Intelligent anesthesiology medicine distribution system based on big data

Abstract

The invention discloses an intelligent anesthesiology medicine distribution system based on big data, and relates to the technical field of intelligent medicine distribution. Comprising a big data distribution module, an intelligent tracking processing module and a medicine positioning device. The big data distribution module generates an order configuration table based on hospital drug demands and inventory data, and dynamically balances drug supply and demand in combination with time sequence smoothing; introducing constraint parameters in distribution path planning, and generating an optimal distribution path through multi-target linear programming and an incremental cost optimization algorithm; the medicine positioning device achieves whole-process position and temperature control monitoring, and the intelligent tracking processing module analyzes path deviation and risk states according to real-time positioning information and adjusts distribution strategies in real time. According to the invention, accurate prediction, path optimization and whole-process monitoring of the anesthesiology medicine are realized, the distribution efficiency and safety are improved, the medicine expiration and delay risk is reduced, and the intelligent level of hospital medicine management is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent medicine distribution, in particular to an intelligent anesthesiology medicine distribution system based on big data. BACKGROUND

[0002] The existing hospital medicine distribution system generally has high degree of manual intervention, delayed information update, and lack of flexibility in path scheduling.

[0003] Especially in the anesthesiology department, due to the large number of drug types, high usage frequency, and strict expiration date and temperature control requirements for some drugs, the traditional manual distribution method is prone to risks such as drug shortage, delayed distribution, or abnormal temperature control.

[0004] The existing system lacks dynamic prediction and intelligent scheduling mechanism based on big data in the medicine distribution process, and cannot quickly respond to the timeliness changes of clinical demand, nor can it realize the whole-process state monitoring of medicines during transportation.

[0005] Therefore, how to use big data analysis and intelligent control technology to realize the precise allocation, path optimization and whole-process tracking of anesthesiology medicines has become a problem that needs to be solved in the intelligent management of hospital medicines. SUMMARY

[0006] The purpose of the present application is to solve the problems of demand response lag, unreasonable path planning, lack of transportation monitoring, and difficulty in predicting distribution risks in the process of anesthesiology medicine distribution in the prior art, and to propose an intelligent anesthesiology medicine distribution system based on big data.

[0007] The purpose of the present application can be achieved by the following technical solution: an intelligent anesthesiology medicine distribution system based on big data, comprising: a big data distribution module, an intelligent tracking processing module, and a medicine positioning device; The big data distribution module obtains the demand information of anesthesiology medicines through a big data platform, retrieves the inventory and demand data of corresponding medicines from the anesthesiology medicine library, and generates a corresponding order configuration table; After completing the demand information order configuration, the optimal distribution path is generated through distribution analysis; According to the demand information order configuration and the optimal distribution path, the corresponding medicine positioning device is installed in the distribution device; In the positioning monitoring process, the intelligent tracking processing module comprehensively analyzes the real-time positioning information and the preset distribution path to obtain the path deviation result of medicine distribution; According to the deviation state and the path deviation result, risk prediction analysis is performed to obtain a risk prediction result set; The big data distribution module dynamically optimizes the distribution process according to the prediction result, adjusts the distribution strategy and time allocation in real time, and calculates a comprehensive risk index, and if the comprehensive risk index is greater than a preset risk threshold, the optimization iteration is continuously executed.

[0008] As a preferred embodiment of the present application, the specific process of generating the corresponding order configuration table is: The big data distribution module obtains a demand data set of anesthesiology drugs from a hospital business system; obtains an inventory data set from an anesthesiology drug warehouse database, and one-to-one matches the demand data set and the inventory data set; In the matching process, an associated query is performed on each drug number in the demand data set to obtain corresponding demand records and inventory records. For each pair of matching records, a corrected drug demand priority sequence is calculated; According to the corrected drug demand priority sequence, the big data distribution module obtains corresponding historical drug consumption records; the historical drug consumption records are classified according to drug numbers to obtain time series data sets of each drug; time series smoothing processing is performed on each drug time series data set to obtain smoothed time series; and trend analysis is performed based on the smoothed time series to generate an order configuration table.

[0009] As a preferred embodiment of the present application, the process of calculating the corrected drug demand priority sequence is: An associated query is performed on each drug number in the demand data set to obtain corresponding demand records and inventory records. For each pair of matching records, a corresponding demand difference value is calculated; if the demand difference value Δ i is greater than 0, it is a drug to be replenished; if the demand difference value is less than or equal to 0, it is sufficient inventory; all drug items with a demand difference value greater than 0 are integrated into a demand difference matrix, and the demand difference matrix is sorted in descending order according to the numerical value of the demand difference value to obtain a drug demand priority sequence; The priority is corrected using the frequency and time period to obtain a corrected priority value; the drug number, demand difference value, and drug demand priority output are obtained, and the demand difference matrix is sorted in descending order according to the demand difference value to obtain a drug demand priority sequence.

