Method and system for dynamic prediction of infusion bottle residual volume based on drop frequency decay gradient

Abstract

The application discloses a dynamic residual volume prediction method and system based on drip frequency attenuation gradient, and relates to the technical field of infusion monitoring. The method comprises the following steps: acquiring a time stamp sequence of a drip event of a vented gravity infusion, an initial effective volume, bottle geometry parameters, an effective single-drip volume, a pressure-flow index and a hydraulic baseline; determining accumulated drop count and local drop frequency according to the time stamp sequence, and dividing a resistance stable interval according to a drop frequency mutation position; determining a drop frequency attenuation gradient of a local drop frequency logarithm relative to accumulated drop count in each interval; simultaneously considering a pressure-flow relationship and a volume conservation relationship, eliminating an equivalent conduction coefficient of a liquid path, establishing a residual volume constraint relationship between the drop frequency attenuation gradient and corresponding liquid level parameters of the residual volume; and combining the bottle geometry parameters to solve a candidate residual volume, and determining a current residual volume according to the constraint that the residual volume decreases or remains unchanged with the increase of accumulated drop count. Therefore, the residual volume of the infusion bottle is predicted, and the prediction stability under the liquid path state change is improved.

Description

Technical Field

[0001] This application relates to the field of infusion monitoring technology, and in particular to a method and system for dynamic prediction of the remaining volume of an infusion bottle based on the drip frequency attenuation gradient. Background Technology

[0002] Gravity-assisted infusion relies primarily on the hydrostatic pressure difference between the fluid level in the infusion container and the patient's end to drive fluid flow. Healthcare staff typically adjust the infusion flow rate by observing the drip rate in the drip chamber, and assess the remaining fluid volume by considering the fluid level in the container, infusion duration, and manual monitoring. In scenarios with a large number of beds, concentrated infusion tasks, or frequent adjustments to the infusion protocol, relying solely on manual judgment can increase the workload for monitoring staff and may lead to errors in judging the remaining volume, premature or delayed bottle changes, and failure to promptly address issues after the infusion is completed.

[0003] Currently, equipment for automatic monitoring of remaining infusion volume typically uses methods such as weighing, pressure, level, capacitance, or ultrasound to obtain observational data related to the remaining fluid volume. Such equipment generally requires the addition of detection components near the infusion stand, infusion container, or infusion tubing, and involves installation, disinfection, calibration, maintenance, and consumable adaptation. Factors such as infusion bottle swaying, tubing tension, and changes in clamping position can also cause fluctuations in the detection signal. For medical facilities that have already deployed drip rate monitoring terminals, adding a separate remaining volume detection component would increase equipment modification costs and nursing procedures.

[0004] Besides the detection methods mentioned above, some devices estimate the remaining fluid volume based on cumulative drops or current drip frequency. The conversion of cumulative drops usually assumes that the drop factor and single-drop volume remain stable, but the actual single-drop volume may vary depending on the drip tip's wetting state, the fluid properties, and the infusion flow rate, easily leading to volume conversion errors over long periods. The current drip frequency is affected not only by the fluid level in the infusion container but also by the roller clamp opening, the infusion container's suspension height, the patient's end pressure, and the air intake status. Therefore, neither cumulative drops nor current drip frequency can consistently and accurately reflect the actual remaining fluid volume in the infusion container under different infusion conditions. Summary of the Invention

[0005] This application provides a method, system, storage medium, computer program product, and electronic device for dynamic prediction of the remaining volume of an infusion bottle based on the drip frequency attenuation gradient, in order to at least solve the problem of the difficulty in continuously and accurately monitoring the remaining volume of an infusion bottle in the prior art.

[0006] In a first aspect, embodiments of this application provide a method for dynamically predicting the remaining volume of an infusion bottle based on a drip frequency decay gradient. The method includes: acquiring a drip event timestamp sequence, initial effective volume, bottle geometric parameters, effective single-drop volume, pressure-flow index, and hydraulic baseline during a ventilated gravity infusion process; the bottle geometric parameters characterize the correspondence between the remaining volume and the liquid level height and cross-sectional area of ​​the liquid level; the hydraulic baseline characterizes the equivalent hydraulic height from the infusion bottle's outlet reference surface to the patient's end; determining the cumulative number of drops and the local drip frequency corresponding to each cumulative drop number based on the drip event timestamp sequence, and determining a resistance stabilization interval based on the abrupt change position of the local drip frequency; the resistance stabilization interval is a range of consecutive drops corresponding to the same value of the equivalent conductance coefficient of the fluid path, where the equivalent conductance coefficient characterizes the equivalent conductance capability of the infusion fluid path for liquid flow; within each resistance stabilization interval, the negative change of the logarithm of the local drip frequency relative to the cumulative number of drops is... The rate is determined as the drip frequency attenuation gradient corresponding to each cumulative drop number; the pressure-flow relationship and volume conservation relationship of gravity infusion are combined to eliminate the equivalent conductivity coefficient of the fluid path and establish a margin constraint relationship; the margin constraint relationship is used to constrain the correspondence between the drip frequency attenuation gradient and the liquid level height and liquid level cross-sectional area corresponding to the remaining volume under the conditions of the effective single drop volume, the pressure-flow index and the hydraulic baseline; according to the drip frequency attenuation gradient corresponding to each cumulative drop number and the bottle geometric parameters, the candidate margin corresponding to each cumulative drop number and satisfying the margin constraint relationship is solved within the range of zero to the initial effective volume; a target margin is determined from the candidate margins corresponding to each cumulative drop number, such that each target margin decreases or remains unchanged as the cumulative drop number increases, the target margin corresponding to the current cumulative drop number is determined as the current margin, and the current margin is used as the dynamic prediction result of the infusion bottle margin.

[0007] Secondly, embodiments of this application provide a dynamic prediction system for the remaining volume of an infusion bottle based on a drip frequency attenuation gradient. The system includes: a basic data acquisition unit, used to acquire the drip event timestamp sequence, initial effective volume, bottle geometric parameters, effective single-drop volume, pressure-flow index, and hydraulic baseline during a ventilated gravity infusion process; the bottle geometric parameters characterize the correspondence between the remaining volume and the liquid level height and liquid level cross-sectional area; the hydraulic baseline characterizes the equivalent hydraulic height from the infusion bottle outlet reference surface to the patient end; a stable interval determination unit, used to determine the cumulative number of drops and the local drip frequency corresponding to each cumulative drop based on the drip event timestamp sequence, and to determine the resistance stable interval based on the abrupt change position of the local drip frequency; the resistance stable interval is the interval of consecutive drops corresponding to the same value of the equivalent conduction coefficient of the fluid path, the equivalent conduction coefficient of the fluid path characterizing the equivalent conduction capability of the infusion fluid path for liquid flow; and an attenuation gradient determination unit, used to determine the negative change of the logarithm of the local drip frequency relative to the cumulative number of drops within each resistance stable interval. The rate is determined as the drip frequency attenuation gradient corresponding to each of the cumulative drops; the margin constraint establishment unit is used to combine the pressure-flow relationship and volume conservation relationship of gravity infusion, eliminate the equivalent conduction coefficient of the fluid path, and establish the margin constraint relationship; the margin constraint relationship is used to constrain the correspondence between the drip frequency attenuation gradient and the liquid level height and liquid level cross-sectional area corresponding to the remaining volume under the conditions of the effective single drop volume, the pressure-flow index, and the hydraulic baseline; the candidate margin solution unit is used to solve the candidate margin corresponding to each of the cumulative drops and satisfying the margin constraint relationship in the range from zero to the initial effective volume according to the drip frequency attenuation gradient corresponding to each of the cumulative drops and the bottle geometric parameters; the target margin determination unit is used to determine a target margin from the candidate margins corresponding to each of the cumulative drops, so that each target margin decreases or remains unchanged as the cumulative drops increase, the target margin corresponding to the current cumulative drops is determined as the current margin, and the current margin is used as the dynamic prediction result of the infusion bottle margin.

[0008] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the dynamic prediction method for the remaining volume of an infusion bottle based on the drip frequency attenuation gradient according to any embodiment of the present application.

[0009] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the dynamic prediction method for the remaining volume of an infusion bottle based on the drip frequency attenuation gradient according to any embodiment of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the dynamic prediction method for the remaining volume of an infusion bottle based on the drip frequency attenuation gradient according to any embodiment of this application.

[0011] The method and system for dynamically predicting the remaining volume of an infusion bottle based on the drip frequency attenuation gradient provided in this application can achieve at least the following technical effects:

[0012] By extracting the attenuation characteristics of local drip frequency as a function of cumulative drops within a stable interval of equivalent fluid flow, and characterizing the drip frequency attenuation gradient as the negative rate of change of the logarithm of the local drip frequency relative to the cumulative drops, the information used for margin determination is transformed from the drip frequency value at a single moment to the variation law of drip frequency as fluid volume is continuously consumed during infusion. Based on this, the pressure-flow relationship is combined with the volume conservation relationship, and the equivalent fluid flow coefficient is eliminated during the coupling process, thus establishing a direct margin constraint relationship between the drip frequency attenuation gradient and the fluid level height and cross-sectional area. Therefore, margin inference can be performed using the drip frequency evolution characteristics without obtaining the specific value of the equivalent fluid flow coefficient, and the influence of different stable fluid flow flow states on the margin calculation results is reduced.

[0013] By combining the aforementioned residual constraint relationship with the bottle's geometric parameters, the hydraulic pressure change reflected by the drip frequency decay gradient can be converted into candidate residuals according to the actual geometric correspondence between the remaining volume, liquid level height, and liquid surface cross-sectional area. This allows the residual inversion to adapt to changes in the cross-sectional area at different liquid level positions on the infusion bottle. For cases where the same drip frequency decay gradient may correspond to multiple candidate residuals, further constraints can be applied using the physical law that the remaining liquid volume decreases or remains constant with the cumulative number of drops during infusion. This eliminates candidate branches that do not conform to the direction of continuous liquid consumption, ensuring that the residual results during continuous infusion maintain a reasonable evolution direction and improving the stability and continuity of dynamic prediction results.

[0014] This technical solution transforms continuously acquired drip event information during infusion into a drip frequency attenuation gradient that reflects the liquid level descent pattern. By coupling the pressure-flow rate relationship and the volume conservation relationship, it eliminates the intermediate influencing factor of the equivalent conductivity of the fluid path. Furthermore, it leverages the bottle's geometric relationship to invert the remaining liquid volume. Thus, the remaining volume in the infusion bottle can be characterized by the dynamic evolution of the drip frequency, rather than simply by the absolute drip frequency or cumulative drip count. This maintains consistency with the physical mechanism of gravity-based infusion while enabling continuous dynamic prediction of the remaining liquid volume in the infusion bottle. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating an example of a dynamic prediction method for the remaining volume of an infusion bottle based on a drip frequency decay gradient according to an embodiment of this application is shown.

[0017] Figure 2 A flowchart illustrating an example of the operation of constructing a margin constraint relationship based on the drop frequency attenuation gradient in the method according to an embodiment of this application is shown.

[0018] Figure 3 A flowchart illustrating an example of determining the target residual under continuous solution of branch constraints in the method according to an embodiment of this application is shown.

[0019] Figure 4 A schematic diagram illustrating the operational mechanism of an example of a dynamic prediction method for the remaining volume of an infusion bottle based on the drip frequency attenuation gradient according to an embodiment of this application is shown.

[0020] Figure 5 A schematic diagram of the comparative experimental simulation results of different methods under combined disturbance conditions is shown.

[0021] Figure 6 A comparative experimental simulation diagram illustrating the performance of different methods in dynamically predicting the remaining volume of infusion bottles under multiple disturbance scenarios is shown.

[0022] Figure 7 A comparative simulation diagram illustrating an example of the dynamic prediction performance of infusion bottle remaining volume under different technology processing configurations is shown.

[0023] Figure 8 A structural block diagram of an example of a dynamic prediction system for the remaining volume of an infusion bottle based on a drip frequency decay gradient, according to an embodiment of this application, is shown. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] It should be noted that some IV (Intravenous Infusion) monitoring systems currently use photoelectric detection to collect the light-blocking signal formed when droplets pass through the drip chamber, and then calculate the cumulative drop count and the number of drops per unit time based on the detection results. Taking the DripOMeter drip rate monitoring device as an example, it uses a light source and photodetector to detect droplet events, which can be used to track the drip rate, identify drip stoppage states, and generate corresponding alarms. Such systems can continuously obtain drop event data during the infusion process, and their output is mainly used for drop count statistics and flow rate monitoring.

[0026] In addition to photoelectric detection, some systems employ CV (Computer Vision) technology to acquire images of the drip chamber and identify droplet events based on image changes during droplet formation and descent. Some visual drip rate monitoring systems also introduce DL (Deep Learning) models to identify different droplet states and perform real-time drop count statistics. Visual detection can reduce direct contact between the detection component and the infusion path, but its recognition results still mainly represent discrete droplet events or the current drop frequency. The recognition effect can also be affected by ambient light, image background, drip chamber clamping position, and changes in droplet morphology.

[0027] Currently, related systems also employ an IoT (Internet of Things) architecture to remotely monitor infusion status. For example, some IoT infusion monitoring systems install optical drop detection units and capacitive level detection units at the drip chamber. The optical drop detection unit acquires drop rate data, and the capacitive level detection unit acquires level data, then transmits the data from different detection channels to a remote terminal. Other remote infusion monitoring systems are equipped with both optical sensors and load sensors to detect the infusion drip rate and the weight of the infusion container, respectively. These systems typically rely on independent physical detection channels to obtain drip rate and remaining volume information, with the drip event data itself primarily serving the functions of infusion counting and flow monitoring.

[0028] During drop event acquisition, changes in ambient light, obstruction of the detection optical path, missed drop detection, duplicate counting, and air bubbles passing through the drip chamber can create abnormal pulses, abnormal drop intervals, or short-term drop stoppages. Drop event monitoring systems typically employ threshold judgment, signal smoothing, or outlier removal to reduce the impact of these factors on drop count statistics. Since the processing primarily serves drop identification and drop rate calculation, changes in infusion conditions or detection status can still lead to count deviations or drop frequency fluctuations in the processed drop event data, further affecting infusion status estimations based on drop count and drop frequency.

[0029] It should be understood that the above description of the relevant technologies is intended only to help the public better understand the inventive spirit and motivation of this application, and is not intended to limit this application. Furthermore, the technical solutions described in the above-mentioned relevant technologies are not prior art, and may also be undisclosed technical solutions, such as those under research or in the laboratory stage.

[0030] Figure 1 A flowchart illustrating an example of a dynamic prediction method for the remaining volume of an infusion bottle based on a drip frequency decay gradient according to an embodiment of this application is shown.

