CN102500013A - Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin - Google Patents

Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin Download PDF

Info

Publication number
CN102500013A
CN102500013A CN2011103727423A CN201110372742A CN102500013A CN 102500013 A CN102500013 A CN 102500013A CN 2011103727423 A CN2011103727423 A CN 2011103727423A CN 201110372742 A CN201110372742 A CN 201110372742A CN 102500013 A CN102500013 A CN 102500013A
Authority
CN
China
Prior art keywords
insulin
blood glucose
patient
design
dose
Prior art date
Application number
CN2011103727423A
Other languages
Chinese (zh)
Inventor
王友清
Original Assignee
北京化工大学
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 北京化工大学 filed Critical 北京化工大学
Priority to CN2011103727423A priority Critical patent/CN102500013A/en
Publication of CN102500013A publication Critical patent/CN102500013A/en

Links

Abstract

The invention provides an intelligent method based on model predictive control for automatic infusion of large doses of insulin, which is characterized in that: a continuous glucose monitoring system (CGMS) and an insulin pump are used as the basis of hardware; online detection of food is carried out based on a strong tracking filter; when the food is detected, large doses of insulin is transfused immediately, and the amount of insulin depends on historical food; whether the dosage is needed to be increased is judged every 30 minutes based on the model predictive control, and the dosage is designed; when the blood glucose concentration decreases, whether the basic amount of insulin is needed to be maintained is determined according to the predicted blood glucose value; and finally the blood glucose concentration is controlled to be within the safe range. Compared with the existing corresponding technology, the invention has the advantages of high degree of intelligence and fully automatic operation, and the blood glucose control effects can be improved significantly in case that self-management cannot be carried out by patients.

Description

基于模型预测控制的大剂量胰岛素全自动智能输注方法和 Based on automatic intelligent high-dose insulin infusion method and model predictive control

装置 Equipment

技术领域 FIELD

[0001] 本发明涉及的是一种自动设计大剂量胰岛素的智能方法,特别是一种基于模型预测控制的智能设计方法,属于生物医学工程领域。 [0001] The present invention relates to a method of automatically designing a smart large doses of insulin, in particular based on intelligent design for model predictive control, belonging to the field of biomedical engineering.

背景技术 Background technique

[0002] 随着物质生活水平的提高和环境的恶化,糖尿病日益成为严重危害人类健康的重要疾病。 [0002] With the deterioration of living standards and improve the environment, diabetes is becoming an important disease of serious harm to human health. 糖尿病成为了失明、肾衰竭和下肢截肢的主要原因,也是心血管并发症引起死亡的主要因素。 Diabetes has become the leading cause of blindness, kidney failure and lower limb amputation, as well as cardiovascular complications caused major factor in the death. 糖尿病是人类第四大死因,糖尿病导致的死亡人数占总死亡人数的6.8%,每年有近400万人死于糖尿病及其并发症。 Diabetes is the fourth leading cause of human deaths caused by diabetes accounted for 6.8% of deaths, nearly 400 million people each year die from diabetes and its complications. 每年全球11. 6%的医疗保健费用用于治疗糖尿病及其并发症。 11.6% of the annual global cost of health care for the treatment of diabetes and its complications.

[0003] 出现并发症的主要原因是糖尿病导致的高血糖(血糖浓度高于10mmol/L(毫摩尔每升))。 Mainly [0003] complications due to diabetes is hyperglycemia (blood glucose concentrations higher than 10mmol / L (millimoles per liter)). 为了降低血糖浓度,1型糖尿病患者和部分2型糖尿病患者需要外源性胰岛素治疗。 To reduce the blood glucose concentration, patients with type 1 and type 2 diabetes require portion exogenous insulin therapy. 目前,最先进的疗法是佩戴胰岛素泵,一天M小时实时输注。 Currently, the most advanced insulin pump therapy is worn day M-hour real-time infusion.

[0004] 为了模拟健康人体的胰岛素分泌规律,目前的胰岛素泵输注疗法包括两种模式: 大剂量胰岛素和基础量胰岛素。 [0004] In order to simulate the body's insulin secretion laws of health, the current insulin pump infusion therapy includes two modes: large doses of insulin and basal insulin. 基础量胰岛素用来补偿人体自身产生的血糖。 Basal insulin to compensate for the amount of glucose produced by the body itself. 大剂量胰岛素主要用来补偿饮食的影响和校正高血糖。 Large doses of insulin is mainly used to compensate for the effects of diet and correction of hyperglycemia. 由于饮食很难被实时测量,输注大剂量胰岛素需要使用者的参与,即使用者的自我管理:使用者在进食前后需要将饮食的时刻和大小输入给胰岛素泵。 Because the diet is difficult to measure in real-time, high-dose insulin infusion requires the participation of users, ie users of self-management: users will always need to eat before and after the diet and the size of the input to the insulin pump. 这里的饮食大小是指饮食中碳水化合物的含量。 Diet size here refers to the content of carbohydrates in the diet.

[0005] 为了准确估计饮食中的碳水化合物的含量,糖尿病患者需要接受严格而系统的食物营养知识教育。 [0005] In order carbohydrate content of the diet accurate estimate, diabetes patients need to undergo a rigorous and systematic knowledge of food and nutrition education. 即便如此,饮食大小的估计偏差也很难避免。 Even so, the size of the estimation error diet is difficult to avoid. 为了克服上述缺陷,不少智能胰岛素泵都引入了饮食数据库。 In order to overcome the above drawbacks, many smart insulin pump have introduced diet database. 患者只需要告诉胰岛素泵,吃了哪些东西以及各自的重量。 Patients only need to tell insulin pump, as well as something to eat what their weight. 胰岛素泵就会调用饮食数据库,自动计算碳水化合物的含量。 Insulin pump will call database diet, the carbohydrate content is automatically calculated. 这些先进的智能胰岛素泵部分地减轻了患者的使用负担。 These advanced smart insulin pump in part to reduce the burden on patients to use.

[0006] 然而,目前所有的胰岛素泵的大剂量都需要患者的参与设计,这给他们及其家人的生活带来了极大不便。 [0006] However, large doses of insulin pump all the needs of patients involved in the design, which gives them and their families lives has brought great inconvenience. 特别是很多1型糖尿病患者都是儿童和青少年,让他们每次进餐前把饮食信息输入给胰岛素泵往往是不现实的。 Particularly type 1 diabetes are a lot of children and adolescents to enable them to enter the food before each meal information to insulin pump are often unrealistic.

发明内容 SUMMARY

[0007] 本发明针对现有技术的不足和缺陷,提供了一种基于模型预测控制的全自动智能设计大剂量的方法。 [0007] The present invention is directed to deficiencies and drawbacks of the prior art, there is provided a method of automatic intelligent design based on high-dose model predictive control. 本发明的一个重要硬件基础是动态血糖监测系统(CGMS或者CGM),它的日益成熟使得实时测量血糖浓度成为了可能。 An important basic hardware of the present invention is glucose monitoring system (CGMS or CGM), such that it has become more sophisticated real-time measurement of blood glucose concentration as possible. 基于实时测量信息和反馈控制原理,本申请设计了一种全自动智能算法来自动设计大剂量胰岛素的时间和大小。 Based on real-time measurement and information feedback control theory, this application designed a fully automatic intelligent algorithm to automatically design time and size of large doses of insulin.

[0008] 与本发明有关的硬件结构简图如图1所示。 [0008] The hardware configuration diagram of the present invention pertains shown in Fig. 患者佩戴胰岛素泵,内含胰岛素,并在泵管理系统(大剂量胰岛素全自动智能算法)的控制下注射胰岛素。 Patient wears an insulin pump, containing insulin and injected insulin pump under the control of the management system (automatic intelligent high-dose insulin algorithm). 患者还佩戴CGMS,实时监测患者血糖浓度。 Patients also wear the CGMS, real-time monitoring of blood glucose concentrations in patients. 系统还可以包括一个手持的泵管理器,患者通过它进行系统的控制和设定。 The system may also include a hand-held pump manager, the patient to control and configure the system through it.

[0009] 本发明是通过以下技术方案实现的:首先,CGMS将当前的血糖值提供给设计算法所在的芯片并存入存储器;其次,利用强跟踪滤波器实时估计血糖浓度的变化率,当其超过某一阈值时,检测出饮食,并输注一个较保守的大剂量胰岛素;然后,启动模型预测控制算法,每半小时决策一次是否追加大剂量胰岛素;最后,当血糖浓度进入下降通道后,停止输注大剂量胰岛素,基于模型预测控制算法判断是否存在低血糖风险,从而决定是否关闭基础量胰岛素。 [0009] The present invention is achieved by the following technical solution: First, the CGMS will be provided to the current blood glucose level chip design and where the algorithm stored in the memory; Secondly, the use of strong tracking filter estimated in real time rate of change in blood glucose concentration, when it is exceeds a certain threshold, diet detected, a more conservative infusion and bolus insulin; then start the model predictive control algorithm, once every half hour decision whether additional bolus insulin; Finally, when the blood glucose concentration dropped into the channel, stop infusion of large doses of insulin, model-based predictive control algorithm to determine whether there is a risk of hypoglycemia, to decide whether to close the base amount of insulin. 最终期望将血糖水平保持在安全的范围内。 The final desired to keep blood sugar levels within a safe range.

[0010] 本发明主要具有饮食检测、大剂量自动设计、暂停基础量三个功能。 [0010] The present invention has detected the diet, large doses of automatic design, the amount of suspended three basic functions. 其中,饮食检测可以用来判断患者有没有进食,如果有进食,就马上输注初始大剂量并启动模型预测控制算法,每半小时判断一下是否追加大剂量;大剂量的自动设计包含两部分:初始大剂量和追加大剂量,用来补偿饮食的影响;暂停基础量是为了预防低血糖的发生。 Among them, dietary detection can be used to determine the patient has not eaten, if there is to eat, the initial high-dose infusion immediately and start the model predictive control algorithm, every half hour determine what additional large doses; large doses of automated design consists of two parts: the initial high-dose and high-dose added, to compensate for the effects of diet; suspend basic amount is to prevent the occurrence of hypoglycemia.

附图说明 BRIEF DESCRIPTION

[0011] 图1是本发明所述大剂量胰岛素全自动智能输注装置示意图; [0011] FIG. 1 is a schematic diagram of the intelligent device of the present invention is an infusion bolus insulin automatic;

[0012] 图2是本发明所述述大剂量胰岛素全自动智能输注方法中估计血糖浓度变化率的强跟踪滤波器算法流程图; [0012] FIG 2 is a flowchart of the invention the filter algorithm described strong tracking the rate of change of blood glucose concentration bolus insulin infusion intelligent automatic estimation method;

[0013] 图3是本发明所述述大剂量胰岛素全自动智能输注方法步骤流程图; [0013] FIG. 3 of the present invention is the automatic intelligent said bolus insulin infusion flow chart of method steps;

[0014] 图4是是本发明所述大剂量胰岛素全自动智能输注装置系统结构图。 [0014] FIG. 4 is that the present invention bolus insulin infusion system of automatic intelligent configuration of FIG.

具体实施方式 Detailed ways

[0015] 下面结合附图和具体实施例对本发明作进一步说明,但不作为对本发明的限定。 [0015] The following examples of the present invention will be further described in conjunction with the accompanying drawings and the specific embodiments, but not limitative of the present invention.

[0016] (1)饮食检测算法。 [0016] (1) eating detection algorithm.

[0017] 记k时刻的血糖值为G (k),血糖变化率为G' (k),假设血糖变化率较稳定,则可以 [0017] referred to the value of blood glucose at time k G (k), the blood glucose rate of change G '(k), assuming the rate of change in blood glucose is stable, can be

得到如下二阶线性动态模型: Second order linear dynamic model obtained as follows:

'\G(k + \)~] 「1 Δί]「(?(Λ)] '\ G (k + \) ~] "1 Δί]" (? (Λ)]

= + w(k) = + W (k)

G\k + \) 0 1 G\k) v , G \ k + \) 0 1 G \ k) v,

[0018] ] 「 ] (1) [0018]] "] (1)

r ]「G⑷ Ί y(k) = [1 o] +v(k) r] "G⑷ Ί y (k) = [1 o] + v (k)

[0019] 其中,At为采样周期,通常为1分钟或5分钟,已知;w(k)为建模不确定性或外部干扰,未知;v(k)为测量噪声,未知;y(k)为CGMS的输出值,已知;G(k)代表真实的血糖浓度值,未知,不过y(k)为它的测量值,因此两者十分接近;G' (k)代表血糖变化率,未知, 且不可测。 [0019] wherein, At is the sampling period, typically 1 minute or 5 minutes, is known; w (k) is a disturbance or a modeling uncertainty, the unknown; v (k) is the measurement noise, is unknown; y (k ) is the output value of CGMS is known; G (k) represent the true values ​​of glucose concentration, is unknown, but y (k) measured for its value, both very close; G '(k) representative of the rate of change in blood glucose, unknown and unpredictable. 基于上述动态模型和实时测量值y(k),可以用强跟踪滤波器来估计血糖变化率G' (k),其估计值记作0'(幻。强跟踪滤波器具有如下优点:1)对模型不确定性具有较强的鲁棒性;幻对突变状态有较强的跟踪能力,甚至在系统达到平衡状态时,仍保持对缓变状态和突变状态的跟踪能力。 (Phantom has the advantage that strong tracking filter: 1) based on the dynamic model and real-time measurements y (k), can be strong tracking filter to estimate the rate of change in blood glucose G '(k), which is referred to as estimated values ​​0' to have model uncertainty robust; the magic of mutation status has a strong ability to track, even when the system reaches an equilibrium state, it remains the ability to track the status and graded mutation status. 因此,强跟踪滤波器是检测饮食的有效工具,特别适合检测血糖变化率。 Thus, strong tracking filter detecting diet is an effective tool, especially adapted to detect the rate of change in blood glucose.

[0020] 下面结合附图具体对如何估计G' (k)作进一步说明。 [0020] The following accompanying drawings of how to estimate G (k) be 'binding further described. [0021 ] 强跟踪滤波器算法估计0»/、「G⑷ 1 t、「1 Δί] [0021] Strong tracking filter algorithm to estimate 0 »/," G⑷ 1 t, "1 Δί]

[0022] 1)令X(Jc)= ;/ F(k)= H(k) = [1 0] [0022] 1) Let X (Jc) =; / F (k) = H (k) = [1 0]

Lr (Jc) , U 1 ' Lr (Jc), U 1 '

[0023] 2)令k = 0 ;选择初始值10), P(0|0) (P(k|k)为估计方差;P(k+l|k)为预测方 [0023] 2) Let k = 0; selecting an initial value of 10), P (0 | 0) (P (k | k) is the estimated variance; P (k + l | k) is the square prediction

差);选择一个合适的弱化因子β。 Difference); selecting a suitable weakening factor β.

[0024] 3)由式 [0024] 3) by the formula

[0025] X(k + l\k) = F(k)X (k\k\ y(k + 1) = y(k +1)-H(k)X(k +1 \k) [0025] X (k + l \ k) = F (k) X (k \ k \ y (k + 1) = y (k +1) -H (k) X (k +1 \ k)

[0026]计算 +1 \k), y(k +1)。 [0026] Calculation +1 \ k), y (k +1).

[0027] 由式 [0027] represented by the formula

Kl)/(1), k = 0 Kl) / (1), k = 0

[0028] ■ + = ] [pS0 (k) + r(k + \)γτ (k +1)] [0028] ■ + =] [pS0 (k) + r (k + \) γτ (k +1)]

-, KL -, KL

{ l + p {L + p

[0029]计算 &(k+l), [0029] Calculation & (k + l),

[0030] 由式 [0030] represented by the formula

[0031 ] N (k+1) = S0 (k+1) -H (k) Q (k) Ht (k) - β R (k+1), [0031] N (k + 1) = S0 (k + 1) -H (k) Q (k) Ht (k) - β R (k + 1),

[0032] M (k+1) = H (k) F (k) P (k I k) Ft (k) Ht (k), [0032] M (k + 1) = H (k) F (k) P (k I k) Ft (k) Ht (k),

[0033] ' [0033] '

[0035] 计算出次优渐消因子λ (k+1)。 [0035] The calculated suboptimal fading factor λ (k + 1).

[0036] 4)根据式 [0036] 4) according to formula

[0037] P (k+11 k) = λ (k+1) F (k) P (k I k) Ft (k) +Q (k), [0037] P (k + 11 k) = λ (k + 1) F (k) P (k I k) Ft (k) + Q (k),

[0038]计算 P (k+1 I k); [0038] calculating P (k + 1 I k);

[0039] 由式 [0039] represented by the formula

[0040] K (k+1) = P (k+11 k) Ht (k) · [H (k) P (k+11 k) Ht (k) +R (k+1) ] [0040] K (k + 1) = P (k + 11 k) Ht (k) · [H (k) P (k + 11 k) Ht (k) + R (k + 1)]

[0041]算出 K (k+1); [0041] The calculated K (k + 1);

[0042] 最终得到由 [0042] finally obtained from the

[0043] X(k + \\k + \) = X(k + \ \k) + K(k + \)y(k +1) [0043] X (k + \\ k + \) = X (k + \ \ k) + K (k + \) y (k +1)

[0044]状态估计值 +1 +1)。 [0044] The state estimate +1 +1).

[0045] 5)更新P (k+1 I k+1) [0045] 5) Update P (k + 1 I k + 1)

[0046] P (k+11 k+1) = [IK (k+1) H] P (k+11 k) [0046] P (k + 11 k + 1) = [IK (k + 1) H] P (k + 11 k)

[0047] 6) k+1 — k转向3),继续循环。 [0047] 6) k + 1 - k steering 3), the cycle continues.

[0048] 上述算法的详细流程图见图2。 Detailed flowchart [0048] The algorithm shown in Figure 2. 通过上述算法,可以得到X(k)的估计值1@|幻,进而得到G' (k)的估计值。 By the above-described algorithm can be obtained by X (k) is an estimated value 1 @ | magic thus obtained estimated value G '(k) of the.

[0049] 空腹状态下,G' (k)的取值较小,进食后,G' (k)就会变得很大。 [0049] The fasting state, G 'value (k) is small, after eating, G' (k) becomes large. 因此,系统可以设计一个阈值η >0。 Thus, a system may be designed threshold η> 0. 当ό'(幻>/;时,检测出饮食。阈值n的取值因人而异,需要针对不同患者的历史数据进行统计分析,找到最佳的阈值(通常,阈值在1. 5-3mg/dL/min之间)。[0050] (2)大剂量自动设计 When ό '(phantom> /; when the detected value of threshold n diet vary, the need for statistical analysis of historical data for different patients, to find the best threshold value (typically, the threshold value 1. 5-3mg. between / dL / min). [0050] (2) high dose automatic design

[0051] 本部分的大剂量包括两部分:检测出饮食后马上输注的大剂量;启动模型预测控制算法后,每半小时判断是否追加大剂量。 [0051] The large doses of this section consists of two parts: detecting the high dose infusion immediately after eating; after starting model predictive control algorithm, determines whether or not an additional half hour every large doses.

[0052] 在设计第一个大剂量时,由于系统对饮食大小未知,因此,可以利用具体患者的历史饮食信息来设计一个保守的大剂量。 [0052] In the design of the first large doses, because the system of food size is unknown, therefore, can be used beverage particular patient history to design a conservative large doses. 例如,该患者的历史平均饮食大小为M,可以根据M/2或M/3来设计大剂量,从而提高系统的安全性。 For example, the historical average size of the patient's diet is M, large doses can be designed according to M / 2 or M / 3, thereby enhancing the security of the system. 为了进一步提高系统的安全性,还可以增加一个判断单元:如果血糖水平高于某个阈值(例如140mg/dL),才输注大剂量胰岛素。 To further enhance the security of the system, a determination can also increase the unit: If the blood sugar level is higher than a certain threshold (e.g. 140mg / dL), large doses of insulin was infused.

[0053] 检测出饮食后,模型预测控制算法就会被启动。 After the [0053] detect the diet, model predictive control algorithm will be started. 使用模型预测控制,就需要一个预测模型,本发明采用如下基于离散传递函数的预测模型: Using the model predictive control, requires a predictive model, the present invention adopts the following prediction based on the discrete transfer function:

K [0054] K [0054]

Figure CN102500013AD00071

[0055] 其中,G(k),I (k)和M(k)分别表示k时刻的血糖浓度、胰岛素输注速度和饮食大小;ζ—1为后移算子,例如Z^3G (k) = G(k-3)为三步后移项。 [0055] wherein, G (k), I (k) and M (k) represent the blood glucose concentration at time k, the size of the insulin infusion rate and diet; ζ-1 after shift operator, e.g. Z ^ 3G (k ) = G (k-3) after three-step transposition. 系数K1, Cl1, ai; bI; KM, dM,aM, bM 由开环辨识得到,因此已知。 Coefficients K1, Cl1, ai; bI; KM, dM, aM, bM obtained by an open loop identification, is therefore known. K1反映了胰岛素输注速度对血糖水平的影响大小;屯代表胰岛素的起效时间大小;&反映了上一步血糖浓度G(kl)对当前血糖浓度G(k)的影响程度反映了G(k-2)对G(k)的影响程度;KM, dM,aM, bM的物理意义类似。 K1 insulin infusion rate reflects the magnitude Effects on blood glucose levels; Tun size representative of onset time of insulin; & reflects the glucose concentration step G (kl) impact on the current blood glucose concentration G (k) reflects the G (k -2) impact on G (k); a similar physical meaning KM, dM, aM, bM of. 由公式(2)我们知道, 当前时刻的血糖浓度可以由胰岛素输注速度、饮食大小、上一步的血糖浓度和上两步的血糖浓度来预测。 By equation (2) We know that current blood glucose concentration time can be, the size of diet, blood glucose concentration in the previous step and two-step on the blood glucose concentration to predict insulin infusion rate.

[0056] 上述系统中,I (k)和G(k)可以实时测量,饮食大小M(k)很难被实时测量,因此可以被看作一个未知输入。 [0056] The system, I (k) and G (k) can be measured in real time, diet size M (k) is measured in real time is difficult, it can be seen as an unknown input. 然而,进食通常在10分钟内完成;采样周期通常选为5分钟,dM为饮食影响血糖浓度的滞后步长,通常为2,即滞后时间为2*5 = 10分钟。 However, eating is usually complete within 10 min; sampling periods usually preferably 5 minutes, to the diet affect blood glucose concentration dM lag step size, typically 2, i.e., the lag time is 2 * 5 = 10 minutes. 因此,大概10+2*5 =20分钟后,饮食就不会直接影响血糖预测值。 Therefore, about 10 + 2 * 5 = 20 minutes, it will not directly affect blood sugar diet predictive value. 换句话说,基于预测模型O),20分钟后的血糖预测值不依赖饮食大小,因此本发明就可以回避掉饮食大小未知的难点,实现准确自动预测。 In other words, based on a prediction model O), 20 minutes after the glucose prediction value does not depend on the size of the diet, so the present invention can avoid unknown size difficulty eating out, automatic accurate prediction.

[0057] 选择如下评价指标(代价函数),评价指标给出判断一个控制策略优劣的标准: [0057] select the following evaluation (cost function), evaluation criteria are given merits of a control strategy:

NI ί 2 -,21 NI ί 2 -, 21

[0058] ^l = Yj!\G(k + dI+i\k)-GR(k + dI +i) +Α[Δ/(Λ + ψ)」 (3) [0058] ^ l = Yj! \ G (k + dI + i \ k) -GR (k + dI + i) + Α [Δ / (Λ + ψ) "(3)

i=0 V ) i = 0 V)

[0059] 其中,N是预测步长,0(7+|幻是在k时刻对j时刻的血糖水平的预测值,(^是血糖设定值,Δ/是设计的未来的胰岛素输注速度的差分值。上述评价指标其实包含两部分:第一部分反映了未来血糖水平与设定值之间的差距;第二部分反映了胰岛素输注速度的变化程度。参数λ用来调整两者之间的权重,由系统来设计和定义。 [0059] where, N is the prediction steps, 0 (7+ | magic on blood glucose levels is the predicted value of the time j at time k, (^ setpoint blood glucose, Δ / future is to design the insulin infusion rate . the difference value evaluation described above actually contains two parts: the first part reflects the gap between future blood glucose level with a set value; the second part reflects the degree of change in insulin infusion rate is used to adjust the parameters λ therebetween. the weight of the system design and definition.