[0010] As a preferred embodiment of the present application, the process of trend analysis based on the smoothed time series is: According to the extracted period fluctuation characteristic value, the current single demand amount, and the frequency of use, the predicted demand amount of the future period is calculated; The re-distribution quantity is calculated according to the predicted demand quantity and the current inventory quantity, if the distribution quantity is greater than the re-distribution quantity and the inventory safety lower limit, then the medicine is marked as a medicine to be distributed; and all the medicines to be distributed are counted to obtain a medicine-to-be-distributed set for order configuration, and an order configuration table is generated.

[0011] As a preferred embodiment of the present application, the specific process of generating the optimal distribution path is: According to all the medicine-to-be-distributed information in the order configuration table, and extracting a target department position set, including the spatial coordinates and the receiving time window of the corresponding department; and according to the hospital building plane path network; and performing distribution analysis to obtain a constraint optimization path set; The overall distribution sequence and vehicle scheduling are calculated for the constraint optimization path set; the path allocation rate is calculated according to the total amount of medicine and the transportation capacity; when the path allocation rate is greater than 1, the path is split into multiple sub-tasks, each sub-task corresponds to a vehicle loading unit; the path in the medicine demand priority sequence is corrected in turn, and the task allocation is performed by using the minimum incremental cost strategy according to the current empty state of the vehicle and the predicted driving time: the vehicle with the minimum incremental cost is selected to perform the path task in turn, until all the paths are allocated, and an initial global scheduling scheme and a path set are obtained; According to the initial global scheduling scheme and the path set, the path set is obtained by performing local path adjustment based on the initial scheme.

[0012] As a preferred embodiment of the present application, the process of performing distribution analysis is: The minimum path is calculated according to the shortest path between the target department of the medicine and the starting distribution point; and the shortest feasible path set is calculated and counted to obtain the shortest feasible path set, which is denoted as the primary path set; On the basis of the primary path set, a comprehensive cost function is constructed by introducing a distribution constraint parameter to output a comprehensive cost value; all path comprehensive cost values are integrated to form a cost matrix, and a multi-objective linear programming model is constructed to output a comprehensive optimization result value corresponding to each feasible path; if the comprehensive optimization result value is greater than or equal to a preset optimization result threshold, then each path is dynamically screened according to the constraint condition to obtain a pass item; the path set with the pass item and the minimum comprehensive cost value is denoted as the constraint optimization path set.

[0013] As a preferred embodiment of the present application, the process of obtaining the path deviation result of the medicine distribution is: At the beginning of the medicine distribution, the positioning data set is collected; and a preset distribution path set is obtained. The real-time coordinates are collected at a sampling period point, and the real-time path deviation value is calculated; the deviation values of all sampling points are combined to form a deviation set, and the average deviation rate is calculated; if the average deviation rate is greater than a safety threshold, the path deviation detection is triggered; In the offset detection, an offset direction vector is calculated, and then an offset duration is calculated, and a cumulative length of the offset section is counted; Based on the sampling time sequence, a real-time speed sequence of the drug delivery is calculated, and the real-time speed is compared with a theoretical speed of the path to obtain a speed deviation; The deviation rate, the offset direction, the offset duration and the speed deviation are analyzed to obtain a comprehensive deviation state value; if the comprehensive deviation state value is greater than or equal to a stable state threshold, a path offset result is generated.

[0014] As a preferred embodiment of the present application, the process of obtaining the risk prediction result set is: The comprehensive deviation state value and the path offset result set output in the previous stage are obtained; The time risk prediction is analyzed: the delivery time sequence in the sampling period is compared with a preset time window, and a time delay amount is calculated; according to the cumulative delay of the offset section and the theoretical time of the delivery path, a time risk index is calculated; the speed deviation is included in the calculation, and a corrected time risk index is obtained through a time correction factor; if the corrected time risk index is greater than a set time risk threshold, it is marked as a high time risk event; The process risk prediction is analyzed: the process abnormality probability is calculated according to the deviation state and the path deviation rate; if the process abnormality probability is greater than or equal to a preset process risk threshold, it is determined as a process abnormality risk; The comprehensive risk index is calculated according to the corrected time risk index and the process abnormality probability; finally, the risk prediction result set is obtained.

