Search

CN-121983334-A - Insulin pump dosage decision system, method, device, medium and program product

CN121983334ACN 121983334 ACN121983334 ACN 121983334ACN-121983334-A

Abstract

The invention belongs to the technical field of medical data processing and intelligent decision support, and particularly relates to an insulin pump dosage decision system, method, equipment, medium and program product; the double-channel dose prediction unit generates initial recommended dose according to the baseline characteristic and the risk signal, the simulation deduction unit receives the dose parameter, constructs a virtual metabolism model and generates blood sugar predicted values and interval lower limits of a plurality of preset time points, the safety closed loop feedback control unit is connected with the simulation deduction unit and the double-channel dose prediction unit, corresponding feedback correction is triggered based on the risk state, accurate dose infusion is achieved, and clinical inertia and excessive treatment risks are avoided.

Inventors

  • RONG XI

Assignees

  • 广西医科大学第一附属医院

Dates

Publication Date
20260505
Application Date
20260107

Claims (10)

  1. 1. An insulin pump dosage decision system, comprising: the multi-source data processing unit is used for acquiring baseline characteristic data of a target object and carrying out risk stratification, and generating a risk suppression signal containing a nonlinear attenuation relation when the low blood sugar high risk state is identified; The dual-channel dose prediction unit comprises a large dose prediction sub-module and a basic rate prediction sub-module which are mutually independent, generates an initial pre-meal large dose and 24-hour basic total amount based on the baseline characteristic data respectively, and carries out safe attenuation on a prediction result based on the risk suppression signal to form an initial recommended dose combination; the simulation deduction unit is configured to receive the initial recommended dose combination, construct a virtual metabolism model of a target object in a calculation space, carry out prospective deduction on a plurality of preset time points in the future after administration, and output a corresponding blood sugar predicted value and a lower limit of a predicted interval thereof; And the safe closed-loop feedback control unit is in communication connection with the simulation deduction unit and the dual-channel dosage prediction unit, and executes corresponding feedback correction steps based on triggering of different risk states before generating a final infusion instruction.
  2. 2. The insulin pump dosage decision system of claim 1, wherein the feedback correction step comprises at least one of a step, b step, and c step; step a, when any measured blood glucose value of a target object is monitored to be lower than a preset extremely low threshold value, the system directly triggers a safety fusing operation; Aiming at the hidden risk that the current blood glucose value is not out of limit but the blood glucose reduction rate exceeds a preset threshold value, reversely pushing the dose to be reduced by utilizing a pharmacodynamic conversion coefficient based on a preset negative feedback correction strategy, and counteracting the overlarge reduction rate; The method comprises the steps of c, monitoring the lower limit of a prediction interval, triggering a safety convergence subroutine when the lower limit of any prediction interval is lower than a preset safety threshold, calculating the dose reduction amount based on a preset negative feedback correction strategy, updating the current recommended dose by using the reduced dose, and forcedly triggering the simulation deduction unit to execute deduction again until the generated lower limit of the prediction interval meets the safety threshold condition, and outputting the safety convergence dose.
  3. 3. The insulin pump dosage decision system according to claim 2, wherein the negative feedback correction strategy comprises at least one of an analytical method and an iterative search method, wherein the analytical method is to determine a difference between a lower limit of a prediction interval and a safety threshold, reversely map the difference into a dosage reduction amount by using a pharmacodynamic parameter or a model, and the iterative search method is to gradually down-regulate a current dosage by using a preset attenuation factor or a fixed step length if an accurate pharmacological parameter is lacking, and re-substitute the simulation deduction unit for verification after each down-regulation until the safety threshold is met.
  4. 4. The insulin pump dosage decision system of claim 1, wherein the dual channel dosage prediction unit comprises a time sequence discretization module that maps to time sequence flow rate instructions executable by the insulin pump based on a preset circadian weight template under the constraint of maintaining a sum of segment dosages equal to a time base total amount of 24 hours; Based on the constraint of an asymmetric loss function, outputting a gain coefficient monotonically decreasing along with the risk intensity when the high-risk identification is identified, and forcedly depressing the output value of the initial recommended dose; The auxiliary decision system also comprises a state resetting and rolling control unit, wherein the state resetting and rolling control unit is used for calibrating the internal state variable of the simulation deduction unit to the real state after the real blood sugar observation value is acquired, and executing rolling prediction on the updated future time window by taking the calibrated state as an anchor point.
  5. 5. An insulin pump dosage decision method, comprising the steps of: S1, acquiring baseline characteristic data of a target object; s2, respectively calculating a large dosage and a basal rate based on the baseline characteristic data; S3, distributing different weights to the basic rate based on a preset period; s4, deducing blood glucose states at a plurality of time points in the future after administration by using the model; And S5, constructing a virtual closed-loop feedback mechanism, triggering a risk state based on the blood sugar state, and carrying out feedback correction on the current risk in a calculation space.
  6. 6. The insulin pump dosage decision method of claim 5, wherein in S5, the risk status comprises at least one of a primary status, a secondary status, and a tertiary status; When any measured blood glucose value of the target object is monitored to be lower than a preset extremely low threshold value, the system directly triggers a safe fusing operation; aiming at the hidden risk that the blood sugar falling rate is not out of limit but exceeds a preset threshold value at present, triggering early-warning level dose rollback based on the falling rate when the system meets the triggering condition; When the prediction interval of any time point in the future of model deduction prediction is smaller than or equal to a safety threshold, triggering conventional prediction out-of-limit rollback; In the secondary state, the triggering condition comprises dose aggressiveness judgment, negative deviation judgment and speed abnormality judgment; The dose aggressiveness judgment is that the currently set infusion dose is higher than the originally recommended dose of the model; The negative deviation judgment is that the measured blood glucose value at the current moment is lower than the median of the model deduction prediction interval; and (3) judging the abnormal speed, namely calculating the statistical deviation Z of the current blood sugar falling track, and when Z exceeds a preset threshold value.
  7. 7. The insulin pump dosage decision method according to claim 5 or 6, wherein in the step S5, the correction strategy of the feedback correction comprises any one of an iterative search method and an analytic method, wherein the iterative search method is used for executing heuristic downregulation according to an initialized attenuation step length or an attenuation factor, bringing a downregulation result into a model again for re-simulation and repeatedly carrying out downregulation until a safety condition is met; the analysis method comprises time sequence independent attribution and PD targeting back-pushing; the timing independence is due to: full control, early risk meal reduction, late risk reduction basal rate for the baseline period of non-pumped meal volume; For the pumped preprandial actual measurement period, a basic rate single-channel compensation mode, and in view of irreversible preprandial dose, the system locks the preprandial, and converts all risk gaps into a reduction amount of the basic rate; The PD targeting reverse-pushing method comprises the steps of calculating out-of-limit amplitude Gap, directly reversely pushing the dose delta D=gap/KPD to be reduced by using PD coefficients, and executing correction, namely D new =D current -delta D.
  8. 8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 5-7 when the computer program is executed.
  9. 9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 5-7.
  10. 10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any of claims 5-7.