[0060] 最优的胰岛素输注速度序列需要使得评价指标最小,S卩,使得未来一段时间内的血糖跟踪残差平方与胰岛素输注速度的变化率的平方之和最小: [0060] Optimal insulin infusion rate required sequence such that the minimum evaluation index, S Jie, so that the next time the blood glucose tracking the square of the square of the residual rate of change and the minimum insulin infusion rate:

[0061] AI\k + i \k)i=^ H = arg min Ω (4) [0061] AI \ k + i \ k) i = ^ H = arg min Ω (4)

Δί Δί

[0062] 得到1后,很容易得到最优胰岛素输注速度序列, 这就是在评价指标Ω下的最优的控制策略。 After the [0062] 1 obtained, it is easy to get the optimal insulin infusion rate sequence, which is the optimal control strategy in the evaluation of Ω. 假设血糖预测值0在时刻k+djj达到最大值。 Suppose the predicted value of blood glucose at time 0 k + djj maximum. 那么,大剂量胰岛素可以设计为: So, large doses of insulin can be designed to:

7[0063] 7 [0063]

Figure CN102500013AD00081

[0064] (3)暂停基础量 [0064] (3) the amount of suspended base

[0065] 当血糖进入下降通道后,就要停止输注大剂量胰岛素了,这是为了避免低血糖事件。 [0065] When the blood sugar to enter a downward spiral, it must stop the infusion of large doses of insulin, which is to avoid hypoglycemic events. 因为低血糖事件的短期危害很大,可以造成脑中风甚至死亡。 Because short-term harm hypoglycemic events is large, it can cause stroke or even death.

[0066] 为了进一步降低低血糖的发生概率,需要在线预测低血糖事件,及时关闭基础量胰岛素。 [0066] In order to further reduce the probability of occurrence of hypoglycemia, we need to predict hypoglycemic events online, close the basal insulin. 令公式O)中的I(k)为基础量、M(k)为零,可以预测未来的血糖浓度,如果预测时长(通常为30-60分钟)后的血糖预测值低于某低血糖阈值(如70mg/dL),就关闭基础量60-90分钟。 Order equation O) of I (k) based on the amount of, M (k) is zero, can predict future glucose levels, blood glucose predicted value if the predicted length of time (typically 30-60 minutes) below a certain threshold hypoglycaemia (e.g. 70mg / dL), the amount of base closed for 60-90 minutes.

[0067] 本发明具有实质性特点和显著进步。 [0067] The present invention has substantive features and notable progress. 基于模型预测控制的大剂量胰岛素自动设计方法是在性能良好的强跟踪滤波器技术、模型预测控制技术、自动检测技术等的基础上研发的。 Based on method of automatically designing bolus insulin model predictive control is based on the good performance of strong tracking filter technique, model predictive control, automatic detection of the developed technology. 该发明方法的详细流程图见图3。 A detailed flowchart of the method of the present invention shown in Figure 3.

[0068] 图3中的算法如下所述: [0068] FIG. 3 of the algorithm is as follows:

[0069] 基于模型预测控制的大剂量自动设计算法: [0069] large doses of automatic design algorithm based on model predictive control:

[0070] 1)系统初始化; [0070] 1) system initialization;

[0071] 2)接收到一个新的CGMS读数后,存入存储器并输入强跟踪滤波器计算最新的血糖变化率; [0071] 2) receives a new reading of the CGMS, and enter into the memory the latest strong tracking filter calculated rate of change of blood glucose;

[0072] 3)如果血糖变化率低于阈值(通常为1. 5-3mg/dL/min之间),等待下一个CGMS 读数,并回到步骤2);如果超过阈值,进入下一步; [0072] 3) If the rate of change in blood glucose is below a threshold (typically between 1. 5-3mg / dL / min), waiting for the next reading CGMS, and returns to step 2); if it exceeds the threshold, the next step;

[0073] 4)注射保守的初始大剂量胰岛素; [0073] 4) the initial injection of bolus insulin conserved;

[0074] 5)每30分钟,基于模型预测控制算法设计一次大剂量胰岛素; [0074] 5) every 30 minutes, model-based predictive control algorithm design of a large dose of insulin;

[0075] 6)实时判断血糖浓度是否开始下降:否,则回到步骤5);是,则进入下一步; [0075] 6) detects whether blood glucose concentration begins to fall: NO, the process returns to step 5); YES, the process proceeds to the next step;

[0076] 7)实时判断未来是否有低血糖风险:否,重复步骤7);是,则暂停基础量胰岛素; [0076] 7) in real time to determine whether the future risk of hypoglycemia: No, repeat step 7); that is suspended basal insulin;

[0077] 8)血糖进入安全范围后,结束上述过程。 [0077] 8) After the blood enters the safety range, the end of the above-described process.

[0078] 与已有的开环胰岛素输注方法相比,该发明具有高智能、全自动的特点,可以在缺乏患者自我管理的情形下显著提高血糖控制效果。 [0078] Compared with the existing method of open-loop insulin infusion, the invention has the highly intelligent, automatic features, can significantly improve glycemic control in the case of absence of patient self-management.

[0079] 图4为本发明优选实施例硬件结构框图,现结合图4进一步详细介绍。 [0079] Figure 4 a block diagram showing a hardware configuration examples of preferred embodiments, is now described in further detail in conjunction with FIG. 4 of the present invention.

[0080] 在患者身体上,设置有胰岛素泵41,内含一定剂量的胰岛素,可以在自动设计大剂量胰岛素系统的控制下进行精确量注射。 [0080] On the body of the patient, insulin pump 41 is provided, containing a dose of insulin can be injected at a precise control of the amount of bolus insulin automatic design system. 患者身体还设置有CGMS动态血糖监测系统42,用于实时测量糖尿病患者血糖浓度,并将数据发送给自动设计大剂量胰岛素系统的控制器。 The patient's body is also provided with CGMS glucose monitoring system 42, for real-time measurement of blood glucose concentration in diabetic patients, the controller sends the data to the automatic design system bolus insulin.

