Cell growth state recognition system

Abstract

The invention relates to the technical field of bioinformatics, in particular to a cell growth state recognition system, which comprises a growth data time sequence construction module for acquiring original cell population growth data under different drug disturbance conditions, sequentially arranging timestamps and cell observation values in the original cell population growth data, and constructing a cell growth state recognition module; and converting to obtain a cell growth curve calibration result. The method comprises the following steps of: sequencing timestamps and cell observation values of original cell population growth data, establishing a cell growth curve calibration result, further extracting an instantaneous growth rate sequence in a continuous time interval on the basis, positioning the maximum cell growth rate, and capturing a microscopic change trend in a cell dynamic proliferation process; and quantifying a cell saturation density result by using the maximum cell growth rate and the curve end point trend, thereby reflecting the real state of the cell growth in the saturation stable stage.

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics, and in particular to a cell growth state recognition system. Background Art

[0002] Bioinformatics is an interdisciplinary subject that uses a variety of methods and tools such as computer science, mathematics, statistics and biology to collect, store, analyze, visualize and interpret biological data in order to reveal biological laws and phenomena in the field of life sciences.

[0003] Existing technologies struggle to accurately distinguish subtle differences in cell populations under different drug conditions, limiting the effectiveness of sensitivity identification and leading to ambiguity in determining cellular drug response patterns. For example, in drug screening experiments, response characteristics under multiple drug treatment conditions are often difficult to clearly distinguish, thus affecting the accuracy of efficacy assessments and the reliability of experimental results. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a cell growth status identification system.

[0005] In order to achieve the above-mentioned object, the present invention adopts the following technical solution: A cell growth state recognition system includes:

[0006] The growth data time series construction module obtains the original cell population growth data under different drug perturbation conditions, sequentially arranges the timestamps and cell observation values ​​in the original cell population growth data, and converts them to obtain the cell growth curve calibration results;

[0007] A core rate parameter extraction module calculates, based on the cell growth curve calibration result, the numerical change of the cell growth curve calibration result at each continuous time interval to obtain an instantaneous growth rate sequence characterizing the speed of cell proliferation, and determines the peak value based on the instantaneous growth rate sequence to obtain the maximum cell growth rate result;

[0008] a saturation state parameter quantification module, which analyzes the growth dynamics of the curve under the target rate condition until the saturation state based on the cell growth curve calibration result and the maximum cell growth rate result, determines the growth stable stage, obtains the endpoint trend of the growth curve, and calculates the cell saturation density result based on the endpoint trend of the growth curve;

[0009] The cell state classification module integrates the maximum cell growth rate result and the cell saturation density result, constructs a vector as the cell growth sensitivity fingerprint of the current disturbance condition, generates a two-parameter growth feature vector, compares and judges the two-parameter growth feature vector with the statistical demarcation standard, identifies the characteristic pattern of cell response to drug disturbance, and outputs the cell growth sensitivity judgment result.

[0010] Preferably, the steps for obtaining the cell growth curve calibration result are:

[0011] For the original cell population growth data under different drug perturbation conditions, the acquisition timestamp and cell observation value corresponding to each growth data are obtained one by one, and a one-to-one mapping relationship between each set of timestamps and the corresponding cell observation value is established to obtain the pairing results of timestamps and cell observation values;

[0012] Based on the pairing results of the timestamps and the cell observation values, the pairing results are sequentially rearranged one by one according to the acquisition timestamps from small to large, and during the rearrangement process, the average of multiple cell observation values ​​corresponding to the same timestamp is calculated as a unique observation value to obtain a rearranged timestamp and cell observation value sequence;

[0013] Based on the rearranged timestamps and cell observation value sequence, growth curve point connection is performed time stamp by time stamp, and adjacent cell observation value points in the sequence are continuously connected in the order of timestamps to form a cell growth curve calibration result.

[0014] Preferably, the steps of obtaining the instantaneous growth rate sequence are:

[0015] Based on the cell growth curve calibration result, three consecutive observation points corresponding to time stamps are extracted to construct a time stamp sequence window. In each time stamp sequence window, the cell observation value corresponding to the time stamp is extracted to form a cell observation value sequence and a time interval sequence.

[0016] Calculating the instantaneous growth rate index corresponding to each timestamp sequence window according to the cell observation value sequence and the time interval sequence;

[0017] Based on the instantaneous growth rate index, the instantaneous growth rate indexes corresponding to all time stamp sequence windows are arranged and summarized one by one in chronological order to generate an instantaneous growth rate sequence.

[0018] Preferably, the steps for obtaining the maximum cell growth rate result are:

[0019] Based on the instantaneous growth rate sequence, each instantaneous growth rate index in the instantaneous growth rate sequence is extracted one by one, and according to the time stamp sequence corresponding to each instantaneous growth rate index, the size of adjacent instantaneous growth rate indexes is compared one by one to determine the change trend of the instantaneous growth rate index, and obtain an index change trend sequence;

[0020] According to the exponential change trend sequence, sequentially identifying the inflection points where the trend changes from rising to falling, and extracting the instantaneous growth rate indices of all corresponding inflection points from the exponential change trend sequence to form a growth rate peak candidate set;

[0021] Based on the growth rate peak candidate set, all instantaneous growth rate indices are compared one by one and the maximum instantaneous growth rate index is determined and designated as the maximum cell growth rate result.

[0022] Preferably, the step of obtaining the endpoint trend of the growth curve is:

[0023] Based on the cell growth curve calibration result and the maximum cell growth rate result, extracting the corresponding three consecutive cell observation values ​​and the corresponding three consecutive time values ​​in each time window, and taking the maximum cell growth rate result as the rate benchmark to form a time window observation parameter set;

[0024] Calculating the curve growth dynamic offset intensity of the window according to the time window observation parameter set;

[0025] Based on the dynamic offset intensity of the curve growth corresponding to each time window, a continuous sliding window analysis is performed on the entire growth curve, and the time period in which the dynamic offset intensity of the curve growth in all continuous time windows is lower than the stability judgment threshold is marked as the growth stable stage. The growth direction trend of the observation value at the end of the stable stage is calculated to obtain the end point trend of the growth curve.

[0026] Preferably, the steps for obtaining the cell saturation density result are:

[0027] Based on the endpoint trend of the growth curve, the last three time points in the endpoint trend section are extracted and the temperature mean is calculated to form an endpoint trend stable observation group;

[0028] Calculate the estimated cell saturation density based on the endpoint trend stability observation group;

[0029] Based on the estimated cell saturation density, the distribution range of the estimated cell saturation density of the same cell type in the same culture environment is compared with the historical records. If the estimated cell saturation density is within ±5% of the upper limit of the range, it is determined as the cell saturation density result.

[0030] Preferably, the steps for obtaining the dual-parameter growth feature vector are:

[0031] Integrating the maximum cell growth rate result and the cell saturation density result, taking the maximum cell growth rate result as the first component of the feature vector and taking the cell saturation density result as the second component of the feature vector, to obtain an initial dual-parameter set;

[0032] According to the initial two-parameter set, a unit check is performed on the maximum cell growth rate result of the first component and the cell saturation density result of the second component, the two components are converted and normalized, and the two components are recombined in the order of the maximum cell growth rate result first and the cell saturation density result second to form a dimensionless two-parameter feature set;

[0033] Based on the dimensionless two-parameter feature set, the dimensionless maximum cell growth rate result and the dimensionless cell saturation density result are set as the horizontal and vertical coordinate values ​​of the feature vector, respectively, and the two values ​​are assembled in the form of a feature vector to generate a two-parameter growth feature vector.

[0034] Preferably, the steps for obtaining the cell growth sensitivity determination result are:

[0035] Based on the dual-parameter growth eigenvector, the maximum cell growth rate result and the cell saturation density result in the vector are respectively extracted, and compared item by item with the statistical demarcation standard values ​​under the corresponding drug perturbation conditions in the existing cell sensitivity database, and the degree of deviation from the standard value is calculated to obtain the eigenvector deviation degree result;

[0036] According to the result of the deviation degree of the characteristic vector, the deviation degree of the maximum cell growth rate result and the cell saturation density result are respectively compared with the allowable deviation threshold value preset in the statistical demarcation standard. When the numerical deviation degree of the two components is less than the allowable deviation threshold value, the characteristic pattern is judged as the cell having low sensitivity to drug perturbation; otherwise, it is judged as high sensitivity, forming an initial judgment state of cell sensitivity;

[0037] Based on the initial cell sensitivity determination state, combined with the consistency check of the cell sensitivity pattern under the same conditions in the current disturbance conditions and historical data statistics, the current initial cell sensitivity determination state is verified by comparing the historical results, and the cell growth sensitivity determination result is output.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are:

[0039] In the present invention, the original cell population growth data is timestamped and sorted in order of cell observation values ​​to establish a cell growth curve calibration result. On this basis, the instantaneous growth rate sequence within continuous time intervals is further extracted, and the maximum cell growth rate is located to capture the microscopic change trend during the dynamic cell proliferation process; the cell saturation density result is then quantified using the maximum cell growth rate and the curve endpoint trend, thereby reflecting the true state of cell growth when it reaches the saturated stable stage; in addition, the maximum cell growth rate result and the cell saturation density result are integrated to construct a dual-parameter growth feature vector to achieve the characterization of cell growth sensitivity, and compared with the statistical demarcation standard to achieve accurate identification of cell sensitivity to drug perturbations, so that the cell state judgment is transformed from a single observation indicator to a multi-dimensional quantitative analysis, thereby improving the resolution precision and accuracy of the drug response characteristic pattern, and providing reliable data support for biopharmaceutical screening and cell behavior prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] See also Figure 1 The present invention provides a technical solution: a cell growth state identification system comprising:

[0043] The growth data time series construction module obtains the original cell population growth data under different drug perturbation conditions, sequentially arranges the timestamps and cell observation values ​​in the original cell population growth data, and converts them to obtain the cell growth curve calibration results;

[0044] The core rate parameter extraction module calculates the numerical change of the cell growth curve calibration result at each continuous time interval based on the cell growth curve calibration result, obtains the instantaneous growth rate sequence that characterizes the speed of cell proliferation, and determines the peak value based on the instantaneous growth rate sequence to obtain the maximum cell growth rate result;

[0045] The saturation parameter quantification module, based on the cell growth curve calibration results and the maximum cell growth rate results, analyzes the growth dynamics of the curve under the target rate conditions until the saturation state, determines the growth stability stage, obtains the endpoint trend of the growth curve, and calculates the cell saturation density result based on the endpoint trend of the growth curve;

[0046] The cell state classification module integrates the maximum cell growth rate results and the cell saturation density results, constructs a vector as the cell growth sensitivity fingerprint of the current disturbance conditions, generates a two-parameter growth feature vector, compares and judges based on the two-parameter growth feature vector with the statistical demarcation standard, identifies the characteristic pattern of cell response to drug disturbance, and outputs the cell growth sensitivity judgment result.

[0047] The steps for obtaining the cell growth curve calibration results are as follows:

[0048] For the original cell population growth data under different drug perturbation conditions, the acquisition timestamp and cell observation value corresponding to each growth data are obtained one by one, and a one-to-one mapping relationship between each set of timestamps and the corresponding cell observation value is established to obtain the pairing results of timestamps and cell observation values;

[0049] Based on the pairing results of timestamps and cell observation values, the pairing results are rearranged one by one according to the acquisition timestamp from small to large, and the average value of multiple cell observation values ​​corresponding to the same timestamp is calculated as the unique observation value during the rearrangement process to obtain the rearranged timestamp and cell observation value sequence;

[0050] Based on the rearranged timestamps and cell observation value sequence, the growth curve points are connected time stamp by time stamp, and the adjacent cell observation value points in the sequence are continuously connected in the order of the timestamps to form the cell growth curve calibration result.

[0051] Specifically, when processing the original cell population growth data collected from different drug perturbation experiments, the system first accesses these data, which usually contain multiple data points. Each data point records the cell state at a specific time point, such as the cell number obtained by microscope image analysis or the fluorescence intensity value obtained by fluorescence detection. First, it is necessary to clarify the source of the data. For example, the data may come from a microplate reader, where each well represents a drug perturbation condition or concentration, and the absorbance or fluorescence value is automatically recorded at preset time intervals. Or it may come from a flow cytometer, which records parameters such as cell counts and activity sampled at specific time points. For each drug perturbation condition, such as drug A with a concentration of 1 micromolar and drug B with a concentration of 5 micromolar, the corresponding original data records are parsed one by one, and two core information are accurately extracted from each record: one is the specific moment when the cell observation event occurs, that is, the acquisition timestamp The timestamp is usually recorded in the format of year-month-day hour:minute:second to ensure time accuracy. The other is the cell observation value corresponding to the timestamp, such as cell density (unit: number of cells / ml) or absorbance value (OD value). The data under each set of drug perturbation conditions are operated independently, and a direct, paired association is created for each acquisition timestamp in the group and its corresponding cell observation value. For example, if under the condition of drug X, time point T1 corresponds to observation value C1, and time point T2 corresponds to observation value C2, then pairs such as (T1, C1) and (T2, C2) are formed. This process does not involve screening or filtering of the data and is only a faithful record of the original correspondence. By establishing this one-to-one correspondence for all timestamps and cell observation values ​​under each drug perturbation condition, all these pairs are finally summarized to form a structured timestamp and cell observation value pairing result.

[0052] Based on the timestamp and cell observation value pairing results obtained in the previous step, the result is a series of drug perturbation conditions, each perturbation condition contains multiple (timestamp, cell observation value) pairing sets in disorder or arranged in the order of acquisition device output. Next, these pairing results are systematically sorted and normalized. The core operation is to rearrange them according to the order of acquisition timestamps, and arrange the timestamp and cell observation value pairing results under each drug perturbation condition in ascending order from the earliest time point to the latest time point recorded. For example, if the pairing results contain (10:05, 150), (10:00, 100), (10:10, 200), they are rearranged to (10:00, 100), (10:05, 150), (10:10, 200). During the rearrangement process, the system will specially handle a situation where, under the same drug perturbation condition, there may be records of multiple cell observation values ​​corresponding to exactly the same timestamp, which can be This may be caused by the instrument taking multiple readings or merging data in a short period of time. In this case, a strategy of calculating the arithmetic mean is adopted. The multiple cell observation values ​​at the same timestamp are summed up and then divided by the number of observation values ​​to obtain a single, representative cell observation value. The average value will be used as the unique cell observation value at this timestamp. For example, if cell observation values ​​V1, V2, and V3 are recorded at the timestamp T_A at the same time, the unique observation value corresponding to the timestamp T_A will be updated to (V1+V2+V3) / 3. This averaging process ensures that each timestamp appears only once in the sequence and corresponds to a specific cell observation value. By performing the above-mentioned time sorting and repeated timestamp data averaging operations on the results of pairing the timestamps and cell observation values ​​under all drug perturbation conditions, a set of one or more groups (depending on the number of drug perturbation conditions) of rearranged timestamps and cell observation value sequences with strict order and unique time points are obtained.

[0053] Based on the rearranged timestamps and cell observation value sequences obtained in the previous processing, the sequence provides a list of chronologically arranged cell observation values ​​for each drug perturbation condition, and each timestamp corresponds to a unique cell observation value. For example, for a certain drug perturbation condition, the sequence is (T1, C1), (T2, C2),…, (T n ,C n ), where T1 <T2<…<T n , the system then performs a time stamp-by-time stamp growth curve point connection operation. Specifically, it conceptually connects each adjacent cell observation value point in the sequence according to the natural order of the timestamps. This means that the cell observation value C at the i-th time point is connected to i and the cell observation value C at the i+1th time point immediately following it i+1The connection process starts from the first point of the sequence, connects to the second point, then connects to the third point, and so on, until the penultimate point in the sequence is connected to the last point. This connection method is manifested at the data level as maintaining the order of these (timestamp, cell observation value) pairs, so that they can be directly used for subsequent rate calculation or visualization, forming a broken line graph that intuitively reflects the growth trend of the cell population over time. The entire process does not involve smoothing or curve fitting of the observation values, but directly uses the original (or averaged) observation points for sequential connection, ensuring the original shape of the curve. By performing this continuous, pairwise connection of all points in the rearranged timestamp and cell observation value sequence in chronological order, a clear cell growth trajectory is finally constructed for each drug perturbation condition. This trajectory is the cell growth curve calibration result.