[0031] The dynamic prediction method for the remaining volume of infusion bottles provided in this application embodiment can be executed by a bedside terminal, a ward infusion monitoring gateway, a server, or other electronic devices. The electronic device executes program instructions stored in its memory via a processor and communicates with the drip monitoring terminal to receive drip event timestamps and perform dynamic prediction of the remaining volume of the infusion bottle.

[0032] The method described in this application can be applied to rigid or semi-rigid infusion bottles with ventilation conditions and a recognizable free liquid surface during infusion, as well as non-pump-controlled infusion processes where liquid flow is driven by gravity. The drop monitoring terminal can be a photoelectric drop rate monitoring terminal or an image recognition-based drop count monitoring terminal. The on-site dynamic observations required for remaining volume prediction are provided by the drop monitoring terminal, thereby enabling the determination of the remaining volume in the infusion bottle while reusing existing drop monitoring terminals.

[0033] like Figure 1 As shown, in step S110, the drop event timestamp sequence, initial effective volume, bottle geometric parameters, effective single drop volume, pressure-flow index, and hydraulic baseline are obtained during the ventilated gravity infusion process; the bottle geometric parameters represent the correspondence between the remaining volume and the liquid level height and the liquid level cross-sectional area, and the hydraulic baseline represents the equivalent hydraulic height from the infusion bottle outlet reference surface to the patient end.

[0034] The drop event timestamp sequence is generated by the drop monitoring terminal when the drop passes the detection position of the drip chamber. According to the chronological order of the drop events, the drop event timestamp sequence can be represented as multiple sequentially arranged time points. Different drop monitoring terminals can use their own internal clocks to record drop events. As long as the drop event timestamps in the same infusion process use a unified timing standard and the time differences between adjacent timestamps can be compared, it can be used for local drop frequency calculation.

[0035] The initial effective volume is the volume of fluid that can be infused at the start of the infusion. This initial effective volume can be obtained from medical orders, infusion bottle labels, bottle label scanning results, or the configuration data of the infusion management terminal. When there is medication added, withdrawn, pre-filled, or dead space fluid that cannot be drained from the infusion bottle, the nominal capacity of the infusion bottle can be corrected based on the corresponding business records, and the corrected result will be used as the initial effective volume.

[0036] The bottle geometry parameters can be retrieved from a preset bottle geometry library based on the bottle type identifier. This library records the internal contours of different bottle types and converts these contours into a correspondence between remaining volume, liquid level height, and liquid surface cross-sectional area. For any given remaining volume, the corresponding liquid level height and liquid surface cross-sectional area can be determined from the bottle geometry parameters.

[0037] In this embodiment, the remaining volume is denoted as The height of the liquid level corresponding to the remaining volume is denoted as The cross-sectional area of ​​the bottle at the location of the liquid surface is denoted as . The liquid level is determined from the outlet reference plane of the infusion bottle. Within the effective volume range, the cross-sectional area of ​​the liquid level is... For a positive value, the remaining volume and the liquid level height satisfy the following geometric relationship:

[0038] Equation (1)

[0039] In the formula, From zero to the remaining volume The volume integral variable. Correspondingly, the rate of change of the liquid level height relative to the remaining volume is the reciprocal of the liquid surface cross-sectional area. The bottle's geometric parameters can be stored using a discrete data table, a piecewise function, or a continuous fitting function. When using a discrete data table, interpolation can be performed between adjacent geometric data nodes to obtain the liquid level height and liquid surface cross-sectional area corresponding to any candidate remaining volume.

[0040] Effective single-drop volume represents the volume of liquid corresponding to one effective droplet under the current infusion set and droplet formation state. The pressure-flow index characterizes the response of the fluid volumetric flow rate in the infusion circuit to changes in the total equivalent static head. The hydraulic baseline represents the equivalent hydraulic height from the infusion bottle outlet reference plane to the patient end, which comprehensively reflects the influence of the infusion bottle suspension position, patient infusion position, patient end pressure, and fixed tubing height on the effective pressure difference.

[0041] Effective single-drop volume, pressure-flow index, and hydraulic baseline can be pre-obtained from infusion set specifications, liquid type, infusion bottle installation information, and historical calibration data, or determined from effective observation data during infusion. Before calibration, initial parameters corresponding to the infusion set specifications can be used for margin calculation. When updated parameters are subsequently obtained, these updated parameters are used in the margin calculation after the corresponding time point. The specific self-calibration method for the parameters can be determined using geometric change information in the non-uniform cross-sectional area of ​​the infusion bottle; this self-calibration process will be described in detail in subsequent embodiments.

[0042] In step S120, the cumulative number of drops and the local drop frequency corresponding to each cumulative number of drops are determined according to the drop event timestamp sequence, and the resistance stability interval is determined according to the abrupt change position of the local drop frequency; the resistance stability interval is the interval of consecutive drop numbers corresponding to the same value of the equivalent conduction coefficient of the liquid path, and the equivalent conduction coefficient of the liquid path characterizes the equivalent conduction capability of the infusion liquid path for liquid flow.

[0043] The cumulative drop count increases sequentially according to the order in which drop events occur. Each valid drop event increases the cumulative drop count accordingly. The difference in timestamps between adjacent drop events constitutes the drop interval between adjacent droplets, and the reciprocal of the drop interval can be used as the instantaneous drop frequency at the corresponding position. Considering that a single drop interval may be affected by timestamp accuracy or droplet formation fluctuations, this embodiment can determine the local drop frequency based on multiple drop intervals within a local drop count window centered on the current cumulative drop count. For example, the local drop frequency corresponding to the local drop count window can be formed by dividing the valid drop count within the local drop count window by the time difference between the first and last valid drop events in that window.

[0044] Using the cumulative drop count as data coordinates allows changes in local drop frequency to be mapped to the drop-by-drop consumption process of the fluid in the infusion bottle. These data coordinates are determined by the order in which drop events occur. Even when the infusion flow rate changes, the change in drop frequency caused by each additional effective drop can still be compared.

[0045] The equivalent fluid flow rate coefficient is used to comprehensively characterize the influence of roller clamp opening, infusion tubing, filter components, catheter, and local bends on fluid flow. Within a continuous drop range where the roller clamp position and infusion tubing condition remain constant, the equivalent fluid flow rate coefficient takes the same value and participates in the pressure-flow relationship. When the nurse readjusts the roller clamp, the infusion tubing bends change, or the catheter resistance undergoes a step change, the local drip frequency can abruptly change between adjacent drop window values.

[0046] The electronic device statistically analyzes the local drip frequency before and after each location to be determined, and judges whether a drip frequency abrupt change occurs based on the change in the local drip frequency before and after. Using these abrupt change locations as interval boundaries, the infusion process can be divided into one or more resistance stability intervals. The local drip frequency within each resistance stability interval can continuously change as the liquid level in the infusion bottle decreases, and each resistance stability interval uses its corresponding equivalent fluid path conductivity coefficient. When dividing a new resistance stability interval, the previously determined cumulative drip count and remaining volume status are retained. When the drip event time interval meets a preset abnormal pause condition, the corresponding time range can be excluded from the data used for gradient calculation within the resistance stability interval.

[0047] In step S130, within each of the resistance stabilization intervals, the negative rate of change of the logarithm of the local drop frequency relative to the cumulative drop number is determined as the drop frequency attenuation gradient corresponding to each of the cumulative drop numbers.

[0048] In this embodiment, the local drop frequency is denoted as The cumulative number of drops is recorded as Let the drop frequency attenuation gradient be denoted as The drop frequency decay gradient satisfies:

[0049] Equation (2)

[0050] The drip frequency decay gradient represents the degree to which the logarithmic value of the local drip frequency decreases when the cumulative number of drops increases by one unit. Using a negative rate of change, the drip frequency decay gradient takes a non-negative value when the liquid level in the infusion bottle drops, causing the local drip frequency to gradually decrease.

[0051] In actual calculations, the local drop frequencies within each resistance stability interval can first be converted into a logarithmic drop frequency sequence. Then, the logarithmic drop frequency sequence is locally smoothed or locally regressed, and the drop frequency attenuation gradient is determined based on the change in the cumulative drop number direction according to the smoothing result. The smoothing operation uses a single resistance stability interval as the data range. The drop frequency attenuation gradient corresponding to the current resistance stability interval is determined only by the local drop frequencies within that interval, so that the frequency step at the interval boundary and the continuous attenuation within the interval are represented separately.

[0052] When the local drop frequency is represented by a time-continuous function, the drop frequency attenuation gradient can also be written as:

[0053] Equation (3)

[0054] In the formula, Let be the infusion time. This expression normalizes the rate of change of drip frequency with respect to time by the square of the current drip frequency, thus obtaining a drop-by-drop decay characterization. In practice, it can be implemented by directly differentiating in the cumulative drop count domain, or by first determining the rate of change of local drip frequency with respect to time and then converting it into a drip frequency decay gradient.

[0055] In step S140, the pressure-flow rate relationship and volume conservation relationship of gravity infusion are combined to eliminate the equivalent conductivity coefficient of the fluid path and establish a margin constraint relationship. The margin constraint relationship is used to constrain the correspondence between the drop frequency attenuation gradient and the liquid level height and liquid surface cross-sectional area corresponding to the remaining volume under the conditions that the effective single drop volume, the pressure-flow rate index and the hydraulic baseline are determined.

[0056] The sum of the hydraulic baseline and the liquid level in the bottle is defined as the total equivalent static head. The total equivalent static head is denoted as... Then we have:

[0057] Equation (4)

[0058] In the formula, For hydraulic baseline. Total equivalent static head is used when performing infusion flow rate calculation. It is a positive value.

[0059] Within any resistance stability range, the liquid volumetric flow rate is denoted as... The equivalent conductivity coefficient of the liquid circuit is denoted as The pressure-flow index is denoted as The pressure-flow relationship can be expressed as:

[0060] Equation (5)

[0061] The local drop frequency is determined by the ratio of the liquid volumetric flow rate to the effective single drop volume. The effective single drop volume is denoted as... Then we have:

[0062] Equation (6)

[0063] As droplets continue to be discharged, the remaining volume decreases over time. According to the law of conservation of volume, the rate of change of the remaining volume with respect to time is equal to the negative of the liquid volumetric flow rate, and also equal to the negative of the product of the local drop frequency and the effective single drop volume.

[0064] Equation (7)

[0065] Since the rate of change of the cumulative drop count with respect to time is the local drop frequency, the rate of change of the remaining volume with respect to the cumulative drop count is:

[0066] Equation (8)

[0067] Based on the bottle's geometry, the rate of change of liquid level height relative to the remaining volume is the reciprocal of the liquid surface cross-sectional area. Substituting the pressure-flow rate relationship and the volume conservation relationship into the definition of the drip frequency decay gradient, we can obtain:

[0068] Equation (9)

[0069] After transforming the above relationships, we obtain the residual constraint relationships used for residual inversion:

[0070] Equation (10)

[0071] In the aforementioned margin constraint relationship, the equivalent conductance coefficient of the liquid path corresponding to the current resistance stability interval has been eliminated. The same liquid volume flow rate determines both the droplet generation rate and the rate at which the liquid level in the bottle decreases with the cumulative number of drops; after taking the negative rate of change of the logarithm of the local drop frequency relative to the cumulative number of drops, the equivalent conductance coefficient of the liquid path is eliminated as a multiplicative parameter.

[0072] The margin constraint relationship preserves the bottle geometry, effective single-drop volume, pressure-flow index, and hydraulic baseline. The bottle geometry determines the liquid surface cross-sectional area and liquid level height corresponding to the candidate remaining volume, while the hydraulic baseline and liquid level height together determine the total equivalent static head. Therefore, the remaining volume corresponding to the same drop frequency decay gradient needs to be solved in conjunction with the specific bottle type and corresponding infusion parameters.

[0073] In step S150, based on the drop frequency attenuation gradient corresponding to each of the cumulative drop counts and the bottle geometry parameters, candidate surpluses corresponding to each of the cumulative drop counts and satisfying the surplus constraint relationship are solved within the range from zero to the initial effective volume.

[0074] For the drip frequency attenuation gradient corresponding to the current cumulative number of drops, the electronic device sets multiple margins to be determined within the range of zero to the initial effective volume. For each margin to be determined, the corresponding liquid surface cross-sectional area and liquid surface height are obtained according to the bottle geometry parameters. The liquid surface height is added to the hydraulic baseline to obtain the total equivalent static head. Then, the drip frequency attenuation gradient, liquid surface cross-sectional area, and total equivalent static head are substituted into the margin constraint relationship.

[0075] Electronic devices can determine whether the remaining margin meets the constraint based on the difference between the left and right sides of the remaining margin constraint relationship. This difference is denoted as the remaining margin constraint residual, which can then be expressed as:

[0076] Equation (11)

[0077] In the formula, margin to be determined The corresponding residual constraint. When the residual constraint is equal to zero, or its absolute value does not exceed the residual tolerance determined by the accuracy of the drip frequency calculation, the accuracy of the bottle geometry data, and the accuracy of the parameters, the corresponding residual to be judged will be determined as a candidate residual.

[0078] Candidate surpluses can be obtained through numerical search. For example, by traversing the bottle geometry data from zero to the initial effective volume according to the remaining volume, the location where the sign of the surplus constraint residual changes or the absolute value reaches a local minimum can be determined, and then interpolation or iterative solution can be performed within the corresponding volume range. For bottle geometry parameters that indicate that the liquid surface cross-sectional area takes the same value within the corresponding remaining volume range, candidate surpluses can also be determined based on the surplus constraint relationship corresponding to the fixed liquid surface cross-sectional area.

[0079] The cross-sectional area of ​​the liquid surface in the constricted region at the shoulder or bottom of the infusion bottle varies with the remaining volume, and the resulting geometric constraint function may obtain the same constraint value at different volume locations. Therefore, a drip frequency decay gradient may correspond to one or more candidate reserves within the range of zero to the initial effective volume.

[0080] In step S160, a target balance is determined from the candidate balances corresponding to each cumulative drop count, such that each target balance decreases or remains unchanged as the cumulative drop count increases, the target balance corresponding to the current cumulative drop count is determined as the current balance, and the current balance is used as the dynamic prediction result of the infusion bottle balance.

[0081] In gravity-assisted infusion processes where only liquid output occurs, the remaining volume in the infusion bottle decreases or remains constant as the cumulative drop count increases. Therefore, the electronic device processes the candidate remaining volume corresponding to each cumulative drop count in the order of occurrence of the cumulative drop count, and uses the previously determined target remaining volume as the continuity benchmark for the current candidate remaining volume.

[0082] When the current cumulative drop count corresponds to only one candidate surplus, and that candidate surplus is not greater than the previous target surplus, the candidate surplus can be determined as the current target surplus. When the current cumulative drop count corresponds to multiple candidate surpluses, the candidate surplus that is not greater than the previous target surplus is taken as the candidate to be selected, and from the candidate to be selected, the candidate surplus that is located in the same continuous solution branch as the previous target surplus is determined.