[0015] As a preferred embodiment of the present application, the process of calculating the comprehensive risk index is: For a high time risk event, the path node time allocation matrix and the target arrival time of each node are analyzed, the time offset amount is calculated; the existence of an advance or delay node is judged according to the positive or negative value of the time offset amount, and the average deviation rate is calculated; if the average deviation rate is greater than a preset deviation rate threshold, the moving speed of the delivery device is adjusted; According to the set of alternative paths available in the historical path data, the path with the smallest delay is selected to replace the current path section; For a process abnormality risk, a set of process node states is extracted, and an abnormal node proportion is calculated; if the abnormal node proportion is greater than the process risk threshold, a process adjustment strategy is triggered; The comprehensive risk index is recalculated according to the optimized corrected time risk index and the process abnormality probability; if the comprehensive risk index is less than or equal to a preset risk threshold, it is determined that the optimization is effective and enters the steady state monitoring; if the comprehensive risk index is greater than the preset risk threshold, the optimization iteration is continuously executed.

[0016] Compared with the prior art, the present application has the beneficial effects that: 1、The present application obtains the demand information of anesthesiology drugs through a big data platform, and combines a time sequence smoothing algorithm and a multi-dimensional weight prediction model to predict future drug use trends and generate an optimal order configuration table, so that dynamic balance of drug inventory and demand is achieved.

[0017] 2、The present application installs a drug positioning module in a distribution device to obtain real-time drug position information, and calculates path deviation rate, deviation direction and speed difference through an intelligent tracking processing module to dynamically analyze and predict the risk of the distribution state, so that the drug is safely delivered within the effective period and temperature control conditions.

[0018] 3、The present application introduces multiple constraint parameters such as drug expiration date, time window, temperature control condition and priority in the distribution path planning, adopts a multi-objective linear programming and incremental cost optimization algorithm to generate an optimal distribution path that meets timeliness and safety, and improves distribution efficiency and path utilization rate. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to facilitate the understanding of those skilled in the art, the present application will be further described below in conjunction with the drawings.

[0020] Figure 1 The present application is a schematic diagram of the principle. DETAILED DESCRIPTION

[0021] The technical solutions of the present application will be described below in conjunction with the embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] It should be understood that the terms "include" and "contain" used in the specification and claims of the present disclosure indicate the presence of the described features, whole, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof.

[0023] It should also be understood that the terms used in the specification of the present disclosure are only for the purpose of describing specific embodiments and are not intended to limit the present disclosure. As used in the specification and claims of the present disclosure, the singular forms "a", "an" and "the" are intended to include plural forms unless the context clearly indicates otherwise. It should be further understood that the term "and / or" used in the specification and claims of the present disclosure means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0024] Referring to Figure 1 As shown in the figure, an intelligent anesthesia department drug delivery system based on big data includes a big data delivery module, an intelligent tracking processing module, and a drug positioning device. The big data delivery module obtains demand information of anesthesia department drugs through a big data platform, and retrieves inventory and demand data of corresponding drugs from the anesthesia department drug library to generate a corresponding order configuration table. After completing the demand information order configuration, perform delivery analysis and generate the optimal delivery path based on a multi-level analysis mechanism. According to the demand information order configuration and the delivery path, install the corresponding drug positioning device in the delivery device for whole-process drug location monitoring. During the positioning monitoring process, the intelligent tracking processing module analyzes the real-time positioning information and the preset delivery path to obtain the real-time deviation state and path deviation result of drug delivery. According to the deviation state and path deviation result, further risk prediction analysis is performed, including time risk and process risk. The big data delivery module dynamically optimizes the delivery process based on the prediction result, adjusts the delivery strategy and time allocation in real time, reduces the time risk and process risk, and controls them within a reasonable range.

[0025] The specific process of generating a corresponding order configuration table is as follows: The big data delivery module first obtains the demand data set D1 = {d1, d2, …, d i , …, d n} of anesthesia department drugs from the hospital business system, where each d i contains drug number ID i , drug name N i , single demand quantity Q i , usage frequency F i , usage period T i and claim department S i information; obtain the inventory data set D2 = {s1, s2, …, s i , …, s n} from the anesthesia department drug library database, where each s i contains drug number ID i , current inventory S i , inventory safety lower limit L i , batch number B i and expiration date E i ; match the demand data set D1 and the inventory data set D2, and match the drug number ID i as the primary key, and match the demand data set D1 and the inventory data set D2 one by one. In the matching process, for each drug ID i Perform a join query to obtain the corresponding demand record d i With each pair of matching records s i , calculate the corresponding demand difference value Δ i , by calculating Δ i = Q i -S i , where Q i is the single demand amount of the drug, S i is the corresponding inventory amount; if the demand difference value Δ i > 0, it indicates that the current inventory is insufficient to meet the demand, and is marked as a drug that needs to be replenished; if the demand difference value Δ i ≤ 0, it is marked as sufficient inventory; all drug entries with demand difference value Δ i > 0 are integrated into a demand difference matrix Δ = {(ID i , Δ i )}, and are sorted in descending order according to the numerical value of the demand difference value Δ i ; According to the sorting result, generate a drug demand priority sequence P = {p1, p2, …, p i , …, p n}, where p1 corresponds to the drug with the largest demand difference; Further correct the priority using frequency F i and time period T i to obtain a corrected priority value p i ', by the formula p i ' = p i × (1 + φ1×F i + φ2×T i ), where φ1 and φ2 are the set adjustment coefficients; then obtain the drug ID i , demand difference value Δ i and drug demand priority p i ' output the corrected demand difference matrix Δ' = {(ID i , Δ i , p i ')}, and sort the difference values in the corrected demand difference matrix in descending order according to the priority, to obtain a corrected drug demand priority sequence P' = {p1', p2', …, p i ', …, p n '}.