[0054] The steps for obtaining the instantaneous growth rate series are:

[0055] Based on the calibration results of the cell growth curve, the observation points corresponding to three consecutive timestamps are extracted to construct a timestamp sequence window. In each timestamp sequence window, the cell observation values ​​corresponding to the timestamps are extracted to form a cell observation value sequence and a time interval sequence.

[0056] According to the cell observation value sequence and time interval sequence, the instantaneous growth rate index corresponding to each timestamp sequence window is calculated. The calculation formula is:

[0057]

[0058] Among them, R k is the instantaneous growth rate index corresponding to the kth timestamp sequence window, C k is the cell observation value corresponding to the kth timestamp, C k+1 is the cell observation value corresponding to the k+1th timestamp, C k-1 is the cell observation value corresponding to the k-1th timestamp, T k is the time value of the kth timestamp, T k+1 is the time value of the k+1th timestamp, C ref It is the maximum cell observation value extracted in the last seven days under the current disturbance conditions;

[0059] Based on the instantaneous growth rate index, the instantaneous growth rate indexes corresponding to all timestamp sequence windows are arranged and summarized one by one in chronological order to generate an instantaneous growth rate sequence.

[0060] Specifically, based on the cell growth curve calibration result obtained in the previous step, the result provides a time-ordered (time stamp, cell observation value) point set for the cell growth data under each drug perturbation condition. The system then extracts data fragments from it by sliding windows. The specific operation is that for a given cell growth curve, starting from the starting point of the curve, three consecutive data points (i.e., three consecutive time stamps and their corresponding cell observation values) are selected in sequence to form a "time stamp sequence window". For example, if the cell growth curve calibration result contains the point (T1, C1), (T2, C2), (T3, C3), (T4, C4), ..., then the first timestamp sequence window is composed of (T1, C1), (T2, C2), (T3, C3), the second window is composed of (T2, C2), (T3, C3), (T4, C4), and so on, until three consecutive points can no longer be selected at the end of the curve. Within each constructed timestamp sequence window, the system further parses out two types of core sequences: one is the cell observation value sequence, that is, directly extracting the cell observation values ​​corresponding to the three time points in the window, such as for the window (T k-1 ,C k-1 ),(T k ,C k ),(T k+1 ,C k+1 ), the cell observation value sequence is (C k-1 ,C k ,C k+1 ), the other type is the time interval series, which is obtained by calculating the time difference between consecutive timestamps in the window. For example, in the above window, the time interval mainly focuses on T k+1 -T k and T k -T k-1 , these two intervals will be used in subsequent calculations. In this way, the original growth curve is decomposed into a series of overlapping small fragments containing local growth dynamic information, thereby forming the basic data structure required for subsequent instantaneous growth rate calculations, that is, each window is associated with a set of cell observation value sequences and corresponding time interval information, and finally multiple cell observation value sequences and time interval sequences are obtained.

[0061] formula: The benefit of the formula is that it comprehensively evaluates the growth activity of cells at a specific time point by combining the first-order change of cell growth (a direct reflection of the proliferation rate) and the second-order change (the trend of the proliferation rate change, i.e., the acceleration), and uses C refNormalization processing is performed to make the growth rates under different experimental conditions or different cell density levels comparable. At the same time, the square root of the sum of squares combines the influence of the two change components and can more sensitively capture the fluctuations and turning points of the growth rate.

[0062] C k The parameter acquisition step is: the parameter represents the kth timestamp T k The number of cell populations or its alternative indicators (such as absorbance values) observed at the time of T is directly derived from the "cell growth curve calibration results" generated in the previous step. This result is time series data after time sorting and repeated value averaging. For example, in a cell culture experiment, a microplate reader is used to automatically detect the absorbance (OD value) of cells in the culture wells every 2 hours. After data sorting, the absorbance (OD value) at T is obtained. k = At this time point of 24 hours, the corresponding cell observation value C k =0.58 (OD value).

[0063] C k+1 The parameter acquisition step is: the parameter represents the k+1th timestamp T k+1 The number of cell populations or their alternative indicators observed at T k At a subsequent observation point, for example, in the above-mentioned microplate reader test, if T k = 24 hours, then T k+1 = 26 hours (for example, the observation interval is 2 hours), the corresponding cell observation value C k+1 =0.65 (OD value).

[0064] C k-1 The parameter acquisition step is: the parameter represents the time at the k-1th timestamp T k-1 The number of cell populations or their alternative indicators observed at T k A previous observation point, for example, in the above-mentioned microplate reader test, if T k = 24 hours, then T k-1 = 22 hours, the corresponding cell observation value C k-1 =0.49 (OD value).

[0065] T k The parameter acquisition steps are as follows: the parameter represents the time value of the kth timestamp, the unit is usually hours or days, and is directly extracted from the "cell growth curve calibration result". The timestamp is the cell observation value C k The exact time recorded, for example, T k =24.0 hours.

[0066] T k+1 The parameter acquisition step is as follows: This parameter represents the time value of the k+1th timestamp, and the acquisition method is the same as T k , is the cell observation value C k+1 The exact time recorded, for example, T k+1 =26.0 hours.

[0067] C ref The steps for obtaining the parameter are as follows: the parameter is defined as the maximum cell observation value that can be observed in the last seven days (168 hours) before calculating the transient growth rate index under the current drug perturbation condition. The acquisition process is as follows: first, determine the current drug perturbation condition (for example, drug A, concentration 10μM), then retrieve the historical cell growth data records that are exactly the same as this condition, and the time range is limited to the current R k In the seven days before the calculation time point, find all the cell observation values ​​from these historical data and select the maximum value as C ref For example, a cell is cultured under a specific drug perturbation and its OD value is monitored. When calculating the instantaneous growth rate on the same day (for example, the 10th day of the experiment), all OD records under the drug perturbation condition from the 3rd day to the 9th day of the experiment (a total of seven days) are found. If the OD values ​​recorded during this period are 0.2, 0.4, 0.7, 1.1, 1.5, 1.3, and 1.2 respectively, then C ref is set to 1.5. If there is no data under exactly the same perturbation conditions in the last seven days, 80% of the maximum cell observation value achieved by the control group (e.g., no drug treatment group) under the current perturbation conditions during a similar culture period can be used as the reference benchmark. Alternatively, if the current experiment has lasted for more than seven days, C ref It can be from the start of the current experiment to T k-1 The maximum observation value in the seven-day window before the moment. In this example, it is assumed that C is obtained by querying historical data. ref =1.50 (OD value).

[0068] Calculation process:

[0069] Take the parameter values ​​obtained above as an example for calculation:

[0070] C k-1 =0.49 (OD value);

[0071] C k =0.58 (OD value);

[0072] C k+1 =0.65 (OD value);

[0073] T k = 24.0 hours;

[0074] T k+1 =26.0 hours;

[0075] C ref =1.50 (OD value);

[0076] Time interval ΔT = T k+1 -T k =26.0-24.0=2.0 hours.

[0077] Calculate the first part (the numerator of the first-order variation):

[0078] |C k+1 -C k |=|0.65-0.58|=|0.07|=0.07;

[0079] Calculate the second part (the numerator of the second-order variation):

[0080] |C k+1 -2C k +C k-1 |=|0.65-2×0.58+0.49|=|0.65-1.16+0.49|=|-0.02|=0.02;

[0081] Compute the square of the first term:

[0082]

[0083] Compute the square of the second term:

[0084]

[0085] Calculate R k :

[0086]

[0087] Therefore, the instantaneous growth rate index R corresponding to the kth timestamp sequence window is k It is approximately 0.024265.