[0083] The same continuous solution branch refers to a situation where, as the cumulative number of drops increases, the continuously selected candidate quantities are connected along the same volume change path in the bottle geometry, resulting in a decrease or no change in the remaining volume. The electronic device can generate a projected volume reduction based on the drop increment between adjacent cumulative drops and the effective single-drop volume, and determine the candidate quantity corresponding to the projected volume reduction from the candidate quantities on the same continuous solution branch.

[0084] The target remaining amount corresponding to each cumulative drop count is arranged in order of cumulative drop count to form the trajectory of the remaining amount of the infusion bottle. The target remaining amount at the end of the trajectory corresponds to the current cumulative drop count. The electronic device determines the target remaining amount at the end as the current remaining amount and outputs the current remaining amount as the dynamic prediction result of the remaining amount of the infusion bottle.

[0085] When the number of candidate remaining quantities satisfying the non-increasing remaining quantity relation and the continuous solution branching condition is zero, the electronic device temporarily suspends updating the dynamic prediction result of the infusion bottle remaining quantity using the current drip frequency decay gradient, and retains the most recent target remaining quantity that satisfies the remaining quantity constraint relation and monotonic change condition. The electronic device continues to receive new drip event timestamps and re-solves for candidate remaining quantities after forming a new effective drip frequency decay gradient. This limits the branching jumps in the remaining quantity change trajectory caused by a single incompatible observation.

[0086] In this embodiment, the electronic device uses the drop frequency attenuation gradient in the cumulative drop count domain as the input for margin inversion. Within the resistance stability range, it eliminates the equivalent conductivity coefficient of the fluid path using the pressure-flow relationship and volume conservation relationship, and then solves for the remaining volume based on the bottle geometry and infusion parameters. After the roller clamp or infusion tubing state changes, the electronic device continues to generate the drop frequency attenuation gradient from the new resistance stability range, and uses the existing margin change trajectory as the target margin continuity benchmark.

[0087] The electronic device can store the processing result as the current remaining amount and the remaining amount change trajectory arranged according to the cumulative number of drops, and send the current remaining amount to the infusion monitoring interface or subsequent processing module for use in infusion status display, remaining amount warning or nursing treatment.

[0088] As an application example, during ventilated gravity infusion in a ward, the nurse suspends the infusion bottle on the infusion stand and installs the photoelectric drop monitoring terminal at the corresponding position in the drip chamber. The electronic device obtains the initial effective volume of the infusion bottle from the infusion management terminal, reads the bottle's geometric parameters according to the bottle type identification, and simultaneously receives the drop event timestamps continuously output by the drop monitoring terminal.

[0089] After the infusion begins, the electronic device determines the cumulative drip count and local drip frequency according to the order of drip events. As the fluid level in the infusion bottle decreases, the local drip frequency continuously decreases as the cumulative drip count increases. When the nurse adjusts the roller clamp to cause a jump in the local drip frequency, the electronic device identifies the corresponding drip count position as the interval boundary and establishes resistance stabilization intervals before and after the interval boundary, while retaining the already accumulated drip count and the previously determined residual state.

[0090] Within each resistance stability range, the electronic device determines the drip frequency attenuation gradient based on the negative rate of change of the local drip frequency logarithm along the direction of the cumulative drip number, and combines the pressure-flow relationship and the volume conservation relationship to eliminate the equivalent conduction coefficient of the liquid path from the margin constraint relationship. Based on the drip frequency attenuation gradient and the liquid level height and cross-sectional area corresponding to the candidate remaining volume, the electronic device solves for the candidate margin that satisfies the margin constraint relationship.

[0091] When the liquid level in the infusion bottle passes through the shoulder area and generates multiple candidate remaining volumes, the electronic device selects a candidate remaining volume that can be connected to the previous target remaining volume along the same continuous solution branch, based on the condition that the remaining volume decreases or remains unchanged during the cumulative drop count, and determines this as the current remaining volume. Therefore, even when the roller clamp adjustment causes changes in the local drip frequency level and the bottle geometry generates multiple candidate solutions, the electronic device can still update the dynamic prediction result of the infusion bottle remaining volume along the direction of continuous liquid consumption.

[0092] Regarding the implementation details of drop event correction and resistance stabilization interval determination in the method of this application embodiment, in some examples of this application embodiment, the drop interval is determined according to the time difference of the timestamps of adjacent drop events, and adjacent drop events with a drop interval less than the preset minimum drop interval are merged into a valid drop event to obtain a valid drop event sequence.

[0093] The preset minimum droplet formation interval can be set based on the shortest time required for the drip tip to complete one droplet formation and detachment, combined with the time resolution of the drop monitoring terminal. The time difference between adjacent timestamps is calculated according to the order of drop events. When the time difference between consecutive drop events is less than the preset minimum droplet formation interval, the corresponding drop events are merged as repeated triggering results generated when the same droplet passes through the detection position. During merging, the timestamp of the earliest occurring drop event among the adjacent events can be retained. The valid drop events obtained after merging are used as the time observation objects for subsequent missed detection correction and local drop frequency calculation.

[0094] Within the sliding drop count window of the effective drop event sequence, a reference drop interval is determined, and the preceding reference drop frequency and the following reference drop frequency are determined according to the sliding drop count windows before and after the drop interval to be corrected. For drop intervals to be corrected that are greater than the reference drop interval, the integer closest to and greater than 1 in the ratio of the drop interval to be corrected to the reference drop interval is determined as the target integer multiple, and the absolute value of the difference between the drop interval to be corrected and the product of the target integer multiple and the reference drop interval is determined as the time matching deviation.

[0095] The sliding drop count window can be set according to a preset number of consecutive valid drop events, and the window length is determined based on the drop frequency update cycle and the number of valid drop events required for drop interval statistics. The reference drop interval can be the median of each drop interval within the sliding drop count window. The front reference drop frequency and the back reference drop frequency are determined based on the valid drop count and time span within the two sliding drop count windows before and after the drop interval to be corrected, respectively, and are used to characterize the drop state on both sides of the drop interval to be corrected. The target integer multiple reflects the number of reference drop cycles contained in the drop interval to be corrected, and the time matching deviation reflects the difference between the drop interval to be corrected and the corresponding number of reference drop cycles. The preset time tolerance can be set according to the timestamp resolution of the drop monitoring terminal and the range of change of the drop interval relative to the reference drop interval under stable infusion conditions; the preset continuity threshold can be set according to the range of local drop frequency changes between adjacent sliding drop count windows when the equivalent conduction coefficient of the fluid path takes the same value.

[0096] When the time matching deviation does not exceed the preset time tolerance, and the absolute value of the difference between the previous reference drop frequency and the subsequent reference drop frequency does not exceed the preset continuity threshold, the number obtained by subtracting one from the target integer multiple is determined as the number of missed drops, and the same number of virtual counter bits are added between the two corresponding valid drop events; the virtual counter bits are only used to correct the cumulative drop count and do not have drop event timestamps.

[0097] The virtual counter is inserted between two valid drop events according to the order of occurrence of the missed drop events and participates in the calculation of the cumulative drop count increment. The local drop frequency is determined based on the time difference between the cumulative drop count increment corrected by the virtual counter within the corresponding sliding drop count window and the first and last valid drop events with actual timestamps. This corrects the cumulative drop count gap caused by missed detections, while the time span of the local drop frequency is still determined by the actual drop event timestamps output by the drop monitoring terminal.

[0098] When the interval between drops to be corrected exceeds a preset pause threshold, and the absolute value of the time matching deviation or the difference between the previous reference drop frequency and the subsequent reference drop frequency exceeds the corresponding threshold, the corresponding drop number range is determined as an abnormal pause window.

[0099] The preset pause threshold can be set according to the upper limit of the allowable drop interval under stable infusion conditions. When the time matching deviation between the drop interval to be corrected and an integer number of reference drop intervals exceeds the preset time tolerance, or when the absolute value of the difference between the previous reference drop frequency and the subsequent reference drop frequency exceeds the preset continuity threshold, the increase in the virtual counter is set to zero, and the corresponding drop number range is marked as the abnormal pause window. The abnormal pause window is used to characterize the observation range corresponding to the interruption of drop event detection or the pause in the infusion process. The abnormal pause window is retained in the drop event processing results, and its boundary is used as the calculation boundary when determining the continuous drop frequency change relationship in the future.

[0100] Furthermore, the cumulative drop count is determined based on the occurrence order of the valid drop events and the virtual counter bits, and the local drop frequency corresponding to each cumulative drop count is determined based on the timestamp of the valid drop events and the corrected cumulative drop count; when the absolute value of the logarithm of the ratio of the median of the local drop frequencies before and after the position to be determined exceeds the preset frequency jump threshold, the position to be determined is determined as the drop frequency segment boundary, and the continuous drop count interval between adjacent drop frequency segment boundaries after excluding the abnormal pause window is determined as the resistance stability interval.

[0101] When determining the local drop frequency, sliding drop count windows can be set before and after each position to be determined. For each sliding drop count window, the cumulative drop count increment after correction by virtual counter bits is determined, and the time difference between the first and last valid drop events with actual drop event timestamps is determined. The ratio of the cumulative drop count increment to the time difference is determined as the local drop frequency of the corresponding sliding drop count window.

[0102] The median local drop frequency within the sliding drop count window before and after the position to be determined is determined, and the absolute logarithm of the ratio of the two local drop frequency medians represents the degree of drop frequency jump on both sides of the position to be determined. A preset frequency jump threshold can be set based on the range of local drop frequency changes between adjacent sliding drop count windows when the equivalent conduction coefficient of the liquid path takes the same value. Positions exceeding the preset frequency jump threshold are defined as drop frequency segment boundaries, allowing drop frequency jumps caused by changes in the liquid path state to be handled separately from continuous drop frequency changes within each interval.

[0103] When an abnormal pause window is present between adjacent drop frequency segment boundaries, the abnormal pause window is used as the continuous calculation boundary. Each continuous drop number interval formed after excluding the abnormal pause window is determined as a resistance stability interval. The resistance stability interval is used to limit the calculation range of subsequent robust smoothing and drop frequency decay gradient, so that the corresponding calculation is limited to the drop number interval where the equivalent conduction coefficient of the liquid path takes the same value and the drop event observation is continuous.

[0104] The above preset parameters can be set according to the specifications of the infusion set, the time recording accuracy of the drip monitoring terminal, and the pre-collected stable infusion data.

[0105] Figure 2 A flowchart illustrating an example of the operation of constructing a margin constraint relationship based on the drop frequency attenuation gradient in the method according to an embodiment of this application is shown.

[0106] like Figure 2 As shown, in step S210, the logarithmic drop frequency sequence arranged along the cumulative drop count is determined based on the ratio of the local drop frequency corresponding to each sliding drop count window to the preset positive reference frequency.

[0107] The number is determined according to the following formula. Logarithmic drop frequency corresponding to each sliding drop window:

[0108] Equation (12)

[0109] in, For the sliding drop window index, For the first The cumulative number of drops corresponding to each sliding drop window. For the first The local drop frequency corresponding to each sliding drop window To preset a positive reference frequency, For the first The logarithmic drop frequency corresponding to each sliding drop window. Preset positive reference frequency. With local drip frequency Using the same frequency unit and satisfying The frequency can be determined based on the target drip frequency of the infusion set, the stable drip frequency at the beginning of the infusion, or pre-collected stable infusion data, and the same value is used throughout the same infusion process. The fixed preset positive reference frequency only changes the calculation basis of the logarithmic drip frequency, and the rate of change of the logarithm of the local drip frequency relative to the cumulative number of drops remains consistent.

[0110] In step S220, the logarithmic drop frequency sequence is robustly smoothed within each resistance stability interval to reduce the impact of logarithmic drop frequencies with fitting deviations exceeding a preset deviation threshold on the smoothing result, limit the second-order change between adjacent cumulative drop number positions, and constrain the smoothing result to decrease or remain unchanged as the cumulative drop number increases, thereby obtaining a smoothed logarithmic drop frequency sequence.

[0111] The first The set of indices of the sliding drop number windows within a resistance stability interval is denoted as Let the sets of indices that can form first-order and second-order differences be denoted as follows: and For candidate logarithmic drop frequency function The smoothing objective function corresponding to the current resistance stability interval is determined according to the following formula:

[0112] Equation (13)

[0113] in, Index for the resistance stability range, For the first The smoothing objective function corresponding to each resistance stability interval For parameters The corresponding Hubel loss function, The preset deviation threshold is positive. A positive smoothing coefficient. This is a second-order difference operator in the direction of cumulative drops. The Hubel loss function is for absolute values ​​not exceeding... Apply a squared penalty to the fit bias, for absolute values ​​exceeding The fitting bias is penalized with a linear relationship to the absolute value of the bias.

[0114] Based on the aforementioned smoothing objective function, the smoothing logarithmic drop frequency function is determined according to the following formula:

[0115] Equation (14)

[0116] in, For the cumulative drop count variable, For a smooth logarithmic drop frequency function, This is a first-order difference operator in the direction of cumulative drops. (Constraints) The smooth logarithmic drop frequency function is constrained to decrease or remain constant as the cumulative drop count increases. In practice, the first-order difference is determined based on the logarithmic drop frequency difference between adjacent sliding drop count windows, and the second-order difference is determined based on the change of adjacent first-order differences. Both are normalized according to the corresponding cumulative drop count difference to accommodate situations where the intervals between adjacent cumulative drop count positions are different.

[0117] Preset deviation threshold The smoothing coefficient can be determined based on the recording accuracy of the drip event timestamps and the fitting deviation distribution of the logarithmic drip frequency under stable infusion conditions. Calibration can be performed based on pre-collected stable infusion data; when When the value increases, the weight of the second-order difference term in the smoothing objective function increases; when When the value decreases, the relative effect of the fitting bias term increases. The solution is performed separately for each resistance stability interval, and the abnormal pause window is used as a continuous smooth boundary to process the data corresponding to drop frequency jumps and abnormal pauses separately from the continuous drop frequency changes caused by the drop level drop.

[0118] In step S230, the negative rate of change of the smooth logarithmic drop frequency sequence relative to the cumulative drop number is determined as the drop frequency attenuation gradient corresponding to each cumulative drop number.

[0119] Specifically, the first is determined according to the following formula. Drop frequency attenuation gradient corresponding to each cumulative drop count:

[0120] Equation (15)

[0121] in, For the first The drop frequency attenuation gradient corresponding to each cumulative drop number. When using a continuous smoothing function, determine the continuous smoothing function at the cumulative drop number. The derivative at the point; when using a discrete smoothing sequence, the smoothed logarithmic drop frequency difference between adjacent cumulative drop number positions is divided by the corresponding cumulative drop number difference, and a one-sided difference is used at the end of the resistance stability interval, pointing inwards from the interval. Both the derivative and the difference are confined within the same resistance stability interval. Because the smoothed logarithmic drop frequency function is subject to a monotonically non-increasing constraint, the determined drop frequency decay gradient is non-negative.

