Method for detecting suction condition, pipetting device, and method for pipetting operation

Abstract

The invention relates to a method for detecting a liquid suction condition, a pipetting device and a method for pipetting operation. The method for detecting the liquid suction condition comprises the following steps of: collecting a plurality of pressure sensing data from the pipetting device when liquid suction is performed for a plurality of times; performing data pre-processing on each piece of pressure sensing data so as to correspondingly obtain each piece of normalized data; retrieving a feature corresponding to each of the normalized data; inputting the plurality of features into a machine learning model; and training with a machine learning model to obtain an imbibition classification model, in which the imbibition classification model comprises: detecting whether an imbibition stage is a normal condition or an abnormal condition, and when the abnormal condition is detected, further classifying the abnormal condition into a normal type, a blocking type, an empty imbibition type, a bubble type, or other types.

Description

Technical Field

[0001] The present disclosure relates to a method for detecting aspiration status, which is used to detect abnormal aspiration status of an instrument, as well as a pipetting device and a pipetting operation method that can distinguish the aspiration status. Background Art

[0002] Currently, testing laboratories have manual, semi-automatic and fully automatic methods for biological sample processing or testing. When a large number of samples need to be tested at the same time, the laboratory will consider using automated equipment (such as a fully automatic cupping system or a fully automatic sample processing system) to pre-process a large number of samples at one time to increase the detection throughput. Automated equipment uses the pipetting module of the liquid handling system to move the biological fluid samples to be tested (for example, saliva / serum / plasma / cerebrospinal fluid, etc.) and the detection reagents between different containers through suction and dispensing operations. Considering the overall performance of the automated machine, the accuracy and precision of the pipetting module become the biggest key.

[0003] However, when an abnormal situation occurs in the existing pipetting module during the aspiration process, the experimenter often needs to intervene to determine the type of abnormality and eliminate various abnormal conditions, which not only consumes manpower, but sometimes may even require shutdown for processing, resulting in the suspension of testing. Summary of the invention

[0004] Some embodiments of the present disclosure provide a method for detecting aspiration conditions, comprising: collecting multiple pressure sensing data (e.g., data A, B, C, D, E) from a pipetting device when performing multiple liquid aspirations; performing data pre-processing on each of the pressure sensing data to obtain corresponding normalized data (e.g., data a, b, c, d, e); extracting a corresponding feature of each normalized data (e.g., data a', b', c', d', e'); inputting the multiple features into a machine learning model; and training with the machine learning model to obtain a pipetting classification model, wherein the pipetting classification model comprises: detecting whether a pipetting stage is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into one of a plurality of types, wherein the types include a normal type, a blockage type, an empty aspiration type, a bubble type, or other types.

[0005] In some implementations, pre-processing the pressure sensing data includes performing min-max normalization.

[0006] In some embodiments, extracting features from the normalized data includes performing differentiation on the normalized data.

[0007] In some embodiments, extracting a plurality of features of the normalized data includes multiplying a minimum value of the differentiated curve by a built-in coefficient to obtain a threshold of the differentiation process.

[0008] In some embodiments, training with the machine learning model includes using normal data and abnormal data from multiple pressure sensing data to perform binary classification modeling, so that the aspiration classification model can distinguish between normal aspiration conditions and abnormal aspiration conditions.

[0009] In some embodiments, training the machine learning model includes performing multi-classification modeling using normal data and abnormal data from multiple pressure sensing data, so that the aspiration classification model can distinguish multiple types of abnormal conditions.

[0010] In some embodiments, the method for establishing a liquid suction classification model further includes using empirical rules to strengthen the liquid suction classification model, wherein the use of empirical rules includes establishing a distinction rule based on normal data and empty suction data, and the distinction rule is used to further distinguish between a normal subtype and an empty suction subtype in the empty suction type.

[0011] In some implementations, inputting the plurality of features into the machine learning model includes combining one of the pressure sensing data and the corresponding plurality of features.

[0012] In some embodiments, inputting a plurality of features into a machine learning model comprises labeling the features as normal labels, empty labels, blocked labels, or bubble labels.

[0013] In some embodiments, collecting a plurality of pressure sensing data from a pipetting device during a plurality of liquid aspirations includes data from different liquid holding container types, different pipetting tips of different volumes, different target aspiration volumes, or different liquid viscosities.

[0014] In some embodiments, training with a machine learning model uses random forests, neural networks, decision trees, gradient boosted trees, or logistic regression.

[0015] Some embodiments of the present disclosure provide a pipetting device, comprising: a pipette, a pressure sensor, and a computer. The pressure sensor is coupled to the pipette and is configured to detect the pressure within the pipette during a pipetting stage and generate a pressure sensing data. The computer is electrically connected to the pressure sensor and comprises: a storage device and a processor. The storage device is configured to store a pipetting classification model, and the pipetting classification model is configured to detect whether the pipetting stage is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into a normal type, a blocked type, an empty suction type, a bubble type, or other types. The processor is electrically connected to the storage device and configured to detect the pipetting condition during the pipetting stage according to the pipetting classification model. In some embodiments, in the pipetting device, the pipetting classification model is established through machine learning.