[0026] According to the modified drug demand priority sequence P', the big data distribution module obtains the corresponding big historical drug consumption record H = {h1, h2, …, h i , …, hn}, where each h i Includes consumption timestamp τ i Quantity C consumed i and department logo S i H is assigned to the drug ID. i The data was classified to obtain the time series dataset H for each drug. i ={(τ1,C1), (τ2,C2),…,(τ k C k )};For each H i Perform time series smoothing processing, and use the moving average algorithm to calculate the smoothed series. (t)=α×C i (t)+(1-α)× (t-1), where α is the smoothing coefficient; based on the smoothed time series (t) Perform trend analysis: Extract the periodic fluctuation characteristic value θ i ,Right now The value represents the growth rate of drug demand over time. It should be noted that the smoothing coefficient α is determined based on the volatility characteristics of drug consumption data, with a value range of 0.1–0.6. Among them, 0.1–0.3 is used for drugs with relatively stable consumption to enhance the smoothing effect of historical data; 0.3–0.6 is used for drugs with rapid consumption fluctuations to improve the response speed to trend changes. In practical applications, the optimal α value is automatically determined through cross-validation.

[0027] Based on the periodic fluctuation characteristic value θ i Current single-order demand Q i and frequency of use F i To calculate the projected demand for the next period, use the formula Qf. i =α×Q i +β×F i +γ×θ i +δ×h pi , where h pi The average historical consumption of the drug is represented by α, β, γ, and δ, which are set as multidimensional weights. These weights are initially assigned by those skilled in the art based on experience and then fine-tuned through small-scale experiments. For example, if historical consumption is the most critical factor for prediction, α is initially set to a larger value (e.g., α=0.5), while other weights can be initially set to β=0.2, γ=0.2, and δ=0.1, and then adjusted according to actual delivery deviations and prediction errors. If seasonal fluctuations or future demand trends have a significant impact on a drug, the weights of β or γ can be appropriately increased. Based on the predicted demand Qf i Compared with the current inventory level S i Calculate the redistribution volume R i =Qfi -S i If the delivery volume is re-delivered, R i > Inventory safety lower limit L i The time is marked as a drug to be delivered; for all drugs to be delivered, R = {R1, R2, ..., R...} m Configure the order and generate the order configuration table O = {(ID)} i N i R i S i T i E i B i )}.

[0028] Furthermore, an order configuration table will be generated for delivery analysis, specifically as follows: According to the order configuration table O = {(ID) i N i R i S i T i E i B i Collect all drug information to be delivered from the set C = {c1, c2, ..., c}, and extract the target department location set C = {c1, c2, ..., c}. j c m}, where each c j Includes the spatial coordinates (X) of the corresponding department j Y j and receiving time window Tw j =[t 1j , t 2j Furthermore, combining this with the hospital building's planar path network G = (V, E), where V is the set of delivery nodes, E is the set of path edges, and each edge e ij Attached path length L ij With travel time T ij .

[0029] The process of conducting delivery analysis is as follows: According to the target department S of the drug i The shortest path between the originating delivery point S0 and the starting delivery point S0 is calculated using the weighted shortest path algorithm, i.e., formula D. ij =min(Σ(L) ij ×w1+T ij ×w2)) Output the minimum path D ijwherein w1, w2 are weight factors of path distance and time, respectively, which are initially assigned by the skilled person according to experience and fine-tuned through small-scale experiments; for example, for a delivery task in which distance is more critical, w1 is initially set to a larger value and w2 is smaller; if time priority is higher, the proportion of w2 can be appropriately increased; w1:w2=0.6:0.4 can be initially taken as a reference, and adjusted according to actual delivery efficiency and path deviation; the shortest feasible paths from all starting points to target nodes are then calculated, and the shortest feasible path set is obtained by statistics, denoted as primary path set P1={p 11 , p 12 , …, p 1i , …, p 1n}.