[0081] 患者体外的部件为自动设计大剂量胰岛素系统控制模块,包括以下部件:通信模块43、存储器44、控制器45,、用户输入模块46、显示模块47、供电模块(未在图4中表明)。 [0081] member outside the patient bolus insulin automatic design system control module, comprising the following components: a communication module 43, a memory 44, a controller 45,, the user input module 46, a display module 47, a power supply module (not shown in FIG. 4 show). 上述体外的控制模块设计为如手机大小,便于携带,操作方便。 Outside the body, such as mobile phones module is designed to control the size, easy to carry, easy to operate. 其中,通信模块43用于胰岛素泵41、CGMS动态血糖监测系统42和控制器45之间的数据传输,控制器45发送指令给胰岛素泵41、胰岛素泵41发送胰岛素注射数据返回给控制器45 ;CGMS42通过它实时传输数据给控制器45。 Wherein the communication module 43 for an insulin pump 41, data transmission between the controller 42 and the CGMS glucose monitoring system 45, controller 45 sends instructions to 41 insulin pump, insulin insulin pump 41 transmits data back to the controller 45; CGMS42 through which real-time transmission of data to the controller 45. 存储器44用于存储系统程序数据,还有患者的历史饮食量数据、历史血糖浓度等病患数据。 System memory 44 for storing program data, as well as food intake historical patient data, patient historical data of the blood glucose concentration. 通过用户输入模块46,患者可以向系统输入控制信息,对胰岛素注射量进行手工控制。 Through the user input module 46, the patient can input information to the system control, for manual control of the amount of insulin injection. 显示模块47,用于向患者显示胰岛素注射信息、血糖浓度信息、血糖过高或过低报警信息等数据。 The display module 47 for displaying information insulin, blood glucose concentration information, too high or too low blood glucose alarm information to the patient data. 控制器45控制整个系统,根据CGMS读数估计血糖变化率,判断患者有进食后输入初始大剂量胰岛素,每隔30分钟注射一次大剂量胰岛素,直到血糖值进入安全范围,其又包括饮食监测模块451、血糖预测模块452和大剂量设计模块453。 The controller 45 controls the entire system, according to the CGMS glucose readings estimated rate of change, the input is determined after the initial bolus insulin in patients with eating, injected once every 30 minutes bolus insulin until blood glucose level into the safe range, which in turn includes monitoring module 451 diets , blood glucose prediction module 452 and high-dose design module 453. 饮食监测模块451用于根据患者血糖变化率判断患者是否有进食;血糖预测模块452用于预测未来时间患者的血糖浓度值;大剂量设计模块453用于根据血糖预测模块452计算得到的未来血糖水平与设定值之间的差距,以及胰岛素输注速度的变化程度,来计算每隔30分钟应该注射的大剂量胰岛素数量。 The diet for blood glucose monitoring module 451 determines whether the rate of change of patient eating a patient; glucose prediction module 452 for predicting a future time value of the blood glucose concentration of the patient; high dose design module 453 to calculate the blood glucose level of blood glucose prediction module 452 next the gap between the set value and the degree of change in the insulin infusion rate to calculate the number of doses of insulin every 30 minutes should be large for injection.

[0082] 另外,基于标准糖尿病新陈代谢模型,在MATLAB上开发了糖尿病仿真平台,该仿真平台中含有100名虚拟病人。 [0082] In addition, the standard model of diabetes metabolism, diabetes developed in the MATLAB simulation platform, the simulation platform contains 100 virtual patients. 本发明方法在全部虚拟病人上进行了测试。 The method of the present invention were tested in all of the virtual patient.

[0083] 仿真测试从0:00开始,全部虚拟病人的初始血糖水平选为110mg/dL ;在1:00,所有病人进食,饮食中碳水化合物的含量根据病人体重不同而有所区别,分布在45-85克之间;约在10-25分钟内,强跟踪滤波器检测出了饮食,并输注了初始大剂量;然后启动模型预测控制设计追加大剂量,追加大剂量的数目为0-3次不等;血糖进入下降通道后,停止大剂量,并根据血糖预测结果来决定是否暂停基础量;仿真试验在12:00结束。 [0083] simulation test from 0:00 all the virtual patient's blood glucose level selected as the initial 110mg / dL; at 1:00, all patients eating carbohydrate content of the diet depending on patient weight and differ, distribution between 45-85 g; within about 10-25 minutes, strong tracking filter diet detected, and an initial high dose infusion; model predictive control then start adding high dose, a large number of doses of an additional 0-3 times vary; blood glucose entered a downward spiral, stop large doses, and to decide whether to suspend the basis of the amount of blood glucose prediction; simulation ended at 12:00. 在整个试验中,平均血糖浓度约为140mg/dL,发生低血糖的时间百分比低于1%。 Throughout the test, the average blood glucose concentration of about 140mg / dL, the percentage of time less than 1% of hypoglycemia.

[0084] 本发明方法经过仿真测试,效果较理想。 [0084] The method of the present invention through simulation test, the effect is desirable. 智能算法的引入,使得自我管理缺失的情形下血糖控制效果有很大的改观。 The introduction of intelligent algorithms, making glycemic control in the case of lack of self-management has greatly improved. 此外,由于该方法简单易行,计算量较低,可以很容易地嵌入到目前的胰岛素泵中,改善其效果。 Further, since the method is simple, relatively low amount of calculation can easily be fitted to existing insulin pump, the improvement effect.

[0085] 以上所述的实施例,只是本发明较优选的具体实施方式,本领域的技术人员在本发明技术方案范围内进行的通常变化和替换都应包含在本发明的保护范围内。 [0085] The above-described embodiments, but the present invention is more preferred embodiments, variations and substitutions typically skilled in the art will be within the technical scope of the present invention should be included within the scope of the present invention.