[0016] In some embodiments, in the pipetting device, the processor is further configured to normalize and feature extract the pressure sensing data from the pressure sensor.

[0017] Some other embodiments of the present disclosure provide a method for pipetting, comprising: aspirating liquid; collecting pressure sensing data including the period of aspirating liquid; performing data pre-processing on the pressure sensing data; after the data pre-processing, extracting multiple features corresponding to the pressure sensing data, and using these features to detect a pipetting condition during the aspirating liquid, including using a pipetting classification model to determine whether the period of aspirating liquid is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into a normal type, a blocked type, an empty aspiration type, a bubble type, or other types. In some embodiments, the pipetting classification model is established by machine learning.

[0018] In some embodiments, the method of pipetting further comprises: performing a post-error detection pipetting process after detecting the obstruction type, the empty aspiration type, or the bubble type.

[0019] In some implementations, performing data pre-processing on the pressure sensing data includes performing minimum-maximum normalization.

[0020] In some implementations, extracting the plurality of features of the pressure sensing data includes a differentiation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to make the above and other objects, features, advantages and implementation modes of the present disclosure more obvious, the accompanying drawings are described as follows.

[0022] Figure 1 shows a pipette according to some embodiments;

[0023] Figure 2A flow chart showing a method for distinguishing imbibition conditions according to some embodiments;

[0024] Figure 3A A pressure curve diagram showing normal conditions after data length adjustment;

[0025] Figure 3B for Figure 3A The pressure curve of is normalized and differentiated once; Figure 3C for Figure 3B The image after feature extraction;

[0026] Figure 4A A pressure graph showing the blockage condition after data length adjustment;

[0027] Figure 4B for Figure 4A The pressure curve of is normalized and differentiated once;

[0028] Figure 4C for Figure 4B The image after feature extraction;

[0029] Figure 5A A pressure curve diagram showing the air suction condition after data length adjustment;

[0030] Figure 5B for Figure 5A The pressure curve of is normalized and differentiated once;

[0031] Figure 5C for Figure 5B The image after feature extraction;

[0032] Fig. 6A A pressure graph showing the bubble condition after data length adjustment;

[0033] Figure 6B for Fig. 6A The pressure curve of is normalized and differentiated once;

[0034] Figure 6C for Figure 6B The image after feature extraction;

[0035] Figure 7 A combined graph is shown, combining a graph of a pressure curve from a normal condition with a graph after normalization, a first differentiation and then feature extraction;

[0036] Figure 8 A flowchart for establishing a liquid imbibition classification model according to some embodiments is shown;

[0037] Fig. 9 shows a confusion matrix according to an example implementation;

[0038] Fig.10 A flow chart showing the use of a liquid imbibition classification model to determine the liquid imbibition status;

[0039] Fig.11 shows a pipetting device according to some embodiments;

[0040] Fig.12 A biological sample detection process according to some embodiments is shown;

[0041] Fig.13 is a curve graph comparing the minimum detection values ​​of empty aspiration conditions at different target aspiration volumes between a built-in aspiration detection module of a commercially available brand of pipetting device and the aspiration model of the present invention according to an exemplary embodiment.

[0042]

Explanation of symbols

[0043] 100: Liquid handling device

[0044] 110:Pipette

[0045] 120: Pressure sensor

[0046] 130: Piston

[0047] 140:Pipette tip

[0048] 150:Container

[0049] 152: Liquid

[0050] 160: Pump

[0051] 200: Method

[0052] 210, 220, 230, 240: Stages

[0053] 212, 214, 216, 222, 232, 242: Operation

[0054] 400: Method

[0055] 410, 420, 430, 440, 450, 460, 470, 480, 490: Operation

[0056] 442: Machine Learning Algorithms

[0057] 500: Liquid handling device

[0058] 510: Computer

[0059] 520: Processor

[0060] 530: Storage device

[0061] 540:Procedure

[0062] 542: Liquid Classification Model

[0063] 544: Post-error detection processing

[0064] 600: Method

[0065] 610, 620, 630, 640, 650, 660, 670, 672: Operation

[0066] 652: Abnormal Condition

[0067] 654, 662, 674: Normal condition

[0068] 664: blocking type

[0069] 666: Bubble Type

[0070] 668, 676: Air suction type

[0071] 680, 682, 684, 686: Post-error detection processing

[0072] 800: Method

[0073] 810, 820, 830, 840, 850,: Steps

[0074] AD: Air displacement channel

[0075] C1: Normal condition

[0076] C2: Abnormal Condition

[0077] CM: Liquid Classification Model

[0078] T1: blocking type

[0079] T2: Air suction type

[0080] T3: Bubble Type

[0081] T4: Normal type

[0082] T5: Other types

[0083] T2S1: Air suction type

[0084] T2S2: Normal subtype DETAILED DESCRIPTION

[0085] The following will be used to clearly illustrate the spirit of the present disclosure with drawings and detailed descriptions. It should be understood that the present disclosure can have various changes in different forms, but they all do not depart from the scope of the present disclosure, and the description and attached drawings are used for illustration, not for limiting the present disclosure.