[0030] On the basis of the primary path set P1, delivery constraint parameters are introduced, including drug expiration date E i , time window Tw j , temperature control requirement Wc i and drug demand priority sequence P; and a comprehensive cost function is constructed, and the comprehensive cost value F ij is output by the formula wherein in the formula, d is a path length normalization value, representing the proportion of the distance from delivery node i to j, is a travel time normalization value, representing the proportion of the time consumed by the delivery path; α2, β2, γ2 and δ2 are adjustment weight coefficients, which are initially assigned by the skilled person according to experience and fine-tuned through small-scale experiments; for example: if the delivery task has the most stringent requirement on time, β2 is set to a larger value (such as α2:β2:γ2:δ2=0.2:0.5:0.2:0.1); if the drug expiration date is short, γ2 is appropriately increased; all path comprehensive cost values are integrated into a cost matrix F=[F ij ], and a multi-objective linear programming model is constructed: Min Z=∑(ω1×F ij +ω2×|E i -E0|+ω3×|Tw j -T j0 |) is output as the comprehensive optimization result value Min Z corresponding to each feasible path; wherein E0 is a system-set drug expiration date threshold, T j0 is a target time window center value, and ω1, ω2 and ω3 are target balance weights; if the comprehensive optimization result value Min Z is greater than or equal to a preset optimization result threshold Min Z0, then each path is dynamically screened to obtain the pass item in combination with the constraint conditions, including: each drug corresponds to only one delivery path; the path time must be within the specified window, the temperature control requirement must meet the drug standard, and the drug expiration date must be within the safe range; the simplex method and a multi-objective linear programming solver are used for iterative solution to obtain the optimal feasible solution X of each path.ij F ij The minimum path set is extracted as P2={p1', p2', …, p k '} which is the constraint optimization path set meeting the time window and temperature control conditions.

[0031] The overall distribution sequence and vehicle scheduling are calculated for the constraint optimization path set P2; the path distribution rate Rr is calculated according to the total amount of drugs R i and the transportation capacity Cap, that is, ; when the path distribution rate Rr>1, the path is split into multiple sub-tasks, each of which corresponds to a vehicle loading unit; the paths in the drug demand priority sequence P' are sequentially corrected, and the task distribution is performed using the minimum incremental cost AG strategy according to the current empty state of the vehicle and the predicted driving time.

[0032] The incremental cost AG is the incremental comprehensive cost of the vehicle after inserting the task, and the calculation of the incremental comprehensive cost AG=F ij ×Rr+λc1×Δtc ij , where λc1 is a delay penalty coefficient, and Δtc ij is the predicted delay time after inserting the task.

[0033] The vehicle with the minimum AG is sequentially selected to perform the path task until all path distribution is completed, and the initial global scheduling scheme G0 and the path set Ph0 are obtained.

[0034] The initial global scheduling scheme G0 and the path set Ph0 are further optimized, and local path adjustment is performed on the basis of the initial scheme; the local adjustment includes task exchange and path insertion operations: if it is found that there is a time intersection or cost unevenness in the scheduling of two paths, the target function variation AG'=(F new ×Rr+λ1×Δt new )-(F old ×Rr+λ1×Δt old ) is calculated after the exchange or insertion operation, and if AG' <0 and the time window Tw and temperature control Wc constraints are met, the adjustment operation is performed. As the iteration criterion, the local adjustment is continuously performed for a limited number of times until the continuous several rounds of target value decrease amplitude is lower than the threshold value ε0 or the maximum iteration number Iter max is reached; finally, the path set after iteration convergence is recorded as Ph={p1*, p2*, …, p m *}.

[0035] The corresponding drug positioning device is installed in the distribution device, and the specific process is as follows: A drug positioning device is fixed on each drug delivery unit (including drug tray, medicine box, or fixed bracket in the delivery vehicle). The positioning device can be an ultra-wideband positioning module, RFID tag, or Bluetooth low power module. It should be noted that the real-time spatial coordinate information of the drug can be continuously collected throughout the delivery process. The positioning device is connected to the internal control unit of the delivery device via wired or wireless communication. The collected location information is transmitted to the intelligent tracking and processing module in real time according to the preset sampling period Δt.