[0122] In step S240, the sum of the hydraulic baseline and the liquid level corresponding to the remaining volume is determined as the total equivalent static head, and a margin constraint relationship is established such that the product of the drop frequency attenuation gradient, the liquid surface cross-sectional area corresponding to the remaining volume, and the total equivalent static head is equal to the product of the pressure-flow index and the effective single drop volume.

[0123] Within any resistance stability range, the pressure-flow rate relationship limits the volumetric flow rate to the product of the equivalent conductivity of the liquid path and the pressure-flow rate exponent of the total equivalent static head. The local drip frequency is determined by the ratio of the volumetric flow rate to the effective single-drop volume. According to the volume conservation law, for every unit increase in the cumulative drop count, the remaining volume decreases by one effective single-drop volume. Based on the bottle geometry, the rate of change of the liquid level height relative to the remaining volume is determined by the corresponding liquid surface cross-sectional area.

[0124] Under the condition that the equivalent conduction coefficient of the liquid path takes the same value within the current resistance stability range, the negative rate of change of the logarithm of the local dripping frequency along the cumulative dripping number is used to eliminate the equivalent conduction coefficient of the liquid path, which is a multiplicative constant, and the following margin constraint relationship is established:

[0125] Equation (16)

[0126] in, For the first The remaining volume corresponding to each cumulative drop count. The candidate remaining volumes participating in the residual inversion satisfy... as well as This ensures that the cross-sectional area of ​​the liquid surface and the total equivalent static head are within the effective range corresponding to the gravity-assisted infusion process.

[0127] The margin constraint relationship uses the drip frequency decay gradient, bottle geometry parameters, effective single-drop volume, pressure-flow index, and hydraulic baseline as inputs, and no longer includes the equivalent conductivity coefficient of the fluid path. Therefore, the drip frequency level can vary separately in different resistance stability intervals, while the corresponding remaining volume is still retrieved according to the same margin constraint relationship within each resistance stability interval.

[0128] Regarding the implementation details of the self-calibration of effective single drop volume, pressure-flow index, and hydraulic baseline in the method of this application embodiment, in some examples of this application embodiment, the self-calibration operation of the effective single drop volume, the pressure-flow index, and the hydraulic baseline can also be performed based on the margin constraint relationship and the bottle geometry parameters.

[0129] Specifically, any of the aforementioned resistance stability intervals is defined as calibration intervals; for each candidate single drop volume within the preset single drop volume range, based on the target margin at the calibration reference position and its corresponding cumulative number of drops, the number of new drops after the calibration reference position and the volume consumption determined by the corresponding candidate single drop volume are deducted to obtain the candidate remaining volume.

[0130] The calibration interval is selected from the established resistance stability interval, and the drop frequency attenuation gradient corresponding to the cumulative drop count position used in the calibration is positive. The position where the target margin was most recently successfully determined before the starting position of the calibration interval is determined as the calibration reference position, and the cumulative drop count and target margin corresponding to the calibration reference position are denoted as follows: and The candidate remaining volume corresponding to each candidate single drop volume is determined according to the following formula:

[0131] Equation (17)

[0132] in, To preset the single drop volume range, For candidate single drop volume, To use candidate single drop volume Time The candidate remaining volume corresponds to each cumulative drop count. During the initial self-calibration, the location where the initial effective volume is obtained is determined as the calibration reference location, and... as well as When re-calibrating after a change in infusion status, the most recently successfully determined target residual volume and its corresponding cumulative drops will be used as the calibration reference. This will satisfy... The candidate remaining volume is used for subsequent bottle geometry parameter queries.

[0133] The preset single-drop volume range can be set based on the nominal drop factor of the infusion set, the properties of the infusion fluid, and pre-collected single-drop volume data. Candidate single-drop volumes can be selected sequentially from the preset single-drop volume range according to a preset search interval, or they can be determined through a numerical search method limited by the preset single-drop volume range.

[0134] Based on the bottle geometry parameters, determine the rate of change of the liquid surface cross-sectional area relative to the remaining volume for each candidate remaining volume. Then, form a set of calibrated drops by accumulating the number of drops whose absolute value of the rate of change reaches a preset area change threshold. Finally, determine the candidate single drop volume corresponding to the number of cumulative drops in the calibrated drop volume set that reaches a preset minimum calibrated drop volume as the effective candidate single drop volume.

[0135] When the bottle's geometric parameters are represented by a continuous function, the rate of change is determined by differentiating the function of the liquid surface cross-sectional area. When the bottle's geometric parameters are represented by a discrete data table, the rate of change is determined by the difference between the liquid surface cross-sectional area of ​​adjacent geometric data nodes of the candidate remaining volume and the difference between the remaining volume. A set of calibrated drop counts is formed according to the following formula, and the validity of the candidate single drop volume is determined:

[0136] Equation (18)

[0137] in, Function of cross-sectional area of ​​liquid surface The first derivative with respect to the remaining volume, The preset area change threshold is positive. For candidate single drop volume In the The set of calibration drop numbers corresponding to each calibration interval. The cumulative number of drops included in the calibrated drop count set. Preset area change threshold. The settings can be configured based on the sampling accuracy of the bottle's geometric parameters and the measurement range of the rate of change of the liquid surface cross-sectional area within a section of the bottle with a uniform cross-section. The minimum number of calibration drops is set to be no less than 2, and the required number of observations is set according to the candidate parameter combination evaluation. The screening process considers liquid surface cross-sectional area change rates with an absolute value lower than... The location was excluded from the calibration drop count set, and the cumulative number of drops participating in the parameter evaluation was limited to no less than [number missing]. .

[0138] Each effective candidate single drop volume is combined with each candidate pressure-flow index within the preset pressure-flow index range to form a candidate parameter combination; for each candidate parameter combination, based on the candidate remaining volume, drop frequency attenuation gradient, liquid surface cross-sectional area and liquid surface height of each cumulative drop number in the corresponding calibrated drop number set, the candidate hydraulic baseline corresponding to each cumulative drop number is determined by the margin constraint relationship.

[0139] The preset pressure-flow index range is denoted as... Let any one of the candidate pressure-flow indexes be denoted as The margin constraint relationship is transformed, and the candidate hydraulic baseline corresponding to each candidate parameter combination is determined according to the following formula:

[0140] Equation (19)

[0141] in, For candidate pressure-flow index, For candidate single drop volume Candidate pressure-flow index Next Candidate hydraulic baseline corresponding to each cumulative drop count. Preset pressure-flow index range. It can be set according to the structure of the infusion set and the pre-collected pressure and flow rate data, and consists of positive candidate pressure-flow rate indices.

[0142] The calculation of candidate hydraulic baselines is based on and The condition is used as the basis for self-calibration. Only resistance stability intervals where all calculation positions in the calibration drop count set meet the condition are used for this self-calibration; if the current resistance stability interval does not meet the condition, other resistance stability intervals are selected or valid drop events are accumulated. Each candidate parameter combination thus forms a set of candidate hydraulic baselines corresponding to different accumulated drop count positions.

[0143] Furthermore, when the number of effective candidate droplet volumes is greater than zero, the dispersion of the candidate hydraulic baselines corresponding to each candidate parameter combination within the same calibration interval is determined. The effective candidate droplet volume and candidate pressure-flow index in the candidate parameter combination with the smallest dispersion are determined as the effective droplet volume and the pressure-flow index, respectively, and the median of the corresponding candidate hydraulic baselines is determined as the hydraulic baseline. When the number of effective candidate droplet volumes is zero, the already obtained effective droplet volume, pressure-flow index, and hydraulic baseline are used.

[0144] When using variance to characterize the dispersion of candidate hydraulic baselines, the effective single-drop volume and pressure-flow index are determined according to the following formula:

[0145] Equation (20)

[0146] in, This refers to the set of calibrated drops. Variance calculations were performed on the candidate hydraulic baselines. The results... and The candidate parameter combination with the smallest variance of the candidate hydraulic baseline within the same calibration interval is selected first. When multiple candidate parameter combinations have the same minimum variance at the preset numerical resolution, the candidate parameter combination with the smallest absolute value of the difference between the candidate single drop volume and the effective single drop volume before self-calibration is selected first. If multiple candidate parameter combinations still exist, the candidate parameter combination with the smallest absolute value of the difference between the candidate pressure-flow index and the pressure-flow index before self-calibration is selected next.

[0147] When multiple resistance stability intervals exist that satisfy the calibration conditions, the above parameter optimization can be performed separately for each resistance stability interval, and the calibration result with the smallest corresponding minimum variance value can be determined as the self-calibration result. This is based on the selected effective single-drop volume. The hydraulic baseline is determined according to the following formula based on the corresponding set of calibration drop numbers:

[0148] Equation (21)

[0149] in, This represents median operations. For effective single drop volume The set of calibration drop counts corresponding to the selected calibration interval. Median calculation limits the impact of individual candidate hydraulic baseline deviations on the hydraulic baseline output value.

[0150] When successful confirmation , and When the self-calibration result is used to solve the candidate residual after the corresponding calibration interval, the residual change trajectory formed before self-calibration is retained; when the number of effective candidate single drop volumes is zero, the parameters obtained before the self-calibration operation are continued to be used.

[0151] Regarding the implementation details of infusion state change identification and disturbance correction in the method of this application embodiment, in some examples of this application embodiment, after determining the resistance stability range, the local drip frequency median and the median of the drip frequency attenuation gradient greater than zero can be determined according to the stable drip number window before and after the mutation position, and the drip frequency transition amount and gradient transition amount are determined according to the logarithm of the ratio of the corresponding median after the mutation to the median before the mutation.

[0152] The stable drop count windows are set on both sides of the mutation location. Each stable drop count window is selected from the continuous drop count range of the corresponding side, and the adjacent mutation location, the boundary of the abnormal pause window, or the current data endpoint are used as the data retrieval boundary. Each stable drop count window contains a preset number of effective local drop frequencies and a drop frequency decay gradient greater than zero. When the number of effective data on either side is less than the preset number, the event classification at the current mutation location is delayed until subsequent drop events form a stable drop count window that meets the requirements before identification is performed. The local drop frequency and drop frequency decay gradient used to calculate the corresponding median are both positive values ​​to satisfy the value conditions of logarithmic operation.

[0153] Specifically, the drop frequency transition amount and the gradient transition amount are determined according to the following formula:

[0154] Equation (22)

[0155] in, and The median local droplet frequency is shown before and after the mutation site, respectively. and These are the median drop frequency decay gradients that are greater than zero before and after the mutation site, respectively. For drop frequency transition quantity, This represents the gradient transition. The logarithm of the ratio of the median before and after the event is used to allow comparisons of changes at different drop frequency scales based on relative transition relationships.

[0156] The following conditions are sequentially determined as matching conditions for changes in fluid resistance, changes in hydraulic baseline, and changes in single drop volume: the absolute value of the gradient transition amount does not exceed a preset gradient continuity threshold; the absolute value of the sum of the drop frequency transition amount and the gradient transition amount adjusted by the pressure-flow index does not exceed a preset hydraulic matching threshold; and the absolute value of the sum of the drop frequency transition amount and the gradient transition amount does not exceed a preset drop volume matching threshold within a preset number of consecutive drop windows.

[0157] In one implementation, the three types of transition relationships are normalized according to their corresponding matching thresholds to obtain:

[0158] Equation (23)

[0159] in, For resistance matching error, For hydraulic baseline matching error, This is due to the single-drop volume matching error. A positive gradient continuous threshold, A positive hydraulic matching threshold. The droplet volume matching threshold is positive. When , or When the value is less than 1, it indicates that the matching conditions for the corresponding changes in fluid resistance, hydraulic baseline, or single drop volume are met. The matching condition for single drop volume change must also be met continuously within a preset number of consecutive drop windows.

[0160] When the hydraulic resistance changes, the local dripping frequency level changes with the equivalent conductivity of the hydraulic path, while the dripping frequency attenuation gradient under the corresponding remaining volume maintains a continuous relationship before and after the event. When the hydraulic baseline changes, the total equivalent hydrostatic head changes in the opposite direction to the dripping frequency attenuation gradient. When the effective single-drop volume changes, the effective single-drop volume changes in the same direction as the dripping frequency attenuation gradient, but in the opposite direction to the local dripping frequency. These three physical relationships correspond to the three matching errors mentioned above.

[0161] The gradient continuity threshold can be set based on the gradient fluctuation range when the total equivalent static pressure head remains at the same value and the fluid path resistance changes; the hydraulic matching threshold can be set based on the transition relationship when the bottle height or the equivalent pressure at the patient end changes; the drop volume matching threshold can be set based on the transition relationship when the droplet formation state changes. Each matching threshold can also be calibrated in conjunction with the drop event recording accuracy and the fluctuation range of the corresponding combination volume in stable infusion data. The length of the preset recovery window is determined based on the drop frequency recovery time after a transient pause and the effective number of drops required to form a stable drop count window.

[0162] For abrupt changes that persist after the preset recovery window has ended, the event type is determined based on the condition that only one change matching condition is met. If the event is determined to be a change in hydraulic resistance, an updated resistance stability interval is formed. If the event is determined to be a change in hydraulic baseline, the target margin before and after the event is kept continuous with the nearest target margin before the event, and the hydraulic baseline is updated according to the inverse proportional relationship between the total equivalent static head and the drip frequency attenuation gradient. If the event is determined to be a change in single drop volume, the effective single drop volume is updated according to the same proportional relationship between the effective single drop volume and the drip frequency attenuation gradient. The updated resistance stability interval, hydraulic baseline, or effective single drop volume is then used for subsequent candidate margin calculations.

[0163] If the abrupt change persists after the preset recovery window ends, and only one of the three change matching conditions is met, the state change corresponding to the met condition is determined as the event type. When it is determined to be a fluid resistance change event, the abrupt change location is used as the boundary of the updated resistance stability interval, and the drop frequency decay gradient is redefined after the boundary. This event uses the current effective single drop volume, pressure-flow index, and hydraulic baseline.

[0164] When a hydraulic baseline change event is identified, the nearest target margin before the event is recorded as... The hydraulic baseline after the event is determined according to the following formula:

[0165] Equation (24)

[0166] in, The hydraulic baseline applicable before the event, This serves as the hydraulic baseline applicable after the event. In equation (24), with... The corresponding liquid level remains constant as a margin continuity condition, making the ratio of the total equivalent static head before and after the event inversely proportional to the ratio of the median of the drip frequency decay gradient before and after the event. When the total equivalent static head before the event is positive, the updated total equivalent static head remains positive.