[0086] See also Figure 1 , depicts a pipetting device according to some embodiments. The pipetting device 100 is an air displacement pipette (ADP), also known as a pressure pipette, and includes a pipette gun 110 and a pressure sensor 120. A piston 130 is disposed inside the pipette gun 110, and the piston 130 can be made of metal, plastic, or ceramic material. When in use, a pipette tip 140 is sleeved at the lower part of the pipette gun 110. The lower part of the pipette gun 110 and the pipette tip 140 are an air displacement channel AD. When the liquid 152 in the container 150 is sucked, the air is discharged by the piston 130, and thus, the vertically moving piston 130 creates a vacuum in the barrel and pipette tip of the sealed pipette gun 110. When the piston is pulled upward, the gas in the rear half is compressed, and the front half of the space near the liquid becomes a vacuum. At this time, the liquid 152 near the lower part of the pipette gun 110 enters the vacuum part. The pressure sensor 120 is coupled to the pipette 110 to sense the pressure inside the barrel of the pipette 110. The working state of the pipette 110 can usually be inferred by observing the changes in the data sensed by the pressure sensor 120. For example, the pressure inside the barrel of the pipette 110 will change when the pipette tip is mounted, removed, aspirated or discharged, when the liquid level is lower than the tip, or when there is an abnormality in aspirating or discharging liquid.

[0087] When an air displacement pipette (ADP) is aspirating liquid, there are usually four conditions, including normal conditions and three abnormal conditions. The three abnormal conditions are blockage, empty aspiration, and bubbles. Figure 3A , Figure 4A , Figure 5A and Fig. 6A , respectively showing the pressure curves of the normal condition, the blocked condition, the empty suction condition, and the bubble condition sensed by a pressure sensor. It can be seen that these four conditions present different pressure curves, so one of the purposes of the present disclosure is to develop a detection method that can automatically identify different liquid aspiration conditions based on the data sensed by the sensor. The blocked condition refers to the condition where the tip of the pipette is blocked by foreign matter during the liquid aspiration process, and the foreign matter includes blood clots, thicker samples, etc. The empty suction condition refers to the condition where air is sucked in while the liquid is being aspirated, resulting in a condition where the liquid aspirated volume is less than expected. The bubble condition refers to the condition where, during liquid level detection, the liquid level height is incorrectly detected due to surface bubbles, causing the aspiration action to inhale some bubbles.

[0088] Usually, commercialized pipetting modules use the principle of needle piston aspiration, and calculate the liquid volume by establishing a corresponding relationship between the value of its internal pressure sensor and the volume of the aspirated liquid. In order to increase the added value of pipetting module products, some manufacturers have built-in related abnormality detection algorithms in their pipetting modules to detect abnormal errors that occur during the pipetting process, especially for abnormalities that may be encountered during the movement of biological fluid samples, including the integrity of the aspirated liquid, blockage (clog), etc.

[0089] Conventional pipetting modules require users to adjust the corresponding algorithm threshold in advance before using different consumables, such as putting on a pipette tip, in order to correctly detect errors. Before the pipetting module sucks liquids of different viscosities, the user needs to set corresponding parameters for different liquids in order to correctly detect errors. In other words, conventional pipetting modules require users to establish relevant parameters for the consumables (such as pipette tips), target liquid characteristics, etc. before operating the pipetting device in order for the algorithm to operate normally.

[0090] One of the purposes of the present disclosure is to establish a suitable machine learning model to distinguish different pipetting conditions, without the need for operators to set parameters for the consumables used in the pipette device in advance. Therefore, pipette tips of different brands and sizes (e.g., 50μL, 200μL, or 1000μL) commonly used in biochemistry can be used without the need to provide the relevant parameters or control conditions required by the general algorithm.

[0091] Multiple embodiments of the present disclosure provide methods and error handling procedures for detecting liquid aspiration errors when performing biological sample pre-processing and testing with a pressure-based liquid transfer device, which can improve the deficiencies of current related technologies in this field.

[0092] The method for detecting the aspiration status of multiple embodiments of the present disclosure is a two-stage process. In the first stage, it is first detected whether an abnormal condition occurs, and in the second stage, the abnormal condition is classified and distinguished into different types. In the first stage, if the characteristics of this data are not similar to the data characteristics of a typical normal condition, this data is first classified as an abnormal condition. This design can use a strategy in the first stage to minimize the occurrence of false alarms (false alarms) when detecting abnormal results while maintaining a good accuracy rate of error detection. In the second stage, the data originally determined to be abnormal conditions are further distinguished into different abnormal types. Since a small part of the data originally (in the first stage) determined to be abnormal conditions is actually normal (that is, the target aspiration amount is sucked), it can also be distinguished in this second stage. The importance of distinguishing different types of abnormal conditions is that corresponding processing procedures can be executed corresponding to specific abnormal types.

[0093] In some embodiments, after detecting an abnormal condition and identifying the type of abnormal condition, a post-detection error process is performed. In some embodiments, the most appropriate and fully automatic error handling process is performed for common errors when the pipetting device draws biological samples, so that no human intervention is required throughout the process.