Description

Insulin pump dosage decision system, method, device, medium and program product Technical Field The invention belongs to the technical field of medical data processing and intelligent decision support, and particularly relates to an insulin pump dosage decision system, an insulin pump dosage decision method, insulin pump dosage decision equipment, insulin pump dosage decision medium and an insulin pump dosage decision program product. Background Continuous Subcutaneous Insulin Infusion (CSII) is the central means of enhanced glycemic control in hospitalized type 2 diabetic (T2 DM) patients, but there is still a significant "dose decision blind zone" in the initial phase of initiation of therapy (i.e. "cold start phase"). The first 24 hours of initiation of treatment is not only the "golden window" for the release of hyperglycemic toxicity, but also the high-incidence period of iatrogenic hypoglycemia. Although the current diagnosis and treatment provides an initial dose estimation framework based on body weight, due to lack of an individuation accurate algorithm based on evidence-based medicine, clinical decisions often fall into the game of clinical inertia and oversherapy, a conservative dose strategy avoids immediate risks, blood sugar standard reaching efficiency is sacrificed, and an aggressive dose strategy can rapidly reduce blood sugar, but places fragile people such as old people, renal insufficiency, low body weight and the like in low blood sugar risks. The "curative effect-safety dilemma" is further amplified due to the difference of experience of basic doctors under the background of uneven distribution of urban and rural medical resources in China, so that the safety application of the insulin pump is challenged. However, existing insulin pump related technologies have focused mainly on "run-time" real-time closed-loop control (i.e., artificial pancreas), such systems rely on real-time feedback of continuous blood glucose monitoring (CGM) data, and typically require a "warm-up period" or accumulation of sufficient blood glucose data to enter an automatic control state, and cannot effectively cover the initial "cold-start" phase of therapy initiation. At the time of treatment initiation (time T0), because of lack of patient's pump administration history data and simple static formula based on body weight neglects individual differences such as age group, past administration background, islet cell function, insulin resistance degree, etc., initial dose setting often depends on subjective experience of doctors, and lacks quantitative safety margin and prediction means. Disclosure of Invention It is an object of the present invention to overcome the above-mentioned deficiencies and to provide an insulin pump dosage decision system, method, device, medium and program product. To solve the above technical problem, an aspect of the present invention provides an insulin pump dosage decision system, including: the multi-source data processing unit is used for acquiring baseline characteristic data of a target object and carrying out risk stratification, and generating a risk suppression signal containing a nonlinear attenuation relation when the low blood sugar high risk state is identified; The dual-channel dose prediction unit comprises a large dose prediction sub-module and a basic rate prediction sub-module which are mutually independent, generates an initial pre-meal large dose and 24-hour basic total amount based on the baseline characteristic data respectively, and carries out safe attenuation on a prediction result based on the risk suppression signal to form an initial recommended dose combination; the simulation deduction unit is configured to receive the initial recommended dose combination, construct a virtual metabolism model of a target object in a calculation space, carry out prospective deduction on a plurality of preset time points in the future after administration, and output a corresponding blood sugar predicted value and a lower limit of a predicted interval thereof; the safe closed-loop feedback control unit is in communication connection with the simulation deduction unit and the dual-channel dosage prediction unit, Before the final infusion instruction is generated, a corresponding feedback correction step is performed based on the triggering of the different risk states. Further, the feedback correction step includes at least one of a step, b step, and c step; step a, when any measured blood glucose value of a target object is monitored to be lower than a preset extremely low threshold value, the system directly triggers a safety fusing operation; Aiming at the hidden risk that the current blood glucose value is not out of limit but the blood glucose reduction rate exceeds a preset threshold value, reversely pushing the dose to be reduced by utilizing a pharmacodynamic conversion coefficient based on a preset negative feedback correction strategy, and counteracting the overlarg