[0088] The results show that at time point T k = Around 24 hours, the normalized transient growth rate index of the cell population is 0.024265. This value itself is a relative value, and its size reflects the activity of cell proliferation. The larger the value, the faster the cell growth or the more drastic the change. The smaller the value, the slower the growth or stagnant.

[0089] Based on a series of instantaneous growth rate indices R calculated in the previous step k , where each R kEach corresponds to a specific time point on the original cell growth curve (usually the midpoint T of the time stamp sequence window k or the starting point). Next, the system integrates these discrete instantaneous growth rate indices. Specifically, strictly according to the time stamps T k corresponding to these indices, arrange them one by one in sequence. This means that if the instantaneous growth rate index R1 corresponds to time T1, R2 corresponds to time T2, and T1 < T2, then the position of R1 in the sequence is prior to R2. This process ensures that the time continuity of the rate change is correctly reflected. Subsequently, the system aggregates the instantaneous growth rate indices calculated for all time stamp sequence windows to form a single, time-ordered sequence. For example, if the rate calculated for window 1 is R1, the rate for window 2 is R2, and so on until the rate for the last window is R m , and their corresponding time order is increasing, then the final formed sequence is (R1, R2,..., R m ). Each value in this sequence is an instantaneous growth rate index, jointly depicting the complete picture of how the cell growth rate changes over time during the entire observation period. This process does not involve numerical adjustment or smoothing of the instantaneous growth rate indices themselves, but is simply collection and organization based on time order. Through this arrangement and aggregation, a sequence of instantaneous growth rates is finally generated, which characterizes the full dynamic process of how fast the cells proliferate under different drug perturbation conditions.

[0090] The steps to obtain the maximum cell growth rate result are as follows:

[0091] Based on the instantaneous growth rate sequence, extract each instantaneous growth rate index in the instantaneous growth rate sequence one by one, and according to the time stamp order corresponding to each instantaneous growth rate index, compare the magnitudes of adjacent instantaneous growth rate indices one by one to determine the change trend of the instantaneous growth rate indices, and obtain the index change trend sequence;

[0092] According to the index change trend sequence, sequentially identify the inflection point positions where the trend changes from rising to falling, and extract all the instantaneous growth rate indices corresponding to the inflection point positions from the index change trend sequence to form a growth rate peak candidate set;

[0093] Based on the growth rate peak candidate set, compare all the instantaneous growth rate indices one by one and determine the largest instantaneous growth rate index, which is designated as the maximum cell growth rate result.

[0094] Specifically, based on the instantaneous growth rate sequence obtained in the previous step, this sequence is a series of instantaneous growth rate indices R k arranged in time order and their corresponding time stamps T kThe system first goes through each instantaneous growth rate index in the sequence one by one, starting from the first index R1 of the sequence to the last index R N In order to judge the changing trend of the index, the system compares the adjacent instantaneous growth rate indices in pairs according to the timestamp sequence corresponding to each instantaneous growth rate index. Specifically, for any instantaneous growth rate index R in the sequence i (where i ranges from 1 to N-1), and then compares it with the instantaneous growth rate index R i+1 Compare to determine the i to R i+1 The trend of change is determined by introducing a "relative change significance threshold" for this trend judgment. The threshold is set based on the inherent fluctuation amplitude of the instantaneous growth rate index in the control experiment or the stable stage of cell growth for statistical analysis. For example, the instantaneous growth rate sequence of the control group cell culture in the late stage without drug perturbation (the growth rate should be stable or slowly declining) is analyzed, and its coefficient of variation or the distribution of relative changes between adjacent points is calculated. If the relative change caused by 95% of the noise fluctuations is less than 3%, the relative change significance threshold can be set as 5%. The calculation method is: if (R i+1 -R i ) / R i >0.05, the trend is judged to be "upward"; if (R i -R i+1 ) / R i >0.05, the trend is judged to be "downward"; if both of the above two conditions are not met, that is, the absolute value of the relative change is less than or equal to 5%, the trend is judged to be "stable", for example, if R i =0.040 and R i+1 =0.043, then the relative change is (0.043-0.040) / 0.040=0.003 / 0.040=0.075 (i.e. 7.5%). Since 0.075>0.05, the trend is determined to be "rising". If R i =0.040 and R i+1 =0.038, then the relative change is (0.040-0.038) / 0.040=0.002 / 0.040=0.05, at this time the trend is judged as "stable" (according to the definition of less than or equal to 5% for stability, if strictly greater than it is a decline, here it is treated as stable according to insignificant change; or adjusted to if (R i -R i+1 ) / R i ≥0.05 is considered a decrease, so this case is a decrease, and the former definition is used here). For each pair of adjacent indices (R i ,R i+1 ) performs this judgment, so that for each interval [T i ,Ti+1 ]Generate a trend mark ("rising", "falling" or "stable"), which are arranged in chronological order to form an exponential change trend sequence.

[0095] According to the exponential change trend sequence obtained in the previous step, the sequence describes the change direction of the instantaneous growth rate between adjacent time points (for example, the sequence is ["increase", "increase", "stable", "decline", "decline"]), and combined with the original instantaneous growth rate sequence (including the specific instantaneous growth rate index value R k ), the system then identifies and extracts all potential growth rate peaks. A growth rate peak (i.e., local maximum) is defined as the inflection point where the trend changes from "upward" to "downward". The specific identification process is as follows: the system traverses the exponential change trend sequence and checks each trend marker TS i , which corresponds to the instantaneous growth rate index R i to R i+1 If there is a point R i Make its previous trend TS i-1 (i.e. from R i-1 to R i The trend of TS is "up" and the following trend is TS i (i.e. from R i to R i+1 The trend of the instantaneous growth rate index R is "downward". i It is identified as the growth rate value corresponding to an inflection point and is extracted. For example, if the instantaneous growth rate sequence is (R1, R2, R3, R4, R5) = (0.02, 0.05, 0.08, 0.06, 0.03), the corresponding exponential change trend sequence (describing R k →R k+1 The trend of the growth rate is TS1 ("rising"), TS2 ("rising"), TS3 ("falling"), and TS4 ("falling"). When the system detects that TS2 is "rising" and TS3 is "falling", it confirms that R3 = 0.08 is an inflection point, because it was an upward trend (from R2 to R3) before and a downward trend (from R3 to R4) after. Therefore, 0.08 is extracted. For the case where the trend contains "stationary", such as "rising" → "stationary" → "falling", the last "stationary" trend end point in the "stationary" segment (that is, the instantaneous growth rate index immediately before the start of the "falling" trend) is regarded as the inflection point. For example, if the sequence is R A ,R B ,R C ,R D The corresponding trend is TS A ("Rise"), TS B("stable"), TS C ("down"), then R C All the instantaneous growth rate indices identified and extracted in this way together constitute a set, which is the growth rate peak candidate set.

[0096] Based on the growth rate peak candidate set formed in the previous step, the set contains all the instantaneous growth rate indices corresponding to the local peak points identified in the instantaneous growth rate sequence according to the "rising" to "falling" trend rule (or the extended rule including "stable"). For example, the candidate set is {0.08, 0.075, 0.09} (these values ​​are the instantaneous growth rate indices). The system then needs to determine the unique maximum cell growth rate from these candidate values. The execution process is very straightforward: first, check whether the growth rate peak candidate set is empty. If it is empty, it indicates that no clear growth rate peak is detected during the entire observation period. At this time, the maximum cell growth rate result may be assigned a preset baseline value (such as 0) or marked as undetected. If If the candidate set is not empty, the system will traverse each instantaneous growth rate index in the set and compare them one by one. The specific method is to initialize a "current maximum value" variable and assign its value to the first element in the candidate set. Then, each subsequent instantaneous growth rate index in the growth rate peak candidate set is taken out in turn and compared with the "current maximum value". If the currently taken index is greater than the "current maximum value", the "current maximum value" is updated to this larger index. This process is repeated until all instantaneous growth rate indices in the candidate set have been compared. The final value held by the "current maximum value" variable is the largest of all candidate peaks. This final maximum instantaneous growth rate index is designated as the maximum cell growth rate result.