[0094] In some embodiments, the pressure data sensed by the pressure sensor 120 during aspiration is used to help distinguish whether the aspiration condition is normal or abnormal, and if it is abnormal, what type of abnormality it is. In some embodiments, a reasonable algorithm, such as an aspiration classification model, is constructed by using a machine learning method and multiple pressure data of different conditions to solve how to automatically distinguish the aspiration condition.

[0095] See also Figure 2 , depicting a flow of liquid imbibition classification according to some embodiments. First, the liquid imbibition classification method 200 includes a stage 210, performing data preparation. The data preparation includes an operation 212, performing data collection; then performing an operation 214, performing data pre-processing; and then performing an operation 216, performing feature extraction.

[0096] the following FIG. 3A to FIG. 6C , showing pressure curves from different liquid aspiration conditions (operation 212), curves after data pre-processing and first differentiation (operation 214), and curves after feature extraction (operation 216).

[0097] First, the raw data from different conditions are pre-processed (operation 214). The data pre-processing includes data length adjustment (resize) and data normalization (normalize).

[0098] Data length adjustment is to adjust each training data to a fixed number of sampling points (because the length varies with the amount of liquid aspirated), so that subsequent data modeling can be performed. Data length adjustment can solve the problem of different sampling lengths due to different amounts of liquid aspirated.

[0099] Figure 3A , Figure 4A , Figure 5A and Fig. 6A , which are the pressure curves of normal condition, obstruction condition, air suction condition and bubble condition obtained from the pressure sensor (i.e. the original data from the pressure sensor) after data length adjustment. The horizontal axis represents different sampling time points, and the vertical axis is the pressure reading. After data length adjustment, Figure 3A , Figure 4A , Figure 5A and Fig. 6A As shown in , each figure has 200 sampling time points.

[0100] Data normalization may utilize, for example, Max-Min Normalization to make the pressure value fall within the range of (-1, 1).

[0101] Figure 3B , Figure 4B , Figure 5B and Figure 6B They are Figure 3A , Figure 4A , Figure 5A and Fig. 6A The curve after data preprocessing and first differentiation.

[0102] Then feature extraction (operation 216) is performed. Figure 3C , Figure 4C , Figure 5C and Figure 6C They are Figure 3B , Figure 4B , Figure 5B and Figure 6B The curve after feature extraction is extracted by using data differential, and the threshold is obtained by multiplying the minimum value of the curve after differentiation by a built-in coefficient (set to 0.1-0.4 for different liquid aspiration speeds). Figure 3B , Figure 4B , Figure 5B and Figure 6B The two straight lines in each represent the positive threshold and the negative threshold. After that, differentiate the curve and take the absolute value greater than |threshold|, then the point value is set to 1 or -1 as a feature. That is, in Figure 3B , Figure 4B , Figure 5B and Figure 6B In each case, if the value of the curve is greater than the positive threshold, the value after differentiation is set to 1; if the value of the curve is less than the negative threshold, the value after differentiation is set to -1. After differentiation, it can be observed that different liquid absorption conditions will have different capture characteristics.

[0103] See also Figure 3C , showing the normal suction curve after feature extraction, it can be seen that at the beginning of normal suction, when the piston of the pipette retreats, the air pressure has a significant decreasing characteristic, and at the end of the final suction, because the liquid is sucked into the lumen, the pressure rebounds and tends to balance, and the curve has an increasing characteristic.

[0104] See also Figure 4C It can be seen that when there is a blockage at the tip of the pipette, the pipetting curve after feature extraction will be in a low pressure state (-1) for a longer period of time.

[0105] See also FIG. 5A to FIG. 5CIt can be seen that in the empty suction condition, the pressure drops due to insufficient suction. Therefore, compared with the normal condition, the suction curve after feature capture will have an extra -1 value.

[0106] See also FIG. 6A to FIG. 6C It can be seen that when there are bubbles, the tip of the pipette sucks up a mixture of air and water, so the liquid aspiration curve after feature extraction will have multiple irregular +1 and -1 values.

[0107] Therefore, the data after feature extraction is more concise than the original pressure sensing data, and retains useful information for classification.

[0108] Return to view Figure 2 , the method 200 for classifying the liquid aspiration includes a stage 220 for detecting an abnormal condition. In the stage 220, it is identified in operation 222 whether the features captured from operation 216 are abnormal. When the features of the data are similar to the features of the normal liquid aspiration condition, the data can be determined to be in the normal condition C1. When the features of the data are not similar to the features of the normal liquid aspiration condition, the data is classified as the abnormal condition C2. When the abnormal condition C2 is detected, the abnormal condition C2 is further classified.

[0109] In operation 222, in some embodiments, features extracted from normal data and abnormal data in the pressure sensing data are used to perform machine learning binary classification modeling, so that the suction classification model can distinguish between normal conditions and abnormal conditions in the first stage. Figure 3C The curve after feature extraction has a value of -1 at the beginning and a value of +1 when the aspiration is about to end. According to this model, most normal conditions can be identified, and the rest are regarded as suspected abnormalities and left to the second stage of abnormal condition classification.