[0036] Based on a comprehensive analysis of real-time location information and preset delivery routes, the specific process is as follows: At the start of drug delivery, the drug positioning device installed in the delivery unit is calibrated to obtain real-time positioning points, and the positioning dataset L = {(x 21 y 21 , t 21 ), (x 22 y 22 , t 22 ), ..., (x 2n y 2n , t 2n )}, where x 2i y 2i These represent the planar coordinates of the drug along its delivery route, t and t. 2i This is the sampling time point; immediately afterwards, the preset delivery route set Ph is obtained, and each route p 1j It consists of several critical path nodes, denoted as p. 1j ={(X 11 Y 11 ), (X 12 Y 12 ), ..., (X 1k Y 1k )}, where X 1k Y 1k These are the theoretical coordinates of the path nodes; Using the sampling period Δt to represent the real-time coordinates (x) 2i y 2i Data is collected and real-time path deviation values ​​are calculated. , where δ 2i This represents the spatial distance difference between the current location of the drug and the theoretical path node; the deviation values ​​of all sampling points are combined into a deviation set Δ1 = {δ 21 δ 22 , …, δ 2n Next, calculate the average deviation rate. This is used to analyze the overall deviation of the drug along the pathway; If the average deviation rate ρ ≤ the safety threshold ρ0, the drug delivery is considered to be within the normal range; if the average deviation rate ρ > the safety threshold ρ0, path deviation detection is triggered. After offset detection, the offset direction vector is calculated using the formula. Output offset direction vector V 2i , used to indicate the directional trend of drug movement deviation; Further calculate the offset duration Δt ij =t j -t i And calculate the cumulative length of the offset segment. This is used to identify the severity of the offset; Following this, based on the sampled time series T = {t1, t2, ..., t...} n} Calculate the real-time velocity sequence v of drug delivery i ,Right now And compare it with the theoretical path speed v0 to obtain the speed deviation η. i ,Right now ; The deviation rate ρ and the offset direction V i Offset duration Δt i and speed deviation η i The parameters are input to the state analysis function as comprehensive judgment parameters, and the function adopts a weighted nonlinear model: Sp = α1·ρ 2 +α2·|η i |+α3·Δt i +α4·|V i The output is a comprehensive deviation state value Sp, used to calculate the stability and reliability of the delivery process. α1, α2, α3, and α4 are the weighting coefficients for position deviation, speed deviation, duration, and offset direction, respectively, and satisfy α1 + α2 + α3 + α4 = 1. If the comprehensive deviation state value Sp < the stable state threshold S0, the delivery status is considered normal. If the comprehensive deviation state value Sp ≥ the stable state threshold S0, a path offset result is generated, including the recorded offset time period, offset area, and corresponding drug ID. i .

[0037] The specific process of risk prediction and analysis is as follows: Obtain the comprehensive deviation state value Sp and path offset result set R from the previous stage output. p ={(δ1, ρ 31 η 31 , Δt 31 ), (δ 32 , ρ 32 η 32 , Δt 32 ), …, (δ 3n , ρ 3nη 3n , Δt 3n )};wherein, δ 3i ρ is the positional deviation distance. 3i η is the average deviation rate. 3i The velocity deviation ratio, Δt 3i For offset duration; Analysis of time risk prediction: The delivery time series T = {t1, t2, ..., t3} within the statistical sampling period. n The time delay ΔT = t is calculated by comparing it with the preset time window T0. n -T0; based on the cumulative delay ΣΔt of the offset segment ij Calculate the time risk index Rt based on the theoretical delivery route time Tp, i.e. ; the speed deviation η 3i Incorporating this into the calculation, through the time correction factor θ, i.e. The corrected time risk index Rt′ is obtained: Rt′=Rt×θ; if the corrected time risk index Rt′>the set time risk threshold Rt0, it is marked as a high time risk event and dynamic scheduling is triggered. Perform process risk prediction analysis: Calculate the probability of process anomalies based on the deviation status Sp and path offset rate ρ, using the formula... Output process abnormal probability Where w1, w2, and w3 are preset process risk weight coefficients, Δt a The cumulative offset time for the current stage; Pf reflects the trend of deviation spread and the frequency of anomalies during the delivery process; when the probability of process anomaly Pf ≥ the preset process risk threshold Pf0, it is judged as a process anomaly risk; Further, based on the modified time risk index Rt′ and the process anomaly probability Pf, the comprehensive risk index Rl=μ1×Rt′+μ2×Pf is calculated, where μ1 and μ2 are preset weight adjustment factors used to balance the impact of time and process; finally, the risk prediction result set R={Rt′,Pf,Rl} is output and transmitted to the big data delivery module in real time.