[0097] The steps to obtain the end point trend of the growth curve are:

[0098] Based on the cell growth curve calibration results and the maximum cell growth rate results, the corresponding three consecutive cell observation values ​​and the corresponding three consecutive time values ​​are extracted in each time window, and the maximum cell growth rate result is used as the rate benchmark to form a time window observation parameter set;

[0099] According to the time window observation parameter set, the curve growth dynamic offset intensity of the window is calculated. The calculation formula is:

[0100]

[0101] Among them, O wis the dynamic offset intensity of the curve growth in the time window, C1, C2, and C3 are the cell observation values ​​at the 1st, 2nd, and 3rd time points in the current time window, T1, T2, and T3 are the time values ​​at the corresponding time points, and R max The maximum cell growth rate result is used for rate normalization;

[0102] Based on the dynamic offset intensity of the curve growth corresponding to each time window, a continuous sliding window analysis was performed on the entire growth curve. The time period in which the dynamic offset intensity of the curve growth in all continuous time windows was lower than the stability judgment threshold was marked as the growth stable stage. The growth direction trend of the observation value at the end of the stable stage was calculated to obtain the end point trend of the growth curve.

[0103] Specifically, based on the cell growth curve calibration results obtained in the previous step, the results provide a series of chronologically arranged (timestamp, cell observation value) data points for the cell growth process under a specific drug perturbation, such as (T1, C1), (T2, C2), ..., (T N ,C N ), and combined with the determined maximum cell growth rate results R max The system uses a sliding window method to analyze the growth curve segment by segment. The specific operation is to define a time window containing three consecutive data points on the cell growth curve calibration result. Starting from the starting position of the curve, the i, i+1, and i+2 data points (where i traverses from 1 to N-2) are selected to form a window. For each such time window, the system extracts three consecutive cell observation values ​​within the window, which are recorded as C1=C i 、C2=C i+1 、C3=C i+2 , and extract the three continuous time values ​​corresponding to these three observation values, recorded as T1=T i 、T2=T i+1 、T3=T i+2 Then, the system retrieves the maximum cell growth rate result R calculated previously. max , the R max As a reference for the growth rate within this time window, these three pairs (time, observation value) and R max Combined together to form a network containing {C1,C2,C3,T1,T2,T3,R max This set of parameters is defined as the observation parameter set of the current time window. This process is repeated as the window slides from the beginning to the end along the entire growth curve, generating a corresponding time window observation parameter set for each possible three-point combination.

[0104] formula: The usefulness of the formula is that it quantifies the intensity of the dynamic changes in the cell growth curve within a specific time window, or the "excursion intensity". The numerator combines the square root of the sum of the squares of the first-order difference (approximate growth rate) and the second-order difference (approximate growth acceleration), reflecting the total fluctuation or curvature of the curve within the window. The denominator (T3-T1) normalizes the time span, and The maximum cell growth rate result R max Normalize the local growth rate of the current window. When the local growth rate is much smaller than R max When this term is close to 1, w Mainly reflects the normalized fluctuation; when the local growth rate is close to R max When , this term increases, thereby reducing O w value, which makes O w The growth rate is much lower than R max The stage with small fluctuation is more sensitive, showing smaller values, which helps to identify the stable growth period;

[0105] The steps for obtaining the C1 parameter are as follows: this parameter represents the cell observation value at the first time point in the current time window, and is directly extracted from the "time window observation parameter set" formed in the previous step. This parameter set comes from the "cell growth curve calibration result". For example, when analyzing a certain section of the growth curve, the selected window starting point is a specific time point, and the corresponding cell observation value (such as OD value) is C1 = 1.45.

[0106] The steps for obtaining the C2 parameter are as follows: this parameter represents the cell observation value at the second time point in the current time window, and is also extracted from the "time window observation parameter set". It is the observation value immediately following C1. For example, corresponding to the next observation time point after C1, the cell observation value is C2 = 1.46.

[0107] The steps for obtaining the C3 parameter are as follows: this parameter represents the cell observation value at the third time point in the current time window, is extracted from the "time window observation parameter set", and is the observation value immediately following C2. For example, corresponding to the next observation time point after C2, the cell observation value is C3 = 1.465.

[0108] The steps for obtaining the T1 parameter are as follows: this parameter represents the time value of the first time point in the current time window, in hours, extracted from the "time window observation parameter set", and is the time when the cell observation value C1 is recorded, for example, T1 = 70.0 hours.

[0109] The steps for obtaining the T2 parameter are as follows: this parameter represents the time value of the second time point in the current time window, in hours, extracted from the "time window observation parameter set", and is the time when the cell observation value C2 is recorded, for example, T2 = 72.0 hours.

[0110] The steps for obtaining the T3 parameter are as follows: this parameter represents the time value of the third time point in the current time window, in hours, extracted from the "time window observation parameter set", and is the time when the cell observation value C3 is recorded, for example, T3 = 74.0 hours.

[0111] R max The parameter acquisition step is as follows: the parameter is the "maximum cell growth rate result" determined in the previous step, which represents the maximum growth rate index that the cell can achieve under the current drug perturbation conditions. For example, the maximum value R obtained by analyzing the entire instantaneous growth rate sequence is max =0.05 hours- 1 .

[0112] Calculation process:

[0113] Take the parameter values ​​obtained above as an example for calculation:

[0114] C1=1.45 (OD value);

[0115] C2=1.46 (OD value);

[0116] C3=1.465 (OD value);

[0117] T1 = 70.0 hours;

[0118] T2 = 72.0 hours;

[0119] T3 = 74.0 hours;

[0120] R max =0.05 hours- 1 ;

[0121] Calculate the molecular part:

[0122] First-order difference: C2-C1=1.46-1.45=0.01;

[0123] Second-order difference: C3-2C2+C1=1.465-2×1.46+1.45=1.465-2.92+1.45=-0.005;

[0124] Numerator fluctuation term:

[0125] Calculate the denominator:

[0126] Time span: T3-T1=74.0-70.0=4.0 hours;

[0127] Local rate: (C2-C1) / (T2-T1)=(1.46-1.45) / (72.0-70.0)=0.01 / 2.0=0.005 (OD value / hour);

[0128] Rate normalization term:

[0129] Denominator adjustment factor:

[0130] Denominator as a whole:

[0131] Calculate O w :

[0132]

[0133] The results show that in the time window from T1 = 70 hours to T3 = 74 hours, the curve growth dynamic shift intensity O w It is about 0.00254, which is a relatively small value, suggesting that cell growth may tend to be slow or stable during this window period. w The value will be calculated with other time windows O w The values ​​are used together for subsequent determination of the growth stability stage.

[0134] Based on the curve growth dynamic offset intensity O calculated for each time window w The system then performs a continuous sliding window analysis on the entire cell growth curve to identify when the cell growth enters a stable or saturated stage. This process depends on a key parameter, namely the "stability judgment threshold". w,threshold The threshold is set by analyzing a large amount of historical cell culture data to identify cells that are clearly in a saturated stable period (for example, the change in the cell observation value for 24 consecutive hours is less than 2% of the total, or the growth rate is close to zero). w The distribution of values, such as the 95th percentile of these values, or the mean plus two standard deviations, are selected to ensure that the threshold can effectively distinguish between stable and unstable states. For example, if the analysis shows that O w The value is generally lower than 0.004, then O w,threshold It can be set to 0.004. The system will check whether there is a continuous time window, such as at least N consecutive stable = 5 windows (if each window represents T3-T1=4 hours, and the step size is 2 hours, then the 5 windows cover a time span of 4+(5-1)*2=12 hours). In these consecutive windows, the curve growth dynamic offset intensity O calculated in each windoww All are less than the preset O w,threshold If such a time segment is found, the segment is preliminarily marked as a "growth stability stage". If there are multiple such segments, the last one is usually selected as the final growth stability stage. At the end of the determined growth stability stage, such as the last three observation points, the growth direction trend of these observation values ​​is further calculated to determine whether it is a slight increase, stability or a slight decrease. This can be determined by comparing the first-order differences between the three points and referring to a more refined "zero growth judgment threshold" (for example, an observation value change of less than 0.5% is considered to be stable). The final trend description is the end point trend of the growth curve.

[0135] The steps to obtain the cell saturation density results are:

[0136] Based on the endpoint trend of the growth curve, the last three time points in the endpoint trend segment were extracted and the temperature mean was calculated to form an endpoint trend stable observation group;

[0137] According to the observation group with stable endpoint trend, the estimated cell saturation density was calculated using the following formula:

[0138]

[0139] Where S is the estimated cell saturation density, ZC2 and ZC3 are the cell observation values ​​at the second and third time points in the end trend section of the growth curve (in cells / mL), K3 is the temperature value at the third time point (in K), and K avg is the mean temperature at three time points in this section (in K);

[0140] Based on the estimated cell saturation density, the distribution range of the estimated cell saturation density of the same cell type in the same culture environment was compared. If the estimated cell saturation density was within ±5% of the upper limit of the range, it was determined as the cell saturation density result.

[0141] Specifically, based on the end point trend of the growth curve determined in the previous step, which indicates the section where the cell growth curve enters a stable state, the system accurately extracts the complete data corresponding to the last three recorded time points from this section marked as the stable growth stage, including the cell observation value at each time point (for example, it has been converted into cell concentration at this time, in units of cell / mL, recorded as ZC1, ZC2, ZC3) and the culture environment temperature value recorded at the same time point (in units of Kelvin K, recorded as K1, K2, K3). These temperature data are collected synchronously with the cell observation, for example, through real-time recording by the temperature sensor built into the incubator or the microprobe in the culture container. After extracting the cell observation values ​​ZC1, ZC2, ZC3 and the corresponding temperature values ​​K1, K2, K3 at these three time points, the system then calculates the arithmetic mean of these three temperature values, i.e., K avg =(K1+K2+K3) / 3, the average temperature K avg Together with the cell observation values ​​ZC1, ZC2, ZC3 at these three time points and their respective temperature values ​​K1, K2, K3, they form a structured data set, which is defined as the endpoint trend stability observation group for subsequent cell saturation density estimation.

[0142] formula: The formula is useful in that it not only estimates the saturation density based on observations of cells in the stationary phase of growth, but also introduces a correction factor based on the stability of the culture temperature. Represents the average cell density of the last two observation points in the stationary phase, which is a direct reflection of the saturation state. The second According to the temperature K3 of the last observation point relative to the average temperature K during the stable period, avg If the temperature is very stable (K3≈K avg ), the correction factor is close to 1, and the saturation density estimate is mainly determined by the cell observation value; if the temperature fluctuates greatly, the correction factor decreases, thereby reducing the saturation density estimate. This reflects the consideration of the stability of environmental conditions, making the evaluation result more reliable and avoiding the misjudgment of the saturation state due to instantaneous temperature disturbances;

[0143] The steps for obtaining the ZC2 parameter are as follows: this parameter represents the cell observation value recorded in the "end point trend stable observation group" corresponding to the second time point of the three time points in this group, and the unit is cell / mL. This value is derived from the extraction of the corresponding time point data in the "cell growth curve calibration results" and has been calibrated as necessary (for example, from the absorbance OD value through the preset OD-cell / mL calibration curve) to obtain the actual cell concentration. For example, in the end point trend stable observation group, the cell observation value at the second time point is ZC2 = 2.50 × 106 cell / mL.

[0144] The steps for obtaining the ZC3 parameter are as follows: this parameter represents the cell observation value recorded in the "end point trend stable observation group" corresponding to the third (i.e., last) time point of the three time points in this group. The unit is cell / mL. Its acquisition method is the same as ZC2, and both are calibrated and extracted from the data set. For example, in the end point trend stable observation group, the cell observation value at the third time point is ZC3 = 2.52 × 10 6 cell / mL.

[0145] The steps for obtaining the K3 parameter are as follows: this parameter represents the culture temperature measurement value recorded in the "end point trend stabilization observation group" corresponding to the third time point of the three time points in this group, and the unit is Kelvin (K). This temperature data is collected from the temperature monitoring system of the culture environment (such as in an incubator or culture container) in synchronization with the cell observation. For example, the incubator temperature is set to 37°C, and the actual temperature recorded at the third time point is 36.9°C, then K3 = 36.9 + 273.15 = 310.05K.

[0146] K avg The parameter acquisition steps are as follows: the parameter represents the arithmetic mean of the temperature values ​​K1, K2, and K3 corresponding to the three time points T1, T2, and T3 in the "end point trend stable observation group", and the unit is Kelvin (K). The specific calculation is completed when the "end point trend stable observation group" is formed, that is, K avg =(K1+K2+K3) / 3. For example, if the temperatures of the three points are K1=37.0℃(310.15K), K2=37.1℃(310.25K), and K3=36.9℃(310.05K), then K avg =(310.15+310.25+310.05) / 3=310.15K.

[0147] Calculation process:

[0148] Take the parameter values ​​obtained above as an example for calculation:

[0149] ZC2=2.50×10 6 cell / mL;

[0150] ZC3=2.52×10 6 cell / mL;

[0151] K3=310.05K;

[0152] K avg =310.15K;

[0153] Calculate the first term (mean cell density):

[0154]

[0155] Calculate the second term (temperature stability correction factor) internally:

[0156]

[0157] Temperature stability correction factor:

[0158]

[0159] Calculate the estimated cell saturation density S:

[0160] S=(2.51×10 6 cell / mL)·(0.9996776)≈2.50919×10 6 cell / mL;

[0161] The results show that after considering the observed value and temperature stability, the estimated cell saturation density is 2.50919×10 6 cell / mL, this value is a quantitative estimate of the maximum population size that cells can achieve under current culture conditions.

[0162] Based on the estimated cell saturation density S calculated in the previous step, for example, S = 2.50919 × 10 6 cell / mL, the system then compares and verifies it with historical data. Specifically, the system accesses a pre-built database that stores the historical records of estimated cell saturation density values ​​and their statistical distribution characteristics measured by multiple independent experiments of the same cell type (for example, HeLa cells of a specific passage) under exactly the same culture environment parameters (including culture medium type, serum concentration, culture container, gas environment, and nominal temperature, etc.), especially the upper limit value S of the historical saturation density distribution range. hist_upper For example, for the currently tested HeLa cells and culture conditions, historical data indicate that the upper limit of the saturation density range is S hist_upper =2.6×10 6 cell / mL, the system will set an acceptable floating range around this historical upper limit, which is plus or minus 5% of the upper limit. This range is calculated as [S hist_upper ×(1-0.05),S hist_upper ×(1+0.05)], and the value is [2.6×10 6 ×0.95,2.6×10 6 ×1.05]=[2.47×10 6 cell / mL,2.73×10 6cell / mL], and then the currently calculated cell saturation density estimate S = 2.50919×10 6 cell / mL was compared with this acceptable range, since 2.47×10 6 ≤2.50919×10 6 ≤2.73×10 6 , that is, the current estimated value falls within the ±5% interval of the upper limit of the historical range. Therefore, the cell saturation density estimate S is confirmed to be valid and is designated as the cell saturation density result under the current disturbance conditions.

[0163] The steps to obtain the dual-parameter growth eigenvector are:

[0164] Integrate the maximum cell growth rate result and the cell saturation density result, use the maximum cell growth rate result as the first component of the feature vector, and use the cell saturation density result as the second component of the feature vector to obtain an initial dual parameter set;

[0165] Based on the initial two-parameter set, the maximum cell growth rate result of the first component and the cell saturation density result of the second component are unit-checked, the two components are unit-converted and normalized, and then recombined in the order of the maximum cell growth rate result first and the cell saturation density result last to form a dimensionless two-parameter feature set;

[0166] Based on the dimensionless two-parameter feature set, the dimensionless maximum cell growth rate result and the dimensionless cell saturation density result are set as the horizontal and vertical coordinate values ​​of the feature vector, respectively, and the two values ​​are assembled in the form of a feature vector to generate a two-parameter growth feature vector.