[0110] In stage 230, the classification of abnormal conditions is performed. In operation 232, the abnormal condition C2 determined from operation 222 is further classified into blocking type T1, air suction type T2, and bubble type T3. Some of the normal data will deviate from Figure 3C The typical normal data is thus judged as abnormal condition C2 in stage 220, but it actually belongs to normal type T4, which can be distinguished from the three abnormal conditions at this stage. In addition, if the amount of liquid aspiration is insufficient but it is not similar to the obstruction type, empty aspiration type, and bubble type, it is classified as other type T5.

[0111] In some embodiments, in operation 232, another machine learning modeling is performed to perform multi-classification modeling using normal data and abnormal data from multiple pressure sensing data, so that the liquid absorption classification model can distinguish multiple types of abnormal conditions C2 that were originally determined.

[0112] In some embodiments, in order to better distinguish multiple types of abnormal conditions C2, the original pressure measurement data is combined with the curve after feature extraction to form a combined graph for machine learning modeling. Figure 7 As shown in , the original data from the same normal pressure sensing data and the data after feature extraction are combined to form a combined graph. The combined graph may have the original time series related features and the extracted pressure curve features.

[0113] In some embodiments, stage 240 may be optionally performed to further enhance the accuracy of classification using empirical rules. For example, if the empty suction type T2 is originally determined in operation 232, the data may be further reviewed in operation 242 to determine whether it is an empty suction subtype T2S1 or a normal subtype T2S2 (i.e., the target suction amount is reached).

[0114] In some extreme situations, such as when the liquid is thick or the amount of liquid remaining in the container is small, the amount of liquid aspirated may reach the target value, but the pressure curve or the captured features are not similar to the pressure curve or the captured features of the typical normal situation, so it may be difficult to distinguish the type of abnormal situation. In some embodiments, the third stage of classification is performed, using empirical rules, based on the observation of the data such as the average value, slope, turning point position, etc. of the pressure, to pick out a very small number of special cases that should belong to normal, classify and determine them as normal conditions.

[0115] See also Figure 8 , depicts a flow chart of a method 400 for establishing a liquid imbibition classification model according to some embodiments. Some embodiments of the present disclosure perform one-time data collection / model training in the development phase, and then deploy the model in a test field to perform a detection process.

[0116] The method 400 for establishing a liquid absorption classification model is to first collect pressure sensing data generated by simulating various liquid absorption states, and use these data to establish a model that can distinguish various liquid absorption states in a machine learning manner. During training, the collected pressure sensing data simulating various liquid absorption states are divided into a training data set and a test data set, wherein the training data set is further divided into training data and verification data.

[0117] In the method 400 , first in operation 410 , training data is collected, including data of normal conditions and three types of abnormal conditions (blockage, empty suction, and bubble).

[0118] In operation 420, the data from operation 410 are labeled with different labels, for example, each piece of data is labeled with one of normal, blocked, empty, and bubble labels according to the conditions simulated by the test.

[0119] In operation 430, data pre-processing is performed and features are extracted. Each data has features of a normal condition, a blocked condition, a vacuum condition, and a bubble condition according to the conditions simulated by the test. Data pre-processing may use normalization, such as Min-Max normalization. Features may be extracted using differentiation, as described above.

[0120] In operation 440 , a training phase is performed to obtain a liquid imbibition classification model CM by means of a machine learning algorithm 442 . The liquid imbibition classification model CM can distinguish four liquid imbibition conditions with different characteristics.

[0121] In operation 450, the imbibition classification model CM is cross-validated with the validation data.

[0122] In some embodiments, the first stage of the imbibition classification model CM is to determine whether the imbibition condition is normal or abnormal, and the second stage of the imbibition classification model CM is to distinguish abnormal conditions into different types. Optionally, the imbibition classification model CM includes a third stage, which enhances the classification effect for atypical data according to empirical rules.

[0123] In some embodiments, the machine learning model is trained using random forests, neural networks, decision trees, gradient boosted trees, logistic regression, or other machine learning modeling methods.

[0124] Then, operation 460 is performed, which is the testing phase, using the test set data for testing. First, in operation 470, the test data is collected. Then, in operation 480, the data is pre-processed and features are extracted, and then input into the liquid absorption classification model CM to test the accuracy of the model classification.

[0125] After the established pipetting classification model CM can well distinguish different pipetting conditions, then, in operation 490 , the pipetting classification model CM can be set in a pipetting device of a sample pre-treatment or detection device to detect the pipetting conditions.

[0126] An example implementation of the process for establishing a pipetting model is described below.

[0127] First, simulate various liquid aspiration conditions and collect various test data from a pressure sensor of a pipette device. Artificially simulate three abnormal conditions: use shaking or repeated aspiration to produce bubbles in the liquid, the test tube is filled with less than the actual aspirated amount to produce insufficient volume (empty aspiration), and add matured cornstarch to the sampling tube to simulate blockage. Please refer to the following Table 1 for the test tube specifications, pipette tip specifications, actual test tube loading range, target volume range of the pipette tip, liquid concentration, and number of data records.