[0038] The specific process of dynamically optimizing the delivery process based on the prediction results is as follows: Receive the risk prediction result set R = {Rt′, Pf, Rl} output by the intelligent tracking processing module; perform threshold judgment on R; if Rt′ > Rt0 or Pf > Pf0, determine that there is an abnormal risk in the current delivery process and enter dynamic optimization. The specific process of dynamic optimization is as follows: When the corrected time risk index Rt′ exceeds the corresponding threshold, analyze the path node time allocation matrix T9={t 91 , t 92 , ..., t 9n} and the target arrival time T0 of each node = {T 01 T 02 ,…,T 0n} Calculate the time offset Δt 9i =t 9i -T 0i According to Δt 9i The positive or negative value is used to determine whether there are premature or delayed nodes, and the average offset rate τ is calculated, i.e. If the average offset rate τ > the preset offset rate threshold τ0, the path acceleration scheduling algorithm is activated by adjusting the moving speed v of the delivery device, i.e., v = v0 × (1 + κ × τ), where v0 is the normal speed and κ is the acceleration adjustment coefficient. Based on the set of feasible alternative paths Pt from historical path data, select the path p with the minimum delay. a ∈Pt replaces the current path segment; When the probability of process anomaly Pf exceeds the threshold, extract the process node state set F8 = {f}. 81 f 82 , ..., f 8n}, each f 8i Includes node types (including outbound, loading, transportation, and handover), execution time, and status identifiers; calculates the proportion of abnormal nodes λ, i.e. Where Ne is the number of nodes that detected anomalies, and No is the total number of nodes in the process; if the proportion of abnormal nodes λ > the process risk threshold λ0, then the process adjustment strategy is triggered, including: adjusting the execution order of nodes to give priority to the delivery of high-priority drugs; starting backup delivery personnel or equipment to participate in task diversion; adjusting the transfer waiting time ΔTw = Tw0 × (1 - σ × λ), where Tw0 is the original waiting time, and σ is the optimization adjustment coefficient; The comprehensive risk index Rall′ = μ1×Rt′′ + μ2×Pf′ is recalculated based on the optimized corrected time risk index Rt′′ and the process anomaly probability Pf′. If the comprehensive risk index Rall′ ≤ the preset risk threshold Rall0, the optimization is deemed effective and steady-state monitoring is initiated. If the comprehensive risk index Rall′ > the preset risk threshold Rall0, the optimization iteration is continuously performed until the risk is reduced to a reasonable range.

[0039] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A big data-based intelligent anesthesiology drug delivery system, comprising: The system comprises a big data delivery module, an intelligent tracking and processing module, and a drug positioning device; its features include: The big data delivery module obtains the demand information of anesthesiology drugs through the big data platform, retrieves the inventory and demand data of the corresponding drugs from the anesthesiology drug database, and generates the corresponding order configuration table; after completing the order configuration of demand information, it performs delivery analysis to generate the optimal delivery route; based on the order configuration of demand information and the optimal delivery route, the corresponding drug positioning device is installed in the delivery device. During the location monitoring process, the intelligent tracking processing module performs a comprehensive analysis based on real-time location information and the preset delivery route to obtain the path deviation result of drug delivery; based on the deviation status and path deviation result, risk prediction analysis is performed to obtain a risk prediction result set; The big data delivery module dynamically optimizes the delivery process based on the prediction results, adjusts the delivery strategy and time allocation in real time, and calculates a comprehensive risk index. If the comprehensive risk index is greater than the preset risk threshold, the optimization iteration will continue to be performed in a loop.

2. The intelligent anesthesiology drug delivery system based on big data according to claim 1, characterized in that, The specific process for generating the corresponding order configuration table is as follows: The big data delivery module obtains the demand dataset for anesthesia drugs from the hospital's business system; it also obtains the inventory dataset from the anesthesia drug database and performs a one-to-one matching between the demand dataset and the inventory dataset. During the matching process, an association query is performed on each drug number in the demand dataset to obtain each pair of matching records between the corresponding demand record and the inventory record, and the corrected drug demand priority sequence is calculated. Based on the revised drug demand priority sequence, the big data delivery module obtains the corresponding historical drug consumption records; classifies the historical drug consumption records by drug number to obtain time series datasets for each drug; performs time series smoothing on the time series datasets for each drug to obtain smoothed time series; and generates an order configuration table based on trend analysis of the smoothed time series.

3. The intelligent anesthesiology drug delivery system based on big data according to claim 2, characterized in that, The process of calculating the revised drug demand priority sequence is as follows: Perform an association query on each drug number in the demand dataset to obtain each pair of matching records between the corresponding demand record and the inventory record, and calculate the corresponding demand difference value. If the demand difference is greater than zero, then medication needs to be replenished. If the demand difference is less than or equal to zero, then the inventory is sufficient; integrate all drug items with a demand difference greater than zero into a demand difference matrix, and sort them in descending order according to the value of the demand difference to obtain a drug demand priority sequence. The priority value is obtained by adjusting the frequency and usage period. Obtain the drug number, demand difference value, and drug demand priority. Output a corrected demand difference matrix and sort the difference values ​​in the corrected demand difference matrix in descending order of priority to obtain the drug demand priority sequence.

4. The intelligent anesthesiology drug delivery system based on big data according to claim 3, characterized in that, The process of trend analysis based on smoothed time series is as follows: Based on the extracted periodic fluctuation characteristic values, the current single demand amount, and the usage frequency, the predicted demand amount for the next period is calculated. The redistribution quantity is calculated based on the predicted demand and the current inventory. If the redistribution quantity is greater than the inventory safety limit, the drug is marked as pending delivery. The system then counts all drugs to be delivered to obtain a set of drugs to be delivered, and then configures the orders to generate an order configuration table.