[0167] The updated hydraulic baseline takes effect from the time the candidate margin is determined after the event type is identified. The target margin and margin change trajectory determined before the event are retained, and the candidate margin after the event is solved using the updated hydraulic baseline.

[0168] When a single-drop volume change event is identified, the effective single-drop volume after the event is determined according to the following formula:

[0169] Equation (25)

[0170] in, The effective single drop volume applicable before the event. Let be the effective single-drop volume applicable after the event. In Equation (25), the ratio of the effective single-drop volume after the event to that before the event is equal to the ratio of the median drop frequency decay gradient after the event to that before the event. When the effective single-drop volume before the event is positive, the updated effective single-drop volume remains positive.

[0171] The updated effective single-drop volume takes effect from the first stable drop count window after the event type is determined. When subsequently entering a bottle segment that meets the geometric self-calibration conditions, the effective single-drop volume can be determined again based on the self-calibration results.

[0172] Furthermore, when the local drop frequency and the drop frequency attenuation gradient are respectively restored to a preset tolerance range based on the median before the corresponding event within the preset recovery window, or when the number of valid change matching conditions is zero or greater than 1, the corresponding drop number interval is determined as a transient abnormal window, and the update of the target margin based on the drop frequency attenuation gradient within the transient abnormal window is paused.

[0173] The preset tolerance range can be determined based on the fluctuation boundaries of the local drop frequency and drop frequency decay gradient relative to their respective medians within the stable drop count window before the event. Within the recovery window, it is determined whether the local drop frequency and drop frequency decay gradient have entered their respective tolerance ranges based on the continuously sliding drop count window. When both have entered their respective tolerance ranges, the corresponding change is treated as a recovery-type disturbance; when the number of valid change matching conditions is zero, the current transition relationship lacks a corresponding preset event type; when the number of valid change matching conditions is greater than one, the current transition relationship is determined to be a multi-type matching state. All of the above drop count intervals are determined as transient anomaly windows.

[0174] The effective drop count generated within the transient anomaly window continues to be included in the cumulative drop count, and the target balance uses the most recently determined result before the update pause. When resolving the candidate balance, the decrease range of the target balance relative to the target balance before the update pause is limited based on the cumulative drop count increased within the transient anomaly window, and the target balance is updated only using the drop frequency attenuation gradient that satisfies the balance constraint relationship after recovery. When the absolute value of the balance constraint residual after recovery exceeds the preset residual threshold, the update pause state continues; when a stable drop count window that satisfies the event classification condition or the candidate balance solution condition is re-formed, the corresponding processing is resumed. Thus, continuous and uniquely matched transitions are used for state or parameter updates, and recovery-type disturbances and multi-type matched states are restricted within the transient anomaly window, preventing the corresponding drop frequency change from being directly converted into an infusion bottle balance change.

[0175] Figure 3 A flowchart illustrating an example of determining the target margin under continuous solution of branch constraints in the method according to an embodiment of this application is shown.

[0176] like Figure 3 As shown, in step S310, the previously successfully determined target balance and its corresponding cumulative drop count are obtained, and the new drop count is determined based on the difference between the current cumulative drop count and the corresponding cumulative drop count; the physical consumption range expanded by the preset volume tolerance is determined based on the new drop count and the preset single drop volume range, and the balance search interval from zero to the previously successfully determined target balance range is determined based on the physical consumption range.

[0177] Specifically, the current cumulative drop count position is recorded as... The remaining amount of the previously successfully determined target is recorded as... The new number of drops between the current cumulative drop count and the cumulative drop count corresponding to the previously successfully determined target balance is recorded as . The remaining search interval is determined according to the following formula:

[0178] Equation (26)

[0179] in, and These are the lower and upper limits of the preset single drop volume range, respectively. The preset volume tolerance is positive. This represents the remaining volume search interval corresponding to the current cumulative drop count position. This interval simultaneously limits the remaining volume to no more than the previously successfully determined target remaining volume, and ensures that the volume decrease corresponding to the new drop count is within the physical consumption range expanded by volume tolerance. (Subscript) This indicates the most recent successful solution state before the current solution position, and the cumulative drop interval between the two solution positions is... express.

[0180] The preset single-drop volume range can be determined based on the nominal drop factor of the infusion set, the calibration results of the effective single-drop volume, and the allowable variation range of droplet formation state, and must meet the following requirements: The preset volume tolerance can be set based on the calibration error of the effective single drop volume, the cumulative drop count correction error, and the allowable volume deviation between adjacent target reserves. When the reserve search interval is empty, the number of candidate reserves corresponding to the current cumulative drop count position is set to zero, and the process proceeds to step S343 for infusion status verification.

[0181] In step S320, based on the drop frequency attenuation gradient corresponding to the current cumulative drop count and the currently applicable effective single drop volume, pressure-flow index, and hydraulic baseline, the residual of each remaining quantity to be determined within the remaining quantity search interval under the remaining quantity constraint relationship is determined.

[0182] The residual function constrained to the current cumulative drop count position is determined by the following formula:

[0183] Equation (27)

[0184] in, The remaining quantity to be determined is within the remaining quantity search interval. This represents the drop frequency attenuation gradient corresponding to the current cumulative drop count. The effective single drop volume currently applicable after identifying changes in infusion status. This is the applicable hydraulic baseline after identifying changes in infusion status. margin to be determined Corresponding margin constraint residuals. Pressure-flow index. The current parameter calibration results are used, and the same value is maintained during the calculation of the remaining amount at the current cumulative drop count position.

[0185] A preliminary screening is performed based on the absolute value of the residuals with remaining constraints, forming a preliminary set of residuals:

[0186] Equation (28)

[0187] in, The preset residual threshold is positive. The residual initial screening set consists of residuals to be judged whose absolute value does not exceed a preset residual threshold, thereby providing a numerical solution that satisfies the residual constraint condition.

[0188] The preset residual threshold can be determined based on the estimation error of the drip frequency attenuation gradient, the interpolation error of the bottle's geometric parameters, and the calibration error of the currently applicable parameters. When obtaining the initial residual screening set, the residual search interval can be divided into multiple volume sub-intervals based on the volume nodes corresponding to the bottle's geometric parameters. The residual constraint residuals at each volume node are calculated, and interpolation or iterative solutions are performed within the volume sub-intervals where the residual sign changes or the absolute value of the residual forms a local minimum. The obtained numerical solutions are substituted into the residual constraint residual function for verification, and numerical solutions whose absolute residual value does not exceed the preset residual threshold are included in the initial residual screening set.

[0189] In step S330, the remaining amount to be determined is determined as a candidate remaining amount if the absolute value of the remaining amount constraint residual does not exceed a preset residual threshold, can be connected with the previously successfully determined target remaining amount along the same continuous solution branch formed by the remaining amount constraint relationship changing with the cumulative number of drops, and the remaining volume on the continuous solution branch decreases or remains unchanged as the cumulative number of drops increases.

[0190] In one implementation, each undetermined residual in the initial residual screening set is subjected to continuous deextension along the direction of increasing cumulative drops, and the final candidate residual set is formed according to the following formula:

[0191] Equation (29)

[0192] in, This represents the final candidate residual set corresponding to the current cumulative drop count position. margin to be determined The branch identifier of the continuous solution branch. This is the branch identifier for the continuous solution branch to which the previously successfully determined target surplus belongs. Only when the surplus to be determined can be continuously extended from the previously successfully determined target surplus along the direction of increasing cumulative drops, and the remaining volume decreases or remains unchanged during the extension process, will this be... Set to with The same applies. This eliminates other numerical solution branches generated by the margin constraint relationship in the section where the cross-sectional area of ​​the bottle changes.

[0193] The continuous solution extension is performed in the order of increasing cumulative drops. After obtaining the initial set of residuals at adjacent valid calculation positions, the difference between the previous numerical solution and the current numerical solution is determined as the volume reduction. When the volume reduction is within the physical consumption range determined by the corresponding increase in drops, and the remaining volume corresponding to the current numerical solution is less than or equal to the remaining volume corresponding to the previous numerical solution, the current numerical solution inherits the branch identifier of the previous numerical solution. The remaining numerical solutions are configured with branch identifiers corresponding to their respective extension paths. For multiple numerical solutions existing simultaneously in the bottle shoulder or bottom contraction region, their respective branch identifiers are maintained according to the extension relationship, restricting the cross-switching of numerical solutions corresponding to different geometric segments between adjacent cumulative drop positions.

[0194] In step S341, when the number of candidate reserves is greater than zero, a drop count conservation reference reserve is determined based on the previously successfully determined target reserve, the newly added drop count, and the currently applicable effective single drop volume. The candidate reserve with the smallest absolute value of the difference between the drop count conservation reference reserve and the target reserve is determined as the target reserve. The target reserve is determined as the current reserve, and the current reserve is used as the dynamic prediction result of the infusion bottle reserve.

[0195] Specifically, the current target margin is determined from the final candidate margin set according to the following formula:

[0196] Equation (30)

[0197] in, This represents the target remaining amount corresponding to the current cumulative drop count position. This is the drop number conservation reference margin determined based on the previously successfully determined target margin, the number of additional drops, and the currently applicable effective single drop volume. After continuously solving for the branch access, the root is selected according to the absolute value of the difference between this and the drop number conservation reference margin.

[0198] When the absolute values ​​of the differences between two or more candidate residuals and the drop count conservation reference residuals are the same, the candidate residual with the smallest absolute value of the residual constraint residual is selected first. When the absolute values ​​of the residual constraint residuals are still the same, the candidate residual obtained by first extending the previous target residual along the current continuous solution branch is selected. After determining the current target residual, its corresponding continuous solution branch identifier is used as the previous branch identifier for the next valid calculation position.

[0199] In step S343, when the number of candidate reserves is zero, an inversion mismatch marker is generated and the infusion status is verified. Based on the verification result of the infusion status, the applicable resistance stability range, effective single drop volume, and hydraulic baseline are determined, and the drop count conservation reference reserve and candidate reserves are redefined. When the number of redefined candidate reserves is greater than zero, the candidate reserve with the smallest absolute value of the difference between it and the redefined drop count conservation reference reserve is determined as the target reserve. The target reserve is determined as the current reserve, the inversion mismatch marker is removed, and the dynamic prediction result of the infusion bottle reserve is updated. When the number of redefined candidate reserves is zero, the inversion mismatch marker is retained, and the most recent dynamic prediction result of the infusion bottle reserve is marked as paused.

[0200] The infusion status verification includes reassessing the event type corresponding to the current drip frequency segment boundary and selecting to update the resistance stability interval, effective single-drop volume, or hydraulic baseline based on the verification results. After parameter updates, the margin search interval, margin constraint residual, continuous solution branch, and drop number conservation reference margin are redefined. When candidate margins are re-formed, the inversion mismatch flag is removed; when the number of candidate margins is zero after verification, the most recently successfully determined margin value is retained and the update pause state is maintained.

[0201] During the pause update period, drop events continue to be received and the cumulative drop count is updated. The most recently successfully determined target surplus is maintained as the current output result until candidate surpluses are re-formed. When a subsequent stable drop count window forms a drop frequency decay gradient that satisfies the positive value condition, the most recently successfully determined target surplus and its corresponding cumulative drop count are used as the benchmark. The drop counts added during the pause update period are included in the new drop count. The surplus constraint residuals of each surplus to be determined within the surplus search interval are re-determined. Candidate surpluses are determined based on the surplus constraint residuals and the continuous solution branch. The target surplus is then determined from the candidate surpluses based on the drop count conservation reference surplus. After re-forming candidate surpluses and determining the target surplus, the pause update state is lifted and the surplus change trajectory continues to be generated. Thus, the surplus search interval limits the physical consumption range of the current solution, the continuous solution branch limits the bottle geometry path to which the current solution belongs, and the drop count conservation reference surplus completes the selection of the target surplus within the same branch.

[0202] Regarding the implementation details of determining the remaining confidence interval, remaining infusion time interval, and early warning result in the method of this application embodiment, in some examples of this application embodiment, the standard uncertainty of the remaining constraint residual is determined based on the drop frequency attenuation gradient corresponding to the current cumulative drop count, bottle geometric parameters, effective single drop volume, pressure-flow index, and error propagation of the hydraulic baseline.

[0203] Specifically, the standard uncertainty of the residual constraint corresponding to the current cumulative drop count position is denoted as... The standard uncertainty of the residual constraint is determined based on the standard uncertainty of each input quantity in the residual constraint residual function and the propagation relationship of each input quantity to the residual constraint residual. It is used to characterize the comprehensive fluctuation range of the input error on the current residual constraint residual.

[0204] The standard uncertainty of each input quantity can be determined based on the fitting residual of the drop frequency decay gradient within the resistance stability interval, the measurement error or interpolation error of the bottle's geometric parameters, and the calibration results of the effective single drop volume, pressure-flow index, and hydraulic baseline. The current margin is used as the linearization position for error propagation, and the rate of change of the margin-constrained residual function relative to each input quantity is determined. When the errors of each input quantity are independent, the squares of the products of each rate of change and the corresponding standard uncertainty are summed, and the square root of the summation is obtained. When the errors of some input quantities are correlated, the corresponding covariance propagation term is added to the sum of squares. When the bottle geometry parameters simultaneously affect the liquid surface cross-sectional area and the liquid surface height, error propagation can be performed directly based on the bottle geometry parameters, or the liquid surface cross-sectional area and the liquid surface height can be used as correlated input quantities and included in their covariance.

[0205] The margin inversion sensitivity is determined based on the absolute value of the rate of change of the margin relative to the remaining volume at the current margin, according to the margin constraint relationship.

[0206] Based on the bottle's geometric relationship that the rate of change of liquid level height relative to the remaining volume is equal to the reciprocal of the corresponding liquid surface cross-sectional area, the residual inversion sensitivity is determined according to the following formula:

[0207] Equation (31)

[0208] in, Current margin The corresponding residual inversion sensitivity, Function of cross-sectional area of ​​liquid surface In the current balance The rate of change of the liquid level relative to the remaining volume. Since the liquid level height function satisfies... For the residual function with residual constraint in After differentiation, the product term corresponding to the change in liquid level height becomes 1. The residual inversion sensitivity is used to characterize the degree of response of the residual constraint residual to changes in the remaining volume.

[0209] The rate of change of the liquid surface cross-sectional area can be obtained by differentiating the continuous function corresponding to the bottle's geometric parameters, or it can be determined by difference based on the geometric data nodes on both sides of the current balance. When the current balance is located at the boundary of the geometric data range, unilateral difference is performed using adjacent data nodes facing the middle of the geometric data range.

[0210] When the margin inversion sensitivity exceeds the preset minimum inversion sensitivity threshold, the margin standard uncertainty is determined according to the ratio of the margin constraint residual standard uncertainty to the margin inversion sensitivity, and the margin confidence interval, which is limited to zero to the initial effective volume range, is determined in combination with the current margin and the preset confidence coefficient.