[0128] Table 1

[0129]

[0130]

[0131] The collected data are labeled with different labels and divided into four states: normal, blocked, empty suction, and bubble, and various operating conditions of the pipette are taken into consideration.

[0132] Then, data preprocessing is performed. This step includes length adjustment (Resize) to solve the problem of different data lengths, and data normalization to make the pressure value fall within the (-1,1) range.

[0133] Next, feature extraction is performed. Feature extraction is to differentiate the data and multiply the minimum value of the curve after differentiation by a built-in coefficient (set to 0.1 to 0.4 for different liquid absorption speeds) to obtain the threshold. If the absolute value of the differential curve is greater than the |threshold|, the point value is set to 1 or -1 as a feature.

[0134] After that, the first classification stage of the model is constructed, which is abnormal error detection. From the graph after feature extraction, it can be seen that the normal liquid absorption condition has a value of -1 at the beginning and a value of +1 when the liquid absorption is about to end. According to this model, most normal conditions can be identified.

[0135] After that, the second classification stage of the model is constructed, that is, the error classification. This step is modeled by, for example, a random forest machine learning method. Before modeling, the data is length-adjusted and normalized, and then the original image and feature data are combined into new data (such as Figure 7 ), in order to train and predict the model. When using random forest modeling, the data is divided into training set and test set at a ratio of 7:3. The training set is further divided into training set data and validation set data at a ratio of 8:2 for cross-validation, training and parameter adjustment. The number of data items with different data labels is shown in Table 2 below.

[0136] Table 2

[0137] Data Labels quantity Training Data Verify data Test Data normal 3143 1760 440 943 block 1030 577 144 309 Empty suction 1656 926 232 498 Bubbles 1060 593 149 318 total 6889 3856 965 2068

[0138] Then, in the third classification stage of the model construction, empirical rules are used to observe the data such as the mean value, slope, and turning point position, converge the relevant rules, and enhance the effect of the liquid absorption classification model for atypical data.

[0139] Finally, the test set is used to test the effectiveness of the constructed liquid absorption classification model. Fig. 9 , shows the confusion matrix. All normal aspiration conditions are predicted as normal, which means that the algorithm of this example does a good job of classifying errors while minimizing the probability of false alarms, which is also the strategy adopted by this method. In addition, the classification accuracy of the three abnormal conditions is (303+492+308) / (2068-943)=0.9804.

[0140] The detection method of the present invention uses the machine learning model that has been trained to make instant judgments, without the need for operators to set parameters for the consumables used in the pipette device in advance. Before the machine learning model makes a judgment, no parameters or control conditions need to be given, so it can be used for pipette tips of different brands and sizes (such as 50μL, 200μL or 1000μL) and other commonly used biochemical specifications.

[0141] Fig.10 According to some embodiments, a pipetting device of the liquid aspiration type is provided. The pipetting device 500 comprises a pipetting device 100 and a computer 510. The pipetting device 100 can be referred to in connection with Figure 1 In some embodiments, the pipetting device 100 further comprises a robot arm (not shown) for controlling the movement of the pipetting device 100 inside the instrument.

[0142] The computer 510 is electrically connected to the pressure sensor 120 and includes a storage device 530 and a processor 520. The storage device 530 may be, for example, a memory configured to store a plurality of programs 540 and associated files, such as storing a liquid aspiration classification model 542, wherein the liquid aspiration classification model 542 is configured to detect whether the liquid aspiration stage is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into a normal type, a blocked type, an empty suction type, a bubble type, or other types. The processor 520 is electrically connected to the storage device 530 and configured to detect the liquid aspiration condition during the liquid aspiration stage according to the liquid aspiration classification model.

[0143] In some embodiments, the imbibition classification model 542 is established via machine learning.

[0144] In some implementations, the processor 520 is further configured to normalize and feature extract the pressure sensing data from the pressure sensor 120 .

[0145] In some embodiments, the storage device also stores instructions associated with post-error detection processing 544 to eliminate the abnormal condition or re-absorb the liquid when an abnormal condition is detected. In some embodiments, the pipetting device further includes a pump 160 configured to connect to the computer 510 and control the movement of the piston 130.

[0146] Fig.11 A method of pipetting operation according to some embodiments is shown, which is used to determine whether an abnormality occurs during a pipetting stage. In the method 600 of pipetting operation, first, in operation 610, the pressure readings sensed by the pressure sensor during the pipetting are collected.

[0147] In operation 620, the data of the pressure readings is subjected to data length adjustment and normalization.

[0148] In operation 630, feature extraction is performed, for example, in a differential manner, as described above.

[0149] At operation 640, the extracted features are input into the anomaly detection method used, for example, in the first stage of the liquid absorption classification model as described above.

[0150] In operation 650, it is determined whether the imbibition condition is normal or abnormal. When the imbibition condition is normal 654, the device can then perform subsequent normal operations. When the imbibition condition is a potential abnormal condition 652, further in operation 660, a classification method is used to distinguish different types of potential abnormal conditions 652.

