Clinical Decision Support System for Inspection and Quality Control

Abstract

A clinical decision support system for laboratory quality control provides real-time quality control and decision support by analyzing patient test data. When skewed distributions are detected in the test data, a second decision point is calculated based on the skewed distribution. This avoids the impact of patient sample variability on the stability and consistency of test results, thereby improving laboratory operational efficiency and the accuracy of test results. It also avoids the reliance on manually set target values ​​in traditional quality control processes, which can lead to deviations in the testing system's performance from expected results.

Description

Technical Field

[0001] A system that monitors internal quality control using patient data, particularly a clinical decision support system for laboratory quality control. Prior Technology

[0002] Internal quality control (IQC) is a management procedure to ensure the quality of healthcare services. It primarily uses a series of standards and processes to monitor, evaluate, and improve every aspect of the healthcare process, ensuring that services such as diagnosis, nursing, and testing meet predetermined standards.

[0003] In traditional quality control (QC) methods, medical laboratory testing systems typically use standard QC materials to ensure the accuracy and reliability of test results. These QC materials include synthetic serum, plasma, urine, