基于风险的胰岛素输送转换的系统和方法
EP3902469B1 (fr) Évaluation et visualisation d'un dysfonctionnement glycémique
US20220061706A1 (en) Techniques for image-based monitoring of blood glucose status
US20220062548A1 (en) Post meal compensation for automatic insulin delivery systems
Young et al. DAMON: A data authenticity monitoring system for diabetes management
Meneghetti et al. Fault detection in artificial pancreas: A model-free approach
CN117280423A (zh) 用于餐后血糖水平预测的方法和装置
Alshalalfah et al. Towards safe and robust closed-loop artificial pancreas using adaptive weighted PID control strategy
JP2022514785A (ja) 患者の血糖値を調節する自動システム
WO2019060991A1 (fr) Prévision de propriétés physiologiques
US11806137B2 (en) Real-time meal detection based on sensor glucose and estimated plasma insulin levels
Davari Benam et al. Estimation and Prediction of Glucose Appearance Rate for Use in a Fully Closed-Loop Dual-Hormone Intraperitoneal Artificial Pancreas
CN118302822A (en) Real-time meal detection based on sensor glucose and estimated plasma insulin levels
CN118302825A (en) Meal delivery to correct bolus
Ellouze et al. A Smart Healthcare: Methods based on WBAN and Multi-engine Artificial Intelligence.

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: 19734897

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19734897

Country of ref document: EP

Kind code of ref document: A1