[0167] Specifically, the system integrates the two core physiological indicators, the maximum cell growth rate and the cell saturation density obtained in the previous steps. The system first pairs these two values ​​in order. Specifically, the maximum cell growth rate result, such as 0.04hr -1 , as the first element in this paired set, and the cell saturation density result, for example 2.5×10 6 cell / mL, as the second element in this pairing set, is directly combined with its original value and unit without any numerical transformation or unit change. The resulting ordered pair containing these two original physiological parameters (0.04hr -1 ,2.5×10 6 cell / mL), which constitutes the initial two-parameter set.

[0168] Based on the initial two-parameter set formed in the previous step, the set includes the maximum cell growth rate result and the cell saturation density result with their respective units, for example (0.04hr -1 ,2.5×10 6 cell / mL), the system first checks the units of these two components to ensure that they are the expected standard units, such as the rate per hour (hr -1 ) and density is the number of cells per milliliter (cell / mL). If non-standard units exist, unit conversion is performed according to the preset conversion relationship (for example, converting the rate per minute to the rate per hour, or converting the number of cells per liter to the number of cells per milliliter) to ensure consistency in subsequent processing. Next, normalization is performed on these two components separately to convert their values ​​into dimensionless relative values. This normalization process refers to the reference range of the corresponding parameter under physiological or experimental conditions. Specifically, for the maximum cell growth rate result R, its normalized value R norm By formula R norm =(RR min_ref ) / (R max_ref -R min_ref ) is calculated, where R min_ref and R max_ref The minimum (e.g., 0) and maximum physiological growth rate reference values ​​of the cell type are predefined (e.g., through analysis of a large number of historical control experimental data, the physiological limit growth rate generally does not exceed 0.1hr -1 , so set R max_ref =0.1hr -1 ; and R min_ref Usually set to 0), similarly, for the cell saturation density result S, its normalized value S norm By formula S norm =(SS min_ref ) / (S max_ref -S min_ref ) calculation, where S min_ref and S max_ref The predefined minimum (e.g., seeding density 0.2×10 6 cell / mL) and the maximum physiological saturation density reference value (for example, for the current cell type, the highest cell density that can be achieved under optimal conditions is recorded as 5.0×10 6 cell / mL, then set S max_ref =5.0×10 6 cell / mL), for example, if R = 0.04 hr -1 , then R norm =(0.04-0) / (0.1-0)=0.4, if S=2.5×10 6 cell / mL, then Snorm =(2.5×10 6 -0.2×10 6 ) / (5.0×10 6 -0.2×10 6 )≈0.479, the dimensionless maximum cell growth rate result is placed in front, and the dimensionless cell saturation density result is placed in the back, and they are recombined into a new ordered pair to form a dimensionless two-parameter feature set.

[0169] Based on the dimensionless two-parameter feature set formed in the previous step, this set contains two dimensionless values, namely the normalized maximum cell growth rate result R norm (e.g. 0.4) and the normalized cell saturation density result S norm (For example, 0.479), the system then formally constructs these two values ​​into a two-dimensional feature vector. In the specific operation, the dimensionless maximum cell growth rate result R norm Explicitly specified as the first component of the feature vector, this component can be conceptually regarded as the projection length of the vector along a specific axis (usually defined as the abscissa axis or x-axis) in the two-dimensional feature space. At the same time, the dimensionless cell saturation density result S norm Explicitly specified as the second component of the eigenvector, correspondingly as the projection length of the vector on the other axis (usually defined as the ordinate axis or y axis), in this way, these two values ​​that have been standardized and have no physical units are strictly in accordance with [R norm ,S norm ] are assembled in the order of [0.4, 0.479] to form a two-dimensional vector that can be directly used for subsequent mathematical operations or geometric space analysis, such as [0.4, 0.479]. This vector is the final generated two-parameter growth eigenvector.

[0170] The steps for obtaining the results of cell growth sensitivity determination are as follows:

[0171] Based on the dual-parameter growth eigenvector, the maximum cell growth rate and cell saturation density results in the vector were extracted respectively. The results were then compared item by item with the statistical demarcation standard values ​​under the corresponding drug perturbation conditions in the existing cell sensitivity database. The degree of deviation from the standard values ​​was calculated to obtain the eigenvector deviation degree result.

[0172] According to the deviation degree of the characteristic vector, the deviation degree of the maximum cell growth rate result and the cell saturation density result is compared with the preset allowable deviation threshold value in the statistical demarcation standard. When the numerical deviation degree of the two components is less than the allowable deviation threshold value, the characteristic pattern is judged as the cell having low sensitivity to drug perturbation; otherwise, it is judged as high sensitivity, forming the initial judgment state of cell sensitivity;

[0173] Based on the initial judgment state of cell sensitivity, combined with the consistency check of the cell sensitivity pattern under the same conditions in the current disturbance conditions and historical data statistics, the current initial judgment state of cell sensitivity is verified by comparing the historical results, and the cell growth sensitivity judgment result is output.

[0174] Specifically, based on the two-parameter growth feature vector generated in the previous step, the vector is composed of the dimensionless maximum cell growth rate result R norm and the dimensionless cell saturation density S norm For example, [0.4, 0.479], the system first extracts the two components from the vector, namely R norm and S norm Then, the system accesses a pre-built "cell sensitivity database". This database stores historically statistically derived "statistical cutoff standard values" for the cell type and specific drug perturbation conditions (including drug name and concentration) used in the current test. These standard values ​​represent the typical feature vector coordinates that distinguish high and low sensitivity reactions. For example, for the current cell and drug combination, the low sensitivity cutoff standard recorded in the database is R std_low ,S std_low ]=[0.75,0.65], the system will get the R norm With R std_low Compare and S norm With S std_low Compare and calculate the degree of deviation of each value. The degree of deviation can be defined as the difference between the observed value and the standard value. For example, the degree of deviation D R =R norm -R std_low and D S =S norm -S std_low , if R norm =0.4 and R std_low =0.75, then D R =0.4-0.75=-0.35, if S norm =0.479 and S std_low =0.65, then

[0175] D S =0.479-0.65=-0.171, these two deviation values ​​D R and D S Together they constitute the eigenvector deviation result.

[0176] According to the deviation degree of the characteristic vector obtained in the previous step, that is, the deviation degree D of the maximum cell growth rate result R (e.g. -0.35) and the degree of deviation D from the cell saturation density resultS (For example, -0.171), the system compares these two deviations with the "allowable deviation threshold" preset for the corresponding "statistical demarcation standard value" in the "cell sensitivity database". The "allowable deviation threshold" defines the maximum allowable difference between the observed parameter and the standard parameter when it is judged as low sensitivity. These two thresholds, such as Threshold R and Threshold S , is determined based on a large amount of historical experimental data through statistical analysis (such as ROC curve analysis to optimize the classification boundary), and aims to balance the accuracy and robustness of classification. For example, for low sensitivity judgment, if the statistical demarcation standard value [R std_low ,S std_low ] represents the "ideal point" or lower limit of low sensitivity, and the actual R norm and S norm Should not be significantly lower than this standard, then the deviation from the threshold is allowed. R and Threshold S It may be set to a small negative value or a negative value close to zero, such as setting Threshold R =-0.1 and Threshold S =-0.1, which means that if R norm R std_low Lower than 0.1, or S norm Than S std_low If it is lower than 0.1, it is no longer considered to be low sensitivity. The judgment rule is: when D R ≥Threshold R (ie R norm -R std_low ≥-0.1) and D S ≥Threshold S (i.e. S norm -S std_low ≥-0.1), the characteristic pattern is judged as the cell has "low sensitivity" to drug perturbation, otherwise, it is judged as "high sensitivity". In the example, D R =-0.35 less than Threshold R =-0.1, and D S =-0.171 less than Threshold S =-0.1, so the low sensitivity condition is not met, so the initial cell sensitivity state is determined to be "high sensitivity".