[0211] Specifically, the standard uncertainty and confidence interval of the margin are determined according to the following formula:

[0212] Equation (32)

[0213] in, This represents the standard uncertainty of the residual at the current cumulative drop count position. This represents the margin confidence interval corresponding to the current margin. The pre-set confidence coefficient is positive. The maximum value operation restricts the lower limit of the margin confidence interval to be greater than or equal to zero, while the minimum value operation restricts its upper limit to be less than or equal to the initial effective volume. .

[0214] The preset minimum inversion sensitivity threshold is positive, and is set based on the numerical accuracy of the residual constraint residual, the volumetric resolution of the bottle's geometric parameters, and the acceptable upper limit of the residual inversion error. The preset confidence coefficient is determined based on the confidence level used. Only... When the preset minimum inversion sensitivity threshold is exceeded, Mapped to This is to limit the impact of error amplification when the margin constraint relationship is in the low sensitivity range.

[0215] When the margin confidence interval is formed, the equivalent fluid circuit conductivity coefficient is determined based on the local drip frequency within the current resistance stability interval, the currently applicable effective single drop volume, the pressure-flow index, the hydraulic baseline, and the current margin. When the fluid circuit conductivity coefficient remains unchanged, the infusion flow rate is determined based on the change in the total equivalent static head as the remaining volume decreases and the pressure-flow relationship. The infusion time consumed from the two boundaries of the margin confidence interval to zero margin is accumulated to obtain the remaining infusion time interval.

[0216] The equivalent conductivity coefficient of the fluid path and the remaining infusion time function corresponding to the current resistance stability interval are determined by the following formula:

[0217] Equation (33)

[0218] in, This represents the equivalent conduction coefficient of the hydraulic circuit corresponding to the current resistance stability range. This represents the local drop frequency corresponding to the current cumulative drop count. The remaining volume to be predicted The infusion time required for the volume to drop to zero. The remaining volume is the integral variable. When the confidence interval of the remaining amount is expressed as... When, since the integral function is positive under the condition that it holds true, Follow Increase and increase, therefore and These are respectively determined as the lower and upper limits of the remaining infusion time interval.

[0219] When determining the remaining infusion time interval, the following methods are used: , and The current infusion status, and ensure that the total equivalent static head corresponding to each remaining volume within the remaining volume confidence interval satisfies The integral can be calculated by numerically accumulating discrete volume nodes in the bottle's geometric parameters, or directly when the liquid level height function has a continuous expression. When there are remaining volume locations within the integration interval where the total equivalent static head is less than or equal to zero, the remaining infusion time interval is marked as temporarily unavailable, and the current prediction result is transferred to the insufficient observation processing.

[0220] When the confidence interval for remaining fluid volume and the remaining infusion time interval are formed, a low fluid volume warning is generated when the upper limit of the confidence interval for remaining fluid volume does not exceed the preset safe fluid volume threshold, and a time limit warning is generated when the upper limit of the remaining infusion time interval does not exceed the sum of the preset nursing response time and the preset bottle change preparation time.

[0221] The upper limit of the confidence interval for remaining capacity is used to determine whether the remaining capacity corresponding to the upper boundary of the confidence interval has entered the safe margin range; the upper limit of the remaining infusion time interval is used to determine whether the longest remaining infusion time corresponding to the confidence interval has entered the treatment range jointly defined by the nursing response time and the bottle change preparation time.

[0222] The preset safety margin threshold can be set according to the allowable volume of fluid to be retained in the infusion bottle and clinical infusion management requirements. The preset nursing response time and preset bottle change preparation time can be configured according to the ward nursing process, bed location, and bottle change operation time. Low remaining volume warnings and time limit warnings are generated according to their respective conditions; when both conditions are met, the low remaining volume warning and the time limit warning are jointly determined as the warning result corresponding to the current dynamic prediction result of the remaining volume of the infusion bottle.

[0223] Furthermore, when the remaining balance inversion sensitivity does not exceed the preset minimum inversion sensitivity threshold, the inversion mismatch flag has been generated but not removed, no valid drip event is obtained for a continuous preset duration, or the width of the formed remaining balance confidence interval exceeds the preset confidence width threshold, the dynamic prediction result of the infusion bottle remaining balance is marked as paused and an insufficient observation prompt is generated.

[0224] Specifically, when the residual inversion sensitivity does not exceed the preset minimum inversion sensitivity threshold, the current state is determined to be a low sensitivity state; when the inversion mismatch marker is not removed, the current state is determined to be an inversion incompatible state; when no valid drop events are obtained for a continuous preset duration, the current state is determined to be a drop event observation interruption state; when the width of the residual confidence interval exceeds the preset confidence width threshold, the current state is determined to be a confidence interval exceeding the limit state. When any of the above states occur, the most recently successfully determined dynamic prediction result of the infusion bottle residual volume is used, and an insufficient observation prompt is output.

[0225] The continuous preset duration can be set according to the maximum allowable normal drip interval of the current infusion protocol and the observation time required to form a local drip frequency. The preset confidence width threshold can be set according to the allowable uncertainty range of the remaining volume prediction results. During the pause update period, drip events continue to be received and the cumulative number of drops is recorded. The most recently successfully determined dynamic prediction result of the remaining volume of the infusion bottle is kept as the current output result. When a valid drip event required to form a local drip frequency is obtained again, the current inversion state passes the compatibility verification, the remaining volume inversion sensitivity exceeds the preset minimum inversion sensitivity threshold, and the width of the newly formed remaining volume confidence interval does not exceed the preset confidence width threshold, the pause update state is lifted, the dynamic prediction result of the remaining volume of the infusion bottle is updated, and the remaining infusion time interval and corresponding warning results are re-determined. Among them, the current inversion state passing the compatibility verification includes the state where the inversion mismatch mark has not been generated or has been removed.

[0226] Figure 4 A schematic diagram illustrating the operational mechanism of an example of a dynamic prediction method for the remaining volume of an infusion bottle based on the drip frequency attenuation gradient according to an embodiment of this application is shown.

[0227] like Figure 4 As shown, the drop events and their timestamps collected during the infusion process are used as the observation input for residual volume prediction. Repeated triggering and merging, missed detection supplementation, and abnormal pause marking are performed on the drop events. Local drop frequencies and resistance stability intervals are determined based on the valid drop event timestamps and the corrected cumulative drop count. Within each resistance stability interval, the local drop frequency is normalized using a preset positive reference frequency. The logarithmic drop frequency sequence arranged along the cumulative drop count is robustly smoothed, and the negative rate of change of the smoothed logarithmic drop frequency relative to the cumulative drop count is determined as the drop frequency attenuation gradient.

[0228] Figure 4 China and Israel This summarizes the relationship between the drop frequency attenuation gradient and the cumulative number of drops. In actual calculations, the local drop frequency... With preset positive reference frequency The ratio is taken logarithmically, and the smoothed logarithmic drop frequency is used to calculate the negative rate of change. Because Using the same positive value during the same infusion process satisfies Therefore, the formula in the figure has the same rate of change as the drip frequency attenuation gradient normalized using a preset positive reference frequency. The drip frequency attenuation gradient characterizes the attenuation relationship of the local drip frequency as the cumulative number of drops increases, and serves as the input for parameter self-calibration, identification of changes in infusion state, and solution of residual constraint relationships.

[0229] A self-calibration operation is performed based on the aforementioned margin constraint relationship, drip frequency decay gradient, and bottle geometry parameters to jointly determine the effective single-drop volume, pressure-flow index, and hydraulic baseline. For sudden changes in drip frequency during infusion, the median local drip frequency and median drip frequency decay gradient before and after the change location are compared to identify changes in fluid path resistance, hydraulic baseline, or effective single-drop volume. Changes in fluid path resistance correspond to the formation of an updated resistance stability range; changes in hydraulic baseline or effective single-drop volume correspond to the updating of the currently applicable margin constraint parameters.

[0230] Substitute the drop frequency attenuation gradient, bottle geometry parameters, and currently applicable margin constraint parameters into the margin constraint relationship, and limit the margin search interval based on the previously successfully determined target margin and the number of new drops. Within the margin search interval, determine candidate margins that satisfy the margin constraint residual condition, the continuous solution branch connection condition, and the non-increasing condition of the remaining volume, and determine the current target margin from the candidate margins.

[0231] When the candidate reserve formation conditions are met, the target reserve is dynamically updated with the continuous input of drip events. The solid line reserve decrease curve in the figure represents the predicted trajectory of the target reserve changing with infusion time, and the dashed lines on both sides of the predicted trajectory represent the corresponding reserve confidence intervals. The remaining infusion time interval is determined based on the reserve confidence interval; a low reserve warning is generated when the upper limit of the reserve confidence interval does not exceed the preset safe reserve threshold, and a time limit warning is generated when the upper limit of the remaining infusion time interval does not exceed the sum of the preset nursing response time and the preset bottle change preparation time.

[0232] Figure 4 The dashed line indicating the pause update trigger path represents the remaining capacity prediction control relationship under abnormal conditions. When the number of matching conditions for changes in infusion status is zero or greater than one, the inversion mismatch marker is valid, the remaining capacity inversion sensitivity does not exceed the preset minimum inversion sensitivity threshold, the number of valid drop events within a consecutive preset time period is zero, or the remaining capacity confidence interval width exceeds the preset confidence width threshold, the dynamic prediction result of the infusion bottle remaining capacity is marked as paused. During the pause update period, drop events and cumulative drop counts continue to be recorded, and the most recently successfully determined target remaining capacity is retained. When subsequent drop events form a stable drop count window, the current inversion status passes compatibility verification, the remaining capacity inversion sensitivity exceeds the preset minimum inversion sensitivity threshold, and the width of the newly formed remaining capacity confidence interval does not exceed the preset confidence width threshold, candidate remaining capacity solving and target remaining capacity updating resume.

[0233] To verify whether the drop frequency attenuation gradient before and after roller clamp readjustment meets the preset gradient continuity threshold, whether the bottle geometry self-calibration operation can determine the effective single drop volume, pressure-flow index, and hydraulic baseline and support residual volume inversion under the condition that the simulation true value is only used as the evaluation benchmark, and the impact of infusion state change identification and continuous solution of branch constraints on the residual error distribution under composite disturbances, this embodiment constructs a verification platform based on digital twin simulation and bench parameter fitting. The underlying physical model of the simulation is established based on the pressure-flow relationship, bottle geometry relationship, and volume conservation relationship, and detection errors, geometric parameter deviations, and non-stationary random noise independent of the residual inversion process are added to the generated drop event data.

[0234] The validation platform included four 500 mL infusion bottle profile geometries: a bottle with a constant cross-section, a bottle with a sudden change in upper shoulder cross-sectional area, a bottle with a continuously varying cross-sectional area on both shoulders, and a bottle with a shrinking bottom cross-sectional area. Infusion sets with nominal drop factors of 15 drops / mL, 20 drops / mL, and 60 drops / mL were matched to these bottles, respectively. Target initial flow rates were set to 50 mL / h, 100 mL / h, 200 mL / h, and 300 mL / h, with pressure-flow index... Continuous sampling was performed within the range of 0.88 to 1.12. The deviation of the effective single-drop volume from the nominal single-drop volume was randomly determined within the range of -8% to +8%. The simulated true value of the hydraulic baseline was set based on the fitting results of the corresponding bench parameters. When performing the bottle geometry self-calibration operation, the preset value ranges of the effective single-drop volume and the pressure-flow index were used as self-calibration inputs. The simulated true values ​​of the effective single-drop volume, pressure-flow index, and hydraulic baseline were only used to calculate the self-calibration error.

[0235] The validation platform generated 600 digital twin sessions of infusion with a margin decreasing from 95% to 5%. Each session included a stable infusion section and randomly set one or more combinations of disturbances, including two roller clamp readjustments, changes in the equivalent height of the infusion stand, missed or repeated drip events, infusion pauses caused by microbubbles, and deviations in bottle geometry parameters. Specifically, the changes in the equivalent height of the infusion stand ranged from 5 to 15 cm, with the direction of increase or decrease randomly determined; the proportion of missed or repeated drip events ranged from 0.5% to 3%; the number of infusion pauses caused by microbubbles ranged from 1 to 5; and the deviation of bottle geometry parameters from the corresponding parameters in the geometric database ranged from 0% to 12%. Timestamp observation noise varying with the drip frequency was also added to the drip event timestamps.

[0236] Using a single infusion digital twin session as the resampling unit, 2000 bootstrap statistical analyses were performed. Comparative methods included Nominal Drip Factor Integration (Group A), Raw Drip Frequency Regression (Group B), DripFrequency Decay Gradient Inversion without Infusion State Change Identification (Group C), and DFDG-DVP (Drip Frequency Decay Gradient-based Dynamic Volume Prediction) as described in this application. DFDG-DVP performs bottle geometry self-calibration, infusion state change identification, continuous solution of branch constraints, and uncertainty propagation. Each method was evaluated using the drip event timestamps, initial effective volume, and bottle geometry parameters within the same infusion digital twin session. In each infusion digital twin session, the same evaluation checkpoint is set, and the predicted margin of each method at the corresponding evaluation checkpoint is compared with the simulation true value. When any method is in a paused update state, the most recent dynamic prediction result of the remaining infusion bottle volume is used as the predicted margin of the corresponding evaluation checkpoint. Evaluation indicators include the absolute logarithm of the ratio of the median of the drop frequency attenuation gradient before and after the roller clamp readjustment and the proportion of sessions that meet the preset gradient continuity threshold, effective single drop volume, pressure-flow index and hydraulic baseline self-calibration absolute error, margin MAE (Mean Absolute Error), margin P95 (95th Percentile Absolute Error), RTAE (Remaining Time Absolute Error), and post-event recovery time. For interference events that cause the target balance update to be paused, 10 mL is determined as the post-event error recovery threshold. The post-event recovery time is the time interval from the occurrence of the combined interference to the completion of the identification of changes in infusion status, the re-formation of candidate balances, the restoration of target balance updates, and the absolute balance error re-entering the 10 mL error band.

[0237] As of the end of the corresponding infusion digital twin session, sessions that meet the above recovery conditions are counted as recovered sessions, and the remaining sessions are counted as unrecovered sessions.