5. The intelligent anesthesiology drug delivery system based on big data according to claim 1, characterized in that, The specific process of generating the optimal delivery route is as follows: Based on all the drug information to be delivered in the order configuration table, the target department location set is extracted, including the spatial coordinates of the corresponding department and the receiving time window; then, based on the hospital building plan path network, delivery analysis is performed to obtain the constrained optimization path set; Calculate the overall delivery sequence and vehicle scheduling for the constrained optimization path set; The path allocation rate is calculated based on the total amount of drugs and the transportation capacity. When the path allocation rate is greater than 1, the path is divided into multiple sub-tasks, and each sub-task corresponds to a vehicle loading unit. The paths in the drug demand priority sequence are corrected sequentially. Based on the current empty status of the vehicles and the estimated travel time, the minimum incremental cost strategy is used for task allocation: the vehicle with the minimum incremental cost is selected to execute the path task in turn until all paths are allocated, and the initial global scheduling scheme and path set are obtained. Based on the initial global scheduling scheme and path set, local path adjustments are made to obtain the path set.

6. The intelligent anesthesiology drug delivery system based on big data according to claim 5, characterized in that, The process of conducting delivery analysis is as follows: The minimum path is calculated and output based on the shortest path between the target department of the drug and the starting delivery point; then the shortest feasible paths from all starting points to the target nodes are calculated and statistically analyzed to obtain the set of shortest feasible paths, which is denoted as the primary path set; Based on the initial path set, delivery constraint parameters are introduced to construct a comprehensive cost function and output the comprehensive cost value; The comprehensive cost of all paths is integrated to form a cost matrix, and a multi-objective linear programming model is constructed. The output is the comprehensive optimization result value corresponding to each feasible path. If the comprehensive optimization result value is greater than or equal to the preset optimization result threshold, then each path is dynamically filtered to obtain the pass items in combination with the constraint conditions; the set of pass items and the path with the smallest comprehensive cost value is denoted as the constraint optimization path set.

7. The intelligent anesthesiology drug delivery system based on big data according to claim 1, characterized in that, The process of obtaining the path offset results for drug delivery is as follows: At the start of drug delivery, a location dataset is collected; a set of preset delivery routes is obtained. Real-time coordinates are collected at sampling periodic points, and real-time path deviation values ​​are calculated; the deviation values ​​of all sampling points are combined into a deviation set, and the average deviation rate is calculated; if the average deviation rate is greater than the safety threshold, path offset detection is triggered. In offset detection, the offset direction vector and offset duration are calculated, and the cumulative length of the offset segment is counted. The real-time speed sequence of drug delivery is calculated based on the sampling time series, and the speed deviation is obtained by comparing the real-time speed with the theoretical speed of the path. The deviation rate, offset direction, offset duration, and velocity deviation are analyzed to obtain a comprehensive deviation state value; if the comprehensive deviation state value is greater than or equal to the stable state threshold, the path offset result is generated.

8. The intelligent anesthesiology drug delivery system based on big data according to claim 1, characterized in that, The process of obtaining the risk prediction result set is as follows: Obtain the comprehensive deviation status value and path offset result set output from the previous stage; Time risk prediction analysis: The delivery time series within the statistical sampling period is compared with the preset time window to calculate the time delay; the time risk index is calculated based on the cumulative delay of the offset segment and the theoretical delivery route time. By incorporating speed deviation into the calculation, a corrected time risk index is obtained through a time correction factor. If the adjusted time risk index is greater than the set time risk threshold, it is marked as a high time risk event; Analysis of process risk prediction: The probability of process anomalies is calculated based on the deviation status and path offset rate; If the probability of process abnormality is greater than or equal to the preset process risk threshold, it is judged as a process abnormality risk; The comprehensive risk index is calculated based on the modified time risk index and the probability of process anomalies. Finally, statistical analysis is performed to obtain a set of risk prediction results.

9. The intelligent anesthesiology drug delivery system based on big data according to claim 1, characterized in that, The process of calculating the comprehensive risk index is as follows: For high-time-risk events, the time allocation matrix of path nodes and the target arrival time of each node are analyzed to calculate the time offset; the positive or negative value of the time offset is used to determine whether there are nodes that are ahead of or behind, and the average offset rate is calculated; if the average offset rate is greater than the preset offset rate threshold, the moving speed of the delivery device is adjusted. Based on the set of feasible alternative paths in historical path data, select the path with the least delay to replace the current path segment; When there is a risk of process abnormality, extract the process node status set and calculate the proportion of abnormal nodes; If the proportion of abnormal nodes exceeds the process risk threshold, a process adjustment strategy will be triggered. The comprehensive risk index is recalculated based on the optimized corrected time risk index and the probability of process anomalies. If the comprehensive risk index is less than or equal to the preset risk threshold, the optimization is deemed effective and the system enters steady-state monitoring. If the overall risk index is greater than the preset risk threshold, the optimization iteration will continue to be performed in a loop.

Images