[0177] Based on the initial cell sensitivity judgment state formed in the previous step, such as "high sensitivity", the system further combines the specific disturbance conditions currently being tested (i.e., cell type, drug name and concentration) with the historical data statistical information for exactly the same disturbance conditions in the "cell sensitivity database" for consistency verification. The database stores the distribution of cell sensitivity patterns observed in previous experiments under the same conditions. For example, for the current cell and drug combination, historical data show that 85% of the cases are "highly sensitive", 10% are "lowly sensitive", and 5% of the results are unclear. The system compares the current "initial cell sensitivity judgment state" ("high sensitivity") with the main sensitivity pattern in the historical data (85% is "high sensitivity"). In order to conduct objective verification, a "historical comparison" is set. The consistency confirmation threshold is determined by analyzing the consistency of historical data and the decision risk. For example, it is set at 70%. If the current preliminary judgment state is consistent with the main pattern in the historical data that exceeds the threshold (in this case, 85%>70%), the preliminary judgment state is strongly supported and verified by the historical data. At this time, the system outputs this verified preliminary judgment state of cell sensitivity ("high sensitivity") as the final cell growth sensitivity judgment result. If the historical data is insufficient or the consistency is lower than the threshold, or there is a significant contradiction with the preliminary judgment state (for example, the preliminary judgment is "low sensitivity", and the vast majority of historical data are "high sensitivity"), the system outputs the preliminary judgment state but adds a low confidence mark, or adjusts and outputs the final result according to preset rules (for example, tending to report higher sensitivity when uncertain).

Claims

1. A cell growth status recognition system, characterized in that: The system comprises: The growth data time series construction module obtains the original cell population growth data under different drug perturbation conditions, sequentially arranges the timestamps and cell observation values ​​in the original cell population growth data, and converts them to obtain the cell growth curve calibration results; A core rate parameter extraction module calculates, based on the cell growth curve calibration result, the numerical change of the cell growth curve calibration result at each continuous time interval to obtain an instantaneous growth rate sequence characterizing the speed of cell proliferation, and determines the peak value based on the instantaneous growth rate sequence to obtain the maximum cell growth rate result; a saturation state parameter quantification module, which analyzes the growth dynamics of the curve under the target rate condition until the saturation state based on the cell growth curve calibration result and the maximum cell growth rate result, determines the growth stable stage, obtains the endpoint trend of the growth curve, and calculates the cell saturation density result based on the endpoint trend of the growth curve; The cell state classification module integrates the maximum cell growth rate result and the cell saturation density result, constructs a vector as the cell growth sensitivity fingerprint of the current disturbance condition, generates a two-parameter growth feature vector, compares and judges the two-parameter growth feature vector with the statistical demarcation standard, identifies the characteristic pattern of cell response to drug disturbance, and outputs the cell growth sensitivity judgment result.

2. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the cell growth curve calibration result are: For the original cell population growth data under different drug perturbation conditions, the acquisition timestamp and cell observation value corresponding to each growth data are obtained one by one, and a one-to-one mapping relationship between each set of timestamps and the corresponding cell observation value is established to obtain the pairing results of timestamps and cell observation values; Based on the pairing results of the timestamps and the cell observation values, the pairing results are sequentially rearranged one by one according to the acquisition timestamps from small to large, and during the rearrangement process, the average of multiple cell observation values ​​corresponding to the same timestamp is calculated as a unique observation value to obtain a rearranged timestamp and cell observation value sequence; Based on the rearranged timestamps and cell observation value sequence, growth curve point connection is performed time stamp by time stamp, and adjacent cell observation value points in the sequence are continuously connected in the order of timestamps to form a cell growth curve calibration result.

3. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the instantaneous growth rate sequence are: Based on the cell growth curve calibration result, three consecutive observation points corresponding to time stamps are extracted to construct a time stamp sequence window. In each time stamp sequence window, the cell observation value corresponding to the time stamp is extracted to form a cell observation value sequence and a time interval sequence. Calculating the instantaneous growth rate index corresponding to each timestamp sequence window according to the cell observation value sequence and the time interval sequence; Based on the instantaneous growth rate index, the instantaneous growth rate indexes corresponding to all time stamp sequence windows are arranged and summarized one by one in chronological order to generate an instantaneous growth rate sequence.

4. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the maximum cell growth rate result are: Based on the instantaneous growth rate sequence, each instantaneous growth rate index in the instantaneous growth rate sequence is extracted one by one, and according to the time stamp sequence corresponding to each instantaneous growth rate index, the size of adjacent instantaneous growth rate indexes is compared one by one to determine the change trend of the instantaneous growth rate index, and obtain an index change trend sequence; According to the exponential change trend sequence, sequentially identifying the inflection points where the trend changes from rising to falling, and extracting the instantaneous growth rate indices of all corresponding inflection points from the exponential change trend sequence to form a growth rate peak candidate set; Based on the growth rate peak candidate set, all instantaneous growth rate indices are compared one by one and the maximum instantaneous growth rate index is determined and designated as the maximum cell growth rate result.

5. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the endpoint trend of the growth curve are: Based on the cell growth curve calibration result and the maximum cell growth rate result, extracting the corresponding three consecutive cell observation values ​​and the corresponding three consecutive time values ​​in each time window, and taking the maximum cell growth rate result as the rate benchmark to form a time window observation parameter set; Calculating the curve growth dynamic offset intensity of the window according to the time window observation parameter set; Based on the dynamic offset intensity of the curve growth corresponding to each time window, a continuous sliding window analysis is performed on the entire growth curve, and the time period in which the dynamic offset intensity of the curve growth in all continuous time windows is lower than the stability judgment threshold is marked as the growth stable stage. The growth direction trend of the observation value at the end of the stable stage is calculated to obtain the end point trend of the growth curve.

6. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the cell saturation density result are: Based on the endpoint trend of the growth curve, the last three time points in the endpoint trend section are extracted and the temperature mean is calculated to form an endpoint trend stable observation group; Calculate the estimated cell saturation density based on the endpoint trend stability observation group; Based on the estimated cell saturation density, the distribution range of the estimated cell saturation density of the same cell type in the same culture environment is compared with the historical records. If the estimated cell saturation density is within ±5% of the upper limit of the range, it is determined as the cell saturation density result.

7. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the dual-parameter growth feature vector are: Integrating the maximum cell growth rate result and the cell saturation density result, taking the maximum cell growth rate result as the first component of the feature vector and taking the cell saturation density result as the second component of the feature vector, to obtain an initial dual-parameter set; According to the initial two-parameter set, a unit check is performed on the maximum cell growth rate result of the first component and the cell saturation density result of the second component, the two components are converted and normalized, and the two components are recombined in the order of the maximum cell growth rate result first and the cell saturation density result second to form a dimensionless two-parameter feature set; Based on the dimensionless two-parameter feature set, the dimensionless maximum cell growth rate result and the dimensionless cell saturation density result are set as the horizontal and vertical coordinate values ​​of the feature vector, respectively, and the two values ​​are assembled in the form of a feature vector to generate a two-parameter growth feature vector.

8. The cell growth status recognition system according to claim 1, characterized in that: The steps for obtaining the cell growth sensitivity determination result are: Based on the dual-parameter growth eigenvector, the maximum cell growth rate result and the cell saturation density result in the vector are respectively extracted, and compared item by item with the statistical demarcation standard values ​​under the corresponding drug perturbation conditions in the existing cell sensitivity database, and the degree of deviation from the standard value is calculated to obtain the eigenvector deviation degree result; According to the result of the deviation degree of the characteristic vector, the deviation degree of the maximum cell growth rate result and the cell saturation density result are respectively compared with the allowable deviation threshold value preset in the statistical demarcation standard. When the numerical deviation degree of the two components is less than the allowable deviation threshold value, the characteristic pattern is judged as the cell having low sensitivity to drug perturbation; otherwise, it is judged as high sensitivity, forming an initial judgment state of cell sensitivity; Based on the initial cell sensitivity determination state, combined with the consistency check of the cell sensitivity pattern under the same conditions in the current disturbance conditions and historical data statistics, the current initial cell sensitivity determination state is verified by comparing the historical results, and the cell growth sensitivity determination result is output.

Images