[0151] When the abnormal classification method determines that the aspiration condition is a normal condition 662, the device can then perform subsequent normal operations. When the abnormal classification method determines that the potential abnormal condition 652 is a blocking type 664, a bubble type 666, or an empty aspiration type 668, then the error detection post-processing 680, 682, and 684 are performed respectively.

[0152] In some embodiments, optionally, when the abnormal classification method determines that the potential abnormal condition 652 is empty suction, an empirical rule can be used in operation 670 to assist in the determination. In operation 672, when the liquid aspiration condition is determined to be a normal condition 674, the device can then perform subsequent normal operations; when the liquid aspiration condition is determined to be an empty suction type 676 of the abnormal condition, the device can then automatically perform post-error detection processing 686.

[0153] Fig.12 Some steps in the process of a method for pre-processing a biological sample using an instrument according to an exemplary embodiment are shown. In different pipetting stages, after detecting an abnormal situation, the instrument will automatically adopt corresponding post-error detection processing.

[0154] The method 800 for pre-processing biological samples includes step 810, transferring the liquid biological sample to the extraction plate. In this step, when a blockage, empty suction, or bubble occurs, the liquid pipetting device will discharge the liquid previously sucked by the pipette tip, and then move the liquid pipetting device downward to re-absorb the liquid.

[0155] The biological sample pre-treatment method 800 also includes step 820, transferring the reagent to the mixing tube. In this step, when empty suction or bubbles occur, the liquid pipetting device will discharge the liquid previously sucked by the pipette tip above the liquid surface, and then move the liquid pipetting device downward to re-absorb the liquid.

[0156] The biological sample pre-processing method 800 further includes step 830 of performing a reagent mixing process. In this step, since the liquid transfer module does not absorb liquid, the pressure sensor does not perform detection.

[0157] The biological sample pre-treatment method 800 also includes step 840, transferring the liquid from the mixing tube to the PCR plate (extraction plate). In this step, when empty suction or bubbles occur, the pipette device will discharge the liquid previously sucked by the pipette tip, and then move the pipette device downward to re-absorb the liquid.

[0158] The biological sample pretreatment method 800 also includes step 850, transferring the liquid from the extraction plate to the PCR plate. In this step, when an empty suction occurs, the pipette device will discharge the liquid previously sucked by the pipette tip on the liquid surface, and then move the pipette device downward by 1 mm to re-absorb the liquid; when a blockage occurs, the pipette device will discharge the liquid previously sucked by the pipette tip on the liquid surface, and then move the pipette device upward by 1 mm to re-absorb the liquid.

[0159] The liquid aspiration classification model and the liquid transfer operation method established by the embodiments of the present disclosure can be applied to liquid transfer devices of different brands without setting different parameter conditions for different brands. Fig.13 , showing two curves, which compare the minimum detection values ​​of the empty suction condition at different target suction volumes between the built-in liquid suction detection module of a commercially available brand of liquid pipetting device and the liquid suction model of this case.

[0160] In this exemplary embodiment, the pipetting classification model of this case is constructed using a pipetting device of brand A. Afterwards, the pressure sensing data of another brand, brand B, is input into the pipetting classification model of this case for detection. The pipetting device of brand B has a built-in algorithm to detect abnormal conditions. Therefore, the minimum lack of empty suction that can be detected by the built-in algorithm of brand B is compared with the minimum lack of empty suction that can be detected by the pipetting classification model of this case. Moreover, this comparison is classified using the same data source, that is, the data are all pressure data detected by the sensor of brand B.

[0161] exist Fig.13 In the graph, the horizontal axis represents the target aspiration volume (μL), and the vertical axis represents the amount of liquid in the container that is missing relative to the target aspiration volume (%). Each point on the curve represents the minimum detection value of the empty aspiration condition at a specific target aspiration volume. For example, the rightmost point represents the minimum empty aspiration amount that the built-in algorithm of brand B can detect when the target aspiration volume is 200 microliters (μL), which is a 4.5% lack of empty aspiration. Fig.13 As shown in , the minimum missing amount of empty suction that can be detected by the suction classification model in this case is small.

[0162] Depend on Fig.13 As can be seen from the figure, the curves of the aspiration classification model of this case are all obviously smaller than the curve of the built-in algorithm of brand B. And at the leftmost point, that is, when the target aspiration volume is 10μL, the built-in algorithm of brand B cannot detect it at this time; in contrast, the aspiration classification model of this case can detect the lack of 10% aspiration volume, that is, when the target aspiration volume is 10μL of liquid, and the container is actually loaded with 9μL of liquid, the aspiration classification model of this case can detect such an empty aspiration state.

[0163] In other words, the pipetting classification model constructed by the pipetting device of brand A can be applied to pipetting devices of different brands (such as brand B), and can achieve more sensitive and accurate abnormal condition detection compared to the built-in algorithm of brand B.

[0164] The aspiration classification model and the method for detecting aspiration status constructed by the multiple embodiments of the present disclosure do not require the operator to set parameters and conditions in advance for the consumables used in the pipette device, the brand of the pipette device, and the different sizes of pipette tips.