[0238] Based on the above evaluation, further ablation experiments, parameter sensitivity experiments, and residual confidence interval coverage evaluations were conducted. The ablation experiments used the complete DFDG-DVP as a baseline, replacing paused target residual updates within the transient anomaly window with continuous updates, replacing candidate residual selection with screening based solely on residual constraint residuals, replacing bottle geometry self-calibration with using preset residual constraint parameters, and replacing updates to the resistance stability interval or residual constraint parameters after a sudden change in infusion state with using parameters from before the event. All other experimental conditions remained the same, and the MAE increment of each ablation configuration relative to the complete DFDG-DVP was determined. The parameter sensitivity experiments set the gradient window length within the range of 60–180 drops and the bottle geometry mismatch within the range of 0–12%, determining the MAE corresponding to each parameter combination and forming contour lines based on the MAE of adjacent parameter combinations. The residual confidence interval coverage evaluation used a pre-set confidence coefficient. Based on whether the true value of the simulated remaining volume corresponding to each evaluation checkpoint is within the margin confidence interval. The coverage status was determined internally, and the coverage rate corresponding to all sessions and sessions with a bottle geometry mismatch of more than 8% was statistically analyzed using the aforementioned 2000 Bootstrap tests.

[0239] Figure 5 A schematic diagram of the comparative experimental simulation results of different methods under combined disturbance conditions is shown. Figure 5 The graph uses time, local drip frequency, and normalized margin prediction deviation as the joint display coordinates. The horizontal axis represents time, the vertical axis represents the local drip frequency determined based on the drip event timestamps of the same infusion digital twin session, and the vertical axis represents the normalized margin prediction deviation corresponding to different methods. During normalization, the simulated true value of the remaining volume of the infusion digital twin session at each time point is first subtracted from the predicted margin output by the corresponding method to obtain the margin prediction deviation. Then, the maximum absolute value of the margin prediction deviation of each displayed method in the session is used as the common normalization scale. This normalization process is only used for trajectory comparison in the graph; the aforementioned margin MAE, margin P95, and RTAE are still determined based on errors with the original volume units or time units. The solid cyan line marked "Method of this paper" in the graph corresponds to DFDG-DVP, and the light cyan band outside it represents the margin confidence interval after transformation by the same scale; the orange dashed line and gray dashed line represent the margin prediction deviation trajectories of group B and group A, respectively. Figure 5 Typical conversations were displayed for groups A, B, and DFDG-DVP, and the results of group C were included in the aforementioned statistical evaluation.

[0240] like Figure 5In the infusion digital twin session shown, the effective single drop volume is 6% larger than the nominal single drop volume, and the roller clamp is tightened and loosened at 52 min and 108 min, respectively. DFDG-DVP identifies changes in fluid path resistance at the two roller clamp adjustment positions, forming an updated resistance stability interval, and determines the subsequent target margin along the original continuous solution branch. Its normalized margin prediction deviation trajectory is continuously connected along the time direction at the two event positions. Group B directly uses the original drop frequency for margin regression. The local drop frequency jump caused by roller clamp adjustment enters the margin regression result, causing a sudden change in the margin prediction deviation, and a reverse change in the predicted margin appears in subsequent segments. Group A, based on the cumulative volume consumption of the nominal drop factor, has a 6% deviation between the actual effective single drop volume and the nominal single drop volume that enters the margin prediction result as the cumulative number of drops increases, forming the continuous deviation shown by the gray dashed line in the figure.

[0241] At 136 minutes, the equivalent height of the infusion stand in the digital twin session decreased by 12 cm. This event caused the local drip frequency transition and drip frequency decay gradient transition to satisfy the hydraulic baseline change matching condition. DFDG-DVP maintains the continuity of the margin before and after the event using the nearest target margin before the event and updates the applicable hydraulic baseline after the event. Affected by observational fluctuations near the event location, the light cyan margin confidence interval in the figure expands after 136 minutes; when subsequent drop events form a stable drop number window and re-satisfy the candidate margin formation conditions, the margin confidence interval converges as the continuous solution branches recover. Therefore, Figure 5 This reflects the evolution of the margin prediction deviation under different methods under conditions of roller clamp readjustment and equivalent height change.

[0242] Figure 6 This diagram illustrates a comparative simulation of the performance of different methods in dynamically predicting the remaining volume of infusion bottles under various disturbance scenarios. Figure 6 The left subplot uses error density distribution and box plots to represent the absolute residual error distribution of Group A (nominal drop factor integration method), Group B (original drop frequency regression method), Group C (drop frequency attenuation gradient inversion method without infusion state change identification), and Group D (DFDG-DVP). A 10 mL red dashed line represents the preset error evaluation threshold of this embodiment. The MAEs of Groups A, B, and C are 21.8 mL, 30.7 mL, and 11.9 mL, respectively, while the MAE of Group D is 4.8 mL, and the P95 is 12.6 mL. The error distribution of Group B has two density peaks extending into the high error region. Group C forms a tail error exceeding 10 mL at the gradient peak generated at the location of abrupt changes in infusion state. The error samples of Group D are concentrated below the 10 mL error evaluation threshold, and tail data exceeding the threshold are retained. The 10 mL error evaluation threshold serves as a unified evaluation boundary for error distribution and post-event recovery time in this embodiment.

[0243] Figure 6 The upper right portion shows the MAE under different infusion scenarios. In a stable infusion scenario, the MAE of group D after bottle geometry self-calibration was 3.1 mL. When using the nominal drop factor integral method with the nominal single drop volume, the remaining error accumulates with the cumulative number of drops. In the roller clamp readjustment scenario, the MAEs of groups B and D were 38.6 mL and 4.5 mL, respectively. In the equivalent height change scenario of the infusion stand, the MAEs of groups C and D were 14.2 mL and 5.6 mL, respectively. In the multiple combined interference scenario, the lowest MAE in the control method was 18.9 mL, and the MAE of group D was 7.3 mL.

[0244] Figure 6 The lower right portion shows the RTAE and post-event recovery time. The RTAE for group D was 2.3 min, while the RTAEs for groups A and B were 10.9 min and 14.7 min, respectively. When the recovery conditions were: completion of infusion state change identification, re-formation of candidate remaining volumes, update of the target remaining volume, and the absolute remaining volume error re-entering the 10 mL error band, the median recovery time for group D was 1.6 min, and for group B it was 9.8 min. These results reflect the impact of bottle geometry self-calibration, infusion state change identification, and continuous solution of branch constraints on remaining volume error, remaining infusion time error, and post-event recovery process under different disturbance scenarios.

[0245] Figure 7 A comparative simulation diagram illustrating an example of the dynamic prediction performance of infusion bottle remaining volume under different technology processing configurations is shown.

[0246] Figure 7 The left subplot uses the MAE of the complete DFDG-DVP as the baseline for ablation experiments, and changes one technical treatment while keeping other experimental conditions the same. In the figure, "No Anomaly, Pause Update" indicates that the target residual volume is updated within the transient anomaly window, corresponding to an increase in MAE of 1.7 mL; "No Continuous Branch Constraint" indicates that candidate residual volumes are selected only based on residual constraint residuals, corresponding to an increase in MAE of 3.2 mL; "No Geometric Self-calibration" indicates that the preset effective single-drop volume, pressure-flow index, and hydraulic baseline are used, corresponding to an increase in MAE of 4.1 mL; and "No State Change Recognition" indicates that the resistance stability range or residual constraint parameters before the event are used after a sudden change in infusion state, corresponding to an increase in MAE of 5.8 mL. Among the four ablation configurations, the configuration with no state change recognition has the largest increase in MAE.

[0247] Figure 7The right-hand subplot shows the MAE contour lines when the gradient window length and bottle geometric mismatch change together. The gradient window length is evaluated from 60 to 180 drops, and the bottle geometric mismatch from 0% to 12%. The contour lines labeled 5, 7, 9, and 11 represent the MAE for the corresponding parameter combinations, in mL. When the gradient window length is between 64 and 128 drops and the bottle geometric mismatch is between 0% and 6%, the corresponding parameter combination falls within the 5-7 mL contour line range. When the gradient window length is 96 drops and the bottle geometric mismatch is 10%, the MAE corresponding to the red marker in the figure does not exceed 9 mL.

[0248] Figure 7 The lower region also displays the statistical results of the margin confidence interval coverage. When the confidence coefficient is preset... At that time, the margin confidence interval obtained by using 2000 Bootstrap iterations The coverage of the true value of the simulated residual volume is 94.1%, which is 0.9 percentage points lower than the 95% nominal confidence level; when the geometric mismatch of the bottle exceeds 8%, the coverage is 92.7%. As the geometric parameter error increases, the standard uncertainty of the residual constraint and the width of its corresponding residual confidence interval are adjusted according to the error propagation results.

[0249] The aforementioned digital twin and ablation experiments demonstrate that the method described in this application is applicable to the formation of a liquid infusion container during the liquid output process, which can be represented by a function of the cross-sectional area of ​​the liquid surface. and liquid level height function This represents the free surface geometry and enables the infusion process to achieve the effective drop events required to form a drop frequency decay gradient within a preset time period. When the number of effective drop events does not meet the gradient window requirement, the pressure-flow index... If the equivalent resistance of the liquid path changes continuously within the same analysis interval, or if the flexible liquid storage container collapses irregularly and loses the corresponding free liquid surface geometry, the dynamic prediction result of the infusion bottle balance is marked as paused, the target balance determined most recently is retained and an insufficient observation prompt is generated; after a stable drop count window that satisfies the balance constraint relationship is re-established, the candidate balance solution and target balance update are resumed.

[0250] Under the aforementioned applicable conditions, the method of this application utilizes existing drop events and their timestamps to form a drop frequency attenuation gradient, eliminating the need for additional weighing sensors, pressure sensors, or level sensors. It combines bottle geometry self-calibration, infusion state change identification, continuous solution of branch constraints, and uncertainty propagation to determine the target margin, margin confidence interval, and remaining infusion time interval. In the aforementioned 600 infusion digital twin sessions and 2000 Bootstrap statistics, the margin MAE of the method of this application was 4.8 mL, the margin P95 was 12.6 mL, the RTAE was 2.3 min, and the margin confidence interval coverage rate was 94.1%. These results reflect the limiting effect of the above-mentioned technical processing on the margin prediction error caused by roller clamp readjustment, changes in the equivalent height of the infusion stand, abnormal pauses, and deviations in bottle geometry parameters.

[0251] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of combined actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Secondly, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0252] Figure 8 A structural block diagram of an example of a dynamic prediction system for the remaining volume of an infusion bottle based on a drip frequency decay gradient, according to an embodiment of this application, is shown.

[0253] like Figure 8 As shown, the infusion bottle residual dynamic prediction system 800 based on drip frequency attenuation gradient includes a basic data acquisition unit 810, a stable interval determination unit 820, an attenuation gradient determination unit 830, a residual constraint establishment unit 840, a candidate residual solution unit 850, and a target residual determination unit 860.

[0254] The basic data acquisition unit 810 is used to acquire the drop event timestamp sequence, initial effective volume, bottle geometric parameters, effective single drop volume, pressure-flow index and hydraulic baseline during the ventilated gravity infusion process; the bottle geometric parameters represent the correspondence between the remaining volume and the liquid level height and the liquid level cross-sectional area, and the hydraulic baseline represents the equivalent hydraulic height from the infusion bottle outlet reference surface to the patient end.

[0255] The stable interval determination unit 820 is used to determine the cumulative number of drops and the local drop frequency corresponding to each cumulative drop number based on the drop event timestamp sequence, and to determine the resistance stable interval based on the abrupt change position of the local drop frequency; the resistance stable interval is the interval of consecutive drop numbers corresponding to the same value of the equivalent conduction coefficient of the liquid path, and the equivalent conduction coefficient of the liquid path characterizes the equivalent conduction capability of the infusion liquid path for liquid flow.

[0256] The attenuation gradient determination unit 830 is used to determine the negative rate of change of the logarithm of the local drop frequency relative to the cumulative drop number as the drop frequency attenuation gradient corresponding to each of the cumulative drop numbers within each of the resistance stability intervals.

[0257] The margin constraint establishment unit 840 is used to combine the pressure-flow rate relationship and the volume conservation relationship of gravity infusion, eliminate the equivalent conductivity coefficient of the fluid path and establish a margin constraint relationship; the margin constraint relationship is used to constrain the correspondence between the drop frequency attenuation gradient and the liquid level height and liquid level cross-sectional area corresponding to the remaining volume under the conditions that the effective single drop volume, the pressure-flow rate index and the hydraulic baseline are determined.

[0258] The candidate surplus calculation unit 850 is used to calculate the candidate surplus corresponding to each cumulative drop number and satisfying the surplus constraint relationship within the range from zero to the initial effective volume, based on the drop frequency attenuation gradient corresponding to each cumulative drop number and the bottle geometry parameters.

[0259] The target balance determination unit 860 is used to determine a target balance from the candidate balances corresponding to each of the cumulative drops, so that each target balance decreases or remains unchanged as the cumulative drops increase, the target balance corresponding to the current cumulative drops is determined as the current balance, and the current balance is used as the dynamic prediction result of the infusion bottle balance.

[0260] In some embodiments, this application provides a non-volatile computer-readable storage medium storing one or more programs including execution instructions. The execution instructions can be read and executed by electronic devices (including but not limited to computers, servers, or network devices) to perform the steps of any of the above-described methods for dynamically predicting the remaining volume of infusion bottles based on drip frequency attenuation gradients.

[0261] In some embodiments, this application also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of any of the above-described methods for dynamic prediction of infusion bottle remaining volume based on drip frequency attenuation gradient.

[0262] In some embodiments, this application also provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of a dynamic prediction method for the remaining volume of an infusion bottle based on a drip frequency attenuation gradient.

[0263] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0264] The electronic devices in this application can exist in various forms, including but not limited to: mobile communication devices, ultra-mobile personal computer devices, portable entertainment devices, or other airborne electronic devices with data interaction functions.

[0265] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0266] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0267] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for dynamically predicting the remaining volume of an infusion bottle based on the drip frequency decay gradient, characterized in that, The method includes: The drop event timestamp sequence, initial effective volume, bottle geometric parameters, effective single drop volume, pressure-flow index, and hydraulic baseline are obtained during the ventilated gravity infusion process. The bottle geometric parameters represent the correspondence between the remaining volume and the liquid level height and the liquid level cross-sectional area. The hydraulic baseline represents the equivalent hydraulic height from the infusion bottle outlet reference surface to the patient end. The cumulative number of drops and the local drop frequency corresponding to each cumulative number of drops are determined based on the timestamp sequence of the drop events, and the resistance stability interval is determined based on the abrupt change position of the local drop frequency; the resistance stability interval is the interval of consecutive drop numbers corresponding to the same value of the equivalent conduction coefficient of the liquid path, and the equivalent conduction coefficient of the liquid path characterizes the equivalent conduction capability of the infusion liquid path for liquid flow; Within each of the resistance stability intervals, the negative rate of change of the logarithm of the local drop frequency relative to the cumulative drop number is determined as the drop frequency attenuation gradient corresponding to each of the cumulative drop numbers; By combining the pressure-flow rate relationship and volume conservation relationship of gravity infusion, the equivalent conductivity coefficient of the fluid path is eliminated and a margin constraint relationship is established; the margin constraint relationship is used to constrain the correspondence between the drop frequency decay gradient and the liquid level height and liquid surface cross-sectional area corresponding to the remaining volume under the conditions that the effective single drop volume, the pressure-flow rate index and the hydraulic baseline are determined. Based on the drop frequency attenuation gradient corresponding to each cumulative drop number and the bottle geometry parameters, candidate margins corresponding to each cumulative drop number and satisfying the margin constraint relationship are solved within the range of zero to the initial effective volume. A target balance is determined from the candidate balances corresponding to each cumulative drop count, and each target balance decreases or remains unchanged as the cumulative drop count increases. The target balance corresponding to the current cumulative drop count is determined as the current balance, and the current balance is used as the dynamic prediction result of the infusion bottle balance.