Claims (9)

1. 一种基于模型预测控制的大剂量胰岛素设计和输注方法,其特征在于本方法包括如下步骤:(1)利用动态血糖监测系统(CGMQ输出的血糖浓度值,估计血糖变化率,根据血糖变化率,判断患者是否有进食;(2)如果判断患者确实进食,则马上输注初始大剂量胰岛素,并进行未来血糖水平预测,每半个小时判断一次是否追加大剂量胰岛素;(3)实时监测血糖浓度是否开始下降:如果没有下降,则返回步骤O);如果开始下降, 则判断未来是否有低血糖风险,如果有低血糖风险,则暂停基础胰岛素注射。 1. Based on the design and bolus insulin infusion model predictive control methods, characterized in that the method comprises the steps of: (1) using a glucose monitoring system (blood glucose concentration value CGMQ output, estimated rate of change of blood glucose, blood glucose the rate of change, to determine whether the patient has to eat; (2) determine if the patient does eating, the initial high-dose insulin infusion immediately, and blood glucose levels to predict the future, once every half hour to determine whether or not additional bolus insulin; (3) Real-time whether monitoring blood glucose levels began to decline: If you do not fall, it returns to step O); if starts to drop, it is determined whether the future risk of hypoglycemia, if there is the risk of hypoglycemia, then pause basal insulin injections. (4)血糖进入安全范围后,停止注射胰岛素。 (4) glucose into the safe range, stop the injection of insulin.
2.根据权利要求1所述的大剂量胰岛素设计方法,其特征在于:步骤(1)中,使用二阶线性动态模型和强跟踪滤波器来估计血糖变化率,根据血糖变化率是否大于某阈值来检测是否有进食。 The bolus insulin design method according to claim 1, wherein: step (1), a second-order linear dynamic model and the Strong tracking filter to estimate the rate of change of blood sugar, blood sugar rate of change is greater than a threshold value to detect whether there eating.
3.根据权利要求2所述的大剂量胰岛素设计方法,其特征在于:判断是否有进食的血糖变化率阈值根据患者的历史数据统计分析得到,通常可以选择1. 5-;3mg/dL/min(毫克每升每分钟)之间。 3. The method of design of the large dose of insulin according to claim 2, wherein: determining whether a change in blood glucose consumption rate threshold based on statistical analysis of historical data of the patient is obtained, usually choose 1. 5-; 3mg / dL / min (milligrams per liter per minute) between.
4.根据权利要求1所述的大剂量胰岛素设计方法,其特征在于:步骤(¾中,在饮食刚被检测出来的时刻,初始大剂量由患者历史平均饮食大小确定;为了提高安全性,利用历史平均饮食大小的1/3至1/2来计算初始大剂量。 The bolus insulin design method according to claim 1, wherein: the step (¾ in the diet just been detected at the time, the initial dose is determined by the large size of the patient historical average diet; To improve security, the use of diet historical average size 1/3 to 1/2 to calculate the initial high dose.
5.根据权利要求1所述的大剂量胰岛素设计方法,其特征在于:步骤O)中,由基于离散传递函数的两输入单输出预测模型来预测未来血糖水平。 The bolus insulin design method according to claim 1, wherein: step O), the basis of the discrete two input single-output transfer function of the prediction model to predict future glucose levels.
6.根据权利要求5所述的胰岛素注射方法,其特征在于:追加大剂量由模型预测控制方法来设计,由下面两个因素决定:未来血糖水平与设定值之间的差距;以及胰岛素输注速度的变化程度。 6. The insulin injection method as claimed in claim 5, wherein: an additional high dose predicted by the model to design the control method is determined by two factors: the gap between the blood glucose level with a set value in the future; and insulin output Note the degree of change of speed.
7.根据权利要求1所述的大剂量胰岛素设计方法,其特征在于:步骤(3)中,血糖进入下降通道后,基于血糖预测值,判断是否暂停注射胰岛素;预测步长选为30-60分钟;暂停时长为60-90分钟。 The bolus insulin design method according to claim 1, wherein: step (3), the blood glucose dropped into the channel, based on the predicted blood glucose value, it is determined whether to suspend the injection of insulin; prediction step preferably 30-60 minutes; 60-90 minutes long pause.
8. 一种基于模型预测控制的大剂量胰岛素设计和输注装置,其特征在于:在患者身体上,设置有:胰岛素泵41,内含一定剂量的胰岛素,可以在自动设计大剂量胰岛素系统的控制下进行精确量注射;和CGMS动态血糖监测系统42,用于实时监测糖尿病患者血糖浓度,并将数据发送给自动设计大剂量胰岛素系统的控制器;患者体外设置有自动设计大剂量胰岛素系统控制模块,包括以下部件:通信模块43用于胰岛素泵41、CGMS动态血糖监测系统42和控制器45之间的数据传输,存储器44用于存储系统程序数据,还有患者的历史饮食量数据、历史血糖浓度等病患数据;用户输入模块46,用于患者向系统输入控制信息,进行控制;显示模块47,用于向患者显示胰岛素注射信息、血糖浓度信息、血糖过高或过低报警信息等数据;控制器45,控制整个系统,根据CGMS读数估计 A model predictive control based on high-dose insulin infusion and design, which is characterized in that: in the body of the patient, is provided with: an insulin pump 41, containing a dose of insulin, insulin dose can be automatic design system in the large under the control of injection quantity accurately performed; and the CGMS glucose monitoring system 42, for real-time monitoring of blood glucose in patients with diabetes concentration, the controller automatically sends the data to the system design of large doses of insulin; outside the patient provided with an automatic control system designed bolus insulin module, comprising the following components: a communication module 43 for insulin pumps 41, 42 and the data transfer between the controller 45 CGMS glucose monitoring system, a system program memory 44 for storing data, as well as food intake historical patient data, history patient blood glucose concentration data; a user input module 46, a patient information input to the system control, control is performed; the display module 47 for displaying information insulin, blood glucose concentration information, low blood sugar or high alarm information and the like to a patient data; and a controller 45, controls the entire system, according to the estimated readings CGMS 糖变化率,判断患者有进食后输入初始大剂量胰岛素,每隔30分钟注射一次大剂量胰岛素,直到血糖值进入安全范围。 Sugar change rate, the input is determined after the initial bolus insulin in patients with eating, injected once every 30 minutes bolus insulin until blood glucose level into the safe range.
9.根据权利要求8所述的大剂量胰岛素设计和输注装置,其特征在于,控制器45包含如下模块:饮食检测模块451,用于根据患者血糖变化率判断患者是否有进食; 血糖预测模块452,用于预测未来时间患者的血糖浓度值;大剂量设计模块453,用于根据血糖预测模块452计算得到的未来血糖水平与设定值之间的差距,以及胰岛素输注速度的变化程度,来计算每隔30分钟应该注射的大剂量胰岛素数量。 According to claim 8, wherein said high-dose insulin infusion and design, wherein, the controller 45 comprises the following modules: Diet detection module 451, according to the rate of change of blood glucose in patients with eating determines whether a patient; glucose prediction module 452, for predicting a future time value of the blood glucose concentration of the patient; high dose design module 453, according to the difference between blood glucose levels and blood glucose prediction set value calculated by the next module 452, and the degree of change of the insulin infusion rate, calculating the number of doses of insulin every 30 minutes should be large for injection.
CN2011103727423A 2011-11-22 2011-11-22 Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin CN102500013A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2011103727423A CN102500013A (en) 2011-11-22 2011-11-22 Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2011103727423A CN102500013A (en) 2011-11-22 2011-11-22 Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin

Publications (1)

Publication Number Publication Date
CN102500013A true CN102500013A (en) 2012-06-20

Family

ID=46212185

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2011103727423A CN102500013A (en) 2011-11-22 2011-11-22 Fully automatic intelligent infusion method and device based on model predictive control for large doses of insulin

Country Status (1)

Country Link
CN (1) CN102500013A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103418053A (en) * 2013-07-24 2013-12-04 上海中医药大学附属龙华医院 Individualized insulin treatment pump and basic infusion rate optimization method thereof
CN104667379A (en) * 2015-03-06 2015-06-03 上海交通大学 Insulin pump with dynamic closed-loop control
CN104739419A (en) * 2015-03-19 2015-07-01 深圳市一体太赫兹科技有限公司 System for regulating blood sugars

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7029443B2 (en) * 2002-10-21 2006-04-18 Pacesetter, Inc. System and method for monitoring blood glucose levels using an implantable medical device
CN101032403A (en) * 2006-03-07 2007-09-12 沈阳众泰科技发展有限公司 Tiny-wound, dynamic and continuous detecting method and system of concentration of sugar in human blood
CN101125086A (en) * 2006-08-18 2008-02-20 胜 刘 Closed-loop automatic controlling insulin-injecting system
CN201186082Y (en) * 2008-05-05 2009-01-28 王安宇 Intelligent input mechanism for insulins and glucose
CN101835505A (en) * 2007-10-02 2010-09-15 B.布朗梅尔松根公司 System and method for monitoring and regulating blood glucose levels
CN201658692U (en) * 2010-04-21 2010-12-01 佛山市利强医疗设备有限公司 Artificial pancreas system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7029443B2 (en) * 2002-10-21 2006-04-18 Pacesetter, Inc. System and method for monitoring blood glucose levels using an implantable medical device
CN101032403A (en) * 2006-03-07 2007-09-12 沈阳众泰科技发展有限公司 Tiny-wound, dynamic and continuous detecting method and system of concentration of sugar in human blood
CN101125086A (en) * 2006-08-18 2008-02-20 胜 刘 Closed-loop automatic controlling insulin-injecting system
CN101835505A (en) * 2007-10-02 2010-09-15 B.布朗梅尔松根公司 System and method for monitoring and regulating blood glucose levels
CN201186082Y (en) * 2008-05-05 2009-01-28 王安宇 Intelligent input mechanism for insulins and glucose
CN201658692U (en) * 2010-04-21 2010-12-01 佛山市利强医疗设备有限公司 Artificial pancreas system

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103418053A (en) * 2013-07-24 2013-12-04 上海中医药大学附属龙华医院 Individualized insulin treatment pump and basic infusion rate optimization method thereof
CN104667379A (en) * 2015-03-06 2015-06-03 上海交通大学 Insulin pump with dynamic closed-loop control
CN104667379B (en) * 2015-03-06 2017-08-15 上海交通大学 Dynamic closed-loop control of an insulin pump
CN104739419A (en) * 2015-03-19 2015-07-01 深圳市一体太赫兹科技有限公司 System for regulating blood sugars

Similar Documents

Publication Publication Date Title
US9782076B2 (en) Smart messages and alerts for an infusion delivery and management system
Hovorka Closed-loop insulin delivery: from bench to clinical practice
Garg et al. Glucose outcomes with the in-home use of a hybrid closed-loop insulin delivery system in adolescents and adults with type 1 diabetes
US10188794B2 (en) Closed loop control and signal attenuation detection
Breton et al. Fully integrated artificial pancreas in type 1 diabetes: modular closed-loop glucose control maintains near normoglycemia
US8219222B2 (en) Expert system for pump therapy
US20090164190A1 (en) Physiological condition simulation device and method
Eren-Oruklu et al. Estimation of future glucose concentrations with subject-specific recursive linear models
Lynch et al. Estimation-based model predictive control of blood glucose in type I diabetics: a simulation study
EP2174128B1 (en) Method and device for assessing carbohydrate-to-insulin ratio
Buckingham et al. Preventing hypoglycemia using predictive alarm algorithms and insulin pump suspension
Tassaux et al. Patient-ventilator interactions during partial ventilatory support: a preliminary study comparing the effects of adaptive support ventilation with synchronized intermittent mandatory ventilation plus inspiratory pressure support
US10194839B2 (en) Method and device for assessment of a series of glucose concentration values of a body fluid of a diabetic for adjustment of insulin dosing
US20090105636A1 (en) Closed Loop Control System With Safety Parameters And Methods
JP6017758B2 (en) Patient information input interface for the treatment system
Chase et al. Model-based glycaemic control in critical care—a review of the state of the possible
ES2670793T3 (en) Apparatus for optimizing a dosing regimen of insulin to a patient
US20030104982A1 (en) Array and method for dosing a hormone regulating blood sugar in a patient
Wolpert et al. Dietary fat acutely increases glucose concentrations and insulin requirements in patients with type 1 diabetes: implications for carbohydrate-based bolus dose calculation and intensive diabetes management
Dassau et al. Detection of a meal using continuous glucose monitoring: implications for an artificial β-cell
Lee et al. A closed-loop artificial pancreas using model predictive control and a sliding meal size estimator
US8622988B2 (en) Variable rate closed loop control and methods
Fabietti et al. Clinical validation of a new control-oriented model of insulin and glucose dynamics in subjects with type 1 diabetes
US20120150144A1 (en) Calculating insulin on board for an extended bolus being delivered by an insulin delivery device
US20140180203A1 (en) Integrated insulin delivery system having safety features to prevent hypoglycemia

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)