[0165] After the pipetting device and the pipetting operation method disclosed herein detect different pipetting errors, the software in the computer can automatically cooperate with the corresponding processing flow, and the pipetting error can be correctly handled without the intervention of the experimenter to determine the type of abnormality and eliminate various abnormal conditions, effectively increasing the processing efficiency and convenience of operating the biological sample pre-processing automated machine.

[0166] Although the present disclosure has been disclosed as above in multiple implementation modes and examples, they are not intended to limit the present disclosure. Anyone familiar with this technology can make various changes and modifications without departing from the spirit and scope of the present disclosure. Therefore, the scope of protection of the present disclosure shall be based on the scope defined by the attached claims.

Claims

1. A method for detecting liquid absorption status, characterized in that: Include: Collecting a plurality of pressure sensing data from a pipetting device when performing a plurality of liquid aspirations; Performing data pre-processing on the plurality of pressure sensing data to obtain a plurality of normalized data; extracting a plurality of features of the plurality of normalized data; Inputting the plurality of features into a machine learning model; and The machine learning model is used for training to obtain a liquid imbibition classification model, wherein the liquid imbibition classification model comprises: Detecting a suction phase is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into one of a plurality of types, the types comprising a normal type, a blocking type, an empty suction type, a bubble type, and other types.

2. The method for detecting the liquid absorption condition according to claim 1, characterized in that: The pre-processing of the plurality of pressure sensing data includes performing minimum-maximum normalization.

3. The method for detecting the liquid absorption condition according to claim 1, characterized in that: The extracting the features of the normalized data includes performing a differentiation process on the normalized data.

4. The method for detecting the liquid absorption condition as claimed in claim 3, characterized in that: Extracting the features of the normalized data includes multiplying a minimum value of the differentiated curve by a built-in coefficient to obtain a threshold of the differentiation process.

5. The method for detecting liquid absorption status according to claim 1, characterized in that: The training with the machine learning model includes using normal data and abnormal data from the multiple pressure sensing data to perform binary classification modeling, so that the liquid absorption classification model can distinguish between normal liquid absorption conditions and abnormal liquid absorption conditions.

6. The method for detecting the liquid absorption condition according to claim 1, characterized in that: The training with the machine learning model includes performing multi-classification modeling using normal data and abnormal data from the multiple pressure sensing data, so that the liquid absorption classification model can distinguish multiple types of the abnormal conditions.

7. The method for detecting the liquid absorption condition according to claim 1, characterized in that: The method further comprises using empirical rules to strengthen the liquid suction classification model, wherein the using empirical rules comprises establishing a distinction rule according to normal data and empty suction data, and the distinction rule is used to further distinguish a normal subtype and an empty suction subtype in the empty suction type.

8. The method for detecting the liquid absorption condition according to claim 1, characterized in that: Inputting the plurality of features into the machine learning model includes combining one of the plurality of pressure sensing data and the corresponding plurality of features.

9. The method for detecting the liquid absorption condition according to claim 1, characterized in that: Inputting the plurality of features into the machine learning model includes labeling the plurality of features as a normal label, an empty label, a blocked label, or a bubble label.

10. The method for detecting liquid absorption status according to claim 1, characterized in that: The plurality of pressure sensing data collected from the pipetting device during a plurality of liquid aspirations include data from different types of liquid-carrying containers, different volumes of pipetting tips, different target aspiration volumes, or different liquid viscosities.

11. The method for detecting liquid absorption status according to claim 1, characterized in that: The training with the machine learning model uses random forest, neural network, decision tree, gradient boosting tree, logistic regression, or other machine learning modeling methods.

12. A liquid transfer device, characterized in that: Include: A pipette; a pressure sensor coupled to the pipette and configured to detect pressure within the pipette during a pipetting phase and generate a pressure sensing data; and A computer, electrically connected to the pressure sensor, comprising: a storage device configured to store a liquid aspiration classification model, the liquid aspiration classification model being established through machine learning and configured to detect whether the liquid aspiration stage is a normal condition or an abnormal condition, wherein when the abnormal condition is detected, the abnormal condition is further classified into a blocking type, an empty aspiration type, a bubble type, or another type; and A processor is electrically connected to the storage device and configured to detect the imbibition status during the imbibition stage according to the imbibition classification model.

13. The liquid transfer device according to claim 12, characterized in that: The processor is also configured to normalize and feature extract the pressure sensing data from the pressure sensor.

14. A method of pipetting, characterized in that: Include: Absorb a liquid; collecting pressure sensing data including the period of sucking the liquid; performing data pre-processing on the pressure sensing data; After the data pre-processing, extracting a plurality of features of the pressure sensing data; The plurality of features are used to detect a liquid suction condition when the liquid is sucked, including using a liquid suction classification model to determine whether the period of sucking the liquid is a normal condition or an abnormal condition, wherein the liquid suction classification model is established through machine learning, and when the abnormal condition is detected, the abnormal condition is further classified into a blockage type, an empty suction type, a bubble type, or another type.

15. The method of pipetting operation according to claim 14, characterized in that: The method further comprises: after detecting the blocking type, the empty suction type, or the bubble type, performing a post-error detection liquid suction process.

Images