2. The method according to claim 1, characterized in that, The step of determining the cumulative number of drops and the local drop frequency corresponding to each cumulative drop based on the drop event timestamp sequence, and determining the resistance stability interval based on the abrupt change position of the local drop frequency, includes: The drop interval is determined based on the time difference of the timestamps of adjacent drop events. Adjacent drop events with a drop interval less than the preset minimum drop interval are merged into a single valid drop event to obtain a valid drop event sequence. Within the sliding drop count window of the effective drop event sequence, a reference drop interval is determined, and the preceding reference drop frequency and the following reference drop frequency are determined according to the sliding drop count windows before and after the drop interval to be corrected, respectively; for the drop interval to be corrected that is greater than the reference drop interval, the integer closest to and greater than 1 in the ratio of the drop interval to be corrected to the reference drop interval is determined as the target integer multiple, and the absolute value of the difference between the drop interval to be corrected and the product of the target integer multiple and the reference drop interval is determined as the time matching deviation; When the time matching deviation does not exceed the preset time tolerance, and the absolute value of the difference between the previous reference drop frequency and the subsequent reference drop frequency does not exceed the preset continuity threshold, the number obtained by subtracting one from the target integer multiple is determined as the number of missed drops, and the same number of virtual counter bits are added between the two corresponding valid drop events; the virtual counter bits are only used to correct the cumulative drop count and do not have drop event timestamps; When the interval between drops to be corrected exceeds a preset pause threshold, and the absolute value of the time matching deviation or the difference between the previous reference drop frequency and the subsequent reference drop frequency exceeds the corresponding threshold, the corresponding drop number range is determined as an abnormal pause window. The cumulative drop count is determined based on the order of occurrence of the valid drop events and the virtual counter bits, and the local drop frequency corresponding to each cumulative drop count is determined based on the timestamp of the valid drop events and the corrected cumulative drop count. When the absolute value of the logarithm of the ratio of the median of the local drop frequencies before and after the position to be determined exceeds the preset frequency jump threshold, the position to be determined is determined as the drop frequency segment boundary, and the continuous drop count interval between adjacent drop frequency segment boundaries after excluding the abnormal pause window is determined as the resistance stability interval.

3. The method according to claim 2, characterized in that, Within each of the resistance stability intervals, determining the negative rate of change of the logarithm of the local drop frequency relative to the cumulative drop number as the drop frequency attenuation gradient corresponding to each of the cumulative drop numbers includes: Based on the ratio of the local drop frequency corresponding to each sliding drop number window to the preset positive reference frequency, the logarithmic drop frequency sequence arranged along the cumulative drop number is determined; Robust smoothing of the logarithmic drop frequency sequence is performed within each of the resistance stability intervals to reduce the impact of logarithmic drop frequencies with fitting deviations exceeding a preset deviation threshold on the smoothing result, limit the second-order change between adjacent cumulative drop number positions, and constrain the smoothing result to decrease or remain unchanged as the cumulative drop number increases, thereby obtaining a smoothed logarithmic drop frequency sequence. The negative rate of change of the smoothed logarithmic drop frequency sequence relative to the cumulative drop number is determined as the drop frequency attenuation gradient corresponding to each cumulative drop number; The pressure-flow rate relationship and volume conservation relationship of gravity infusion are combined, the equivalent conductivity coefficient of the fluid path is eliminated, and a margin constraint relationship is established, including: The sum of the hydraulic baseline and the liquid level corresponding to the remaining volume is determined as the total equivalent static head, and a margin constraint relationship is established such that the product of the drop frequency attenuation gradient, the liquid surface cross-sectional area corresponding to the remaining volume, and the total equivalent static head is equal to the product of the pressure-flow index and the effective single drop volume.

4. The method according to claim 3, characterized in that, The method further includes performing a self-calibration operation on the effective single drop volume, the pressure-flow index, and the hydraulic baseline based on the margin constraint relationship and the bottle geometry parameters; the self-calibration operation includes: Any of the aforementioned resistance stability intervals is defined as calibration intervals; for each candidate single drop volume within the preset single drop volume range, based on the target margin at the calibration reference position and its corresponding cumulative number of drops, the number of new drops after the calibration reference position and the volume consumption determined by the corresponding candidate single drop volume are deducted to obtain the candidate remaining volume; Based on the bottle geometry parameters, determine the rate of change of the liquid surface cross-sectional area relative to the remaining volume for each candidate remaining volume. Build a set of calibrated drops when the absolute value of the rate of change reaches a preset area change threshold. Then, determine the candidate single drop volume corresponding to the number of cumulative drops in the calibrated drop volume set that reaches a preset minimum calibrated drop volume as the effective candidate single drop volume. Each effective candidate single drop volume is combined with each candidate pressure-flow index within the preset pressure-flow index range to form a candidate parameter combination; for each candidate parameter combination, based on the candidate remaining volume, drop frequency attenuation gradient, liquid surface cross-sectional area and liquid surface height of each cumulative drop number in the corresponding calibrated drop number set, the candidate hydraulic baseline corresponding to each cumulative drop number is determined by the margin constraint relationship. When the number of effective candidate droplet volumes is greater than zero, the dispersion of the candidate hydraulic baselines corresponding to each candidate parameter combination within the same calibration interval is determined. The effective candidate droplet volume and the candidate pressure-flow index in the candidate parameter combination with the smallest dispersion are determined as the effective droplet volume and the pressure-flow index, respectively, and the median of the corresponding candidate hydraulic baselines is determined as the hydraulic baseline. When the number of effective candidate droplet volumes is zero, the already obtained effective droplet volume, pressure-flow index, and hydraulic baseline are used.

5. The method according to claim 4, characterized in that, After determining the cumulative number of drops and the local drop frequency corresponding to each cumulative drop based on the drop event timestamp sequence, and determining the resistance stability interval based on the abrupt change position of the local drop frequency, the method further includes: The local drop frequency median and the drop frequency attenuation gradient median greater than zero are determined based on the stable drop number windows before and after the mutation position. The drop frequency transition amount and gradient transition amount are determined based on the logarithm of the ratio of the corresponding median after the mutation to that before the mutation. The absolute value of the gradient transition amount does not exceed a preset gradient continuity threshold, the absolute value of the sum of the drop frequency transition amount and the gradient transition amount adjusted by the pressure-flow index does not exceed a preset hydraulic matching threshold, and the absolute value of the sum of the drop frequency transition amount and the gradient transition amount does not exceed a preset drop volume matching threshold within a preset number of consecutive drop windows are sequentially determined as matching conditions for fluid resistance change, hydraulic baseline change and single drop volume change. For abrupt changes that persist after the preset recovery window has ended, if only one change matching condition is met, the event type is determined based on the condition. If it is determined to be a hydraulic resistance change event, an updated resistance stability interval is formed. If it is determined to be a hydraulic baseline change event, the target margin before and after the event is kept continuous with the nearest target margin before the event, and the hydraulic baseline is updated according to the inverse proportional relationship between the total equivalent static head and the drip frequency attenuation gradient. If it is determined to be a single drop volume change event, the effective single drop volume is updated according to the same proportional relationship between the effective single drop volume and the drip frequency attenuation gradient. The updated resistance stability interval, hydraulic baseline, or effective single drop volume is used for subsequent candidate margin solutions. When the local drop frequency and the drop frequency attenuation gradient are respectively restored to a preset tolerance range based on the median before the corresponding event within the preset recovery window, or when the number of valid change matching conditions is zero or greater than 1, the corresponding drop number interval is determined as a transient abnormal window, and the update of the target margin based on the drop frequency attenuation gradient within the transient abnormal window is paused.

6. The method according to claim 5, characterized in that, The step of determining a target surplus from the candidate surpluses corresponding to each of the cumulative drop counts includes: Obtain the previously successfully determined target balance and its corresponding cumulative drop count; determine the new drop count based on the difference between the current cumulative drop count and the corresponding cumulative drop count; determine the physical consumption range expanded by the preset volume tolerance based on the new drop count and the preset single drop volume range; and determine the balance search interval from zero to the previously successfully determined target balance range based on the physical consumption range. Based on the drop frequency attenuation gradient corresponding to the current cumulative drop count and the currently applicable effective single drop volume, pressure-flow index, and hydraulic baseline, determine the residual constraint of each remaining quantity to be determined within the remaining quantity search interval under the remaining quantity constraint relationship; The remaining amount to be determined is determined as a candidate remaining amount if the absolute value of the remaining amount constraint residual does not exceed the preset residual threshold, can be connected with the previously successfully determined target remaining amount along the same continuous solution branch formed by the change of the remaining amount constraint relationship with the cumulative number of drops, and the remaining volume on the continuous solution branch decreases or remains unchanged with the increase of the cumulative number of drops. When the number of candidate reserves is greater than zero, a drop count conservation reference reserve is determined based on the previously successfully determined target reserve, the newly added drop count, and the currently applicable effective single drop volume. The candidate reserve with the smallest absolute value of the difference between the drop count conservation reference reserve and the target reserve is determined as the target reserve. The target reserve is determined as the current reserve, and the current reserve is used as the dynamic prediction result of the infusion bottle reserve. When the number of candidate reserves is zero, an inversion mismatch flag is generated and the infusion status is reviewed. Based on the infusion status review results, the currently applicable resistance stability range, effective single drop volume, and hydraulic baseline are determined, and the drop count conservation reference reserve and candidate reserves are re-determined. When the number of re-determined candidate reserves is greater than zero, the candidate reserve with the smallest absolute value of the difference between it and the re-determined drop count conservation reference reserve is determined as the target reserve, and the target reserve is determined as the current reserve. The inversion mismatch flag is removed, and the dynamic prediction result of the infusion bottle reserve is updated. When the number of re-determined candidate reserves is zero, the inversion mismatch flag is retained, and the most recent dynamic prediction result of the infusion bottle reserve is marked as paused.

7. The method according to claim 6, characterized in that, After using the current remaining amount as the dynamic prediction result of the infusion bottle remaining amount, the method further includes: Based on the drop frequency attenuation gradient corresponding to the current cumulative number of drops, the bottle's geometric parameters, the effective single drop volume, the pressure-flow index, and the error propagation of the hydraulic baseline, determine the standard uncertainty of the margin constraint residual; The margin inversion sensitivity is determined based on the absolute value of the rate of change of the margin constraint relationship relative to the remaining volume at the current margin. When the margin inversion sensitivity exceeds the preset minimum inversion sensitivity threshold, the margin standard uncertainty is determined according to the ratio of the margin constraint residual standard uncertainty to the margin inversion sensitivity, and the margin confidence interval limited to zero to the initial effective volume is determined in combination with the current margin and the preset confidence coefficient. When the margin confidence interval is formed, the equivalent conduction coefficient of the fluid path is determined based on the local drip frequency within the current resistance stability interval, the currently applicable effective single drop volume, the pressure-flow index, the hydraulic baseline, and the current margin. When the fluid path equivalent conduction coefficient remains unchanged, the infusion flow rate is determined based on the change in the total equivalent static head as the remaining volume decreases and the pressure-flow relationship. The infusion time consumed from the two boundaries of the margin confidence interval to zero margin is accumulated to obtain the remaining infusion time interval. When the remaining amount confidence interval and the remaining infusion time interval are formed, a low remaining amount warning is generated when the upper limit of the remaining amount confidence interval does not exceed the preset safe remaining amount threshold, and a time limit warning is generated when the upper limit of the remaining infusion time interval does not exceed the sum of the preset nursing response time and the preset bottle changing preparation time. When the remaining balance inversion sensitivity does not exceed the preset minimum inversion sensitivity threshold, the inversion mismatch flag has been generated and not removed, no valid drip event is obtained for a continuous preset duration, or the width of the formed remaining balance confidence interval exceeds the preset confidence width threshold, the dynamic prediction result of the infusion bottle remaining balance is marked as paused and an insufficient observation prompt is generated.

8. A dynamic prediction system for the remaining volume of an infusion bottle based on the drip frequency decay gradient, characterized in that, The system includes: The basic data acquisition unit is used to acquire the drop event timestamp sequence, initial effective volume, bottle geometric parameters, effective single drop volume, pressure-flow index, and hydraulic baseline during the ventilated gravity infusion process; the bottle geometric parameters represent the correspondence between the remaining volume and the liquid level height and the liquid level cross-sectional area, and the hydraulic baseline represents the equivalent hydraulic height from the infusion bottle outlet reference surface to the patient end; The stable interval determination unit is used to determine the cumulative number of drops and the local drop frequency corresponding to each cumulative drop number based on the drop event timestamp sequence, and to determine the resistance stable interval based on the abrupt change position of the local drop frequency; the resistance stable interval is the interval of consecutive drop numbers corresponding to the same value of the equivalent conduction coefficient of the liquid path, and the equivalent conduction coefficient of the liquid path characterizes the equivalent conduction capability of the infusion liquid path for liquid flow; The attenuation gradient determination unit is used to determine the negative rate of change of the logarithm of the local drop frequency relative to the cumulative drop number as the drop frequency attenuation gradient corresponding to each cumulative drop number within each resistance stability interval. The margin constraint establishment unit is used to establish the pressure-flow rate relationship and volume conservation relationship of gravity infusion, eliminate the equivalent conductivity coefficient of the fluid path, and establish the margin constraint relationship. The margin constraint relationship is used to constrain the correspondence between the drop frequency decay gradient and the liquid level height and liquid level cross-sectional area corresponding to the remaining volume under the conditions of the effective single drop volume, the pressure-flow rate index, and the hydraulic baseline. The candidate surplus calculation unit is used to calculate the candidate surplus corresponding to each cumulative drop number and satisfying the surplus constraint relationship within the range of zero to the initial effective volume, based on the drop frequency attenuation gradient corresponding to each cumulative drop number and the bottle geometric parameters. The target balance determination unit is used to determine a target balance from the candidate balances corresponding to each of the cumulative drops, such that each target balance decreases or remains unchanged as the cumulative drops increase, the target balance corresponding to the current cumulative drops is determined as the current balance, and the current balance is used as the dynamic prediction result of the infusion bottle balance.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor, characterized in that the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the method as described in any one of claims 1-7.

10. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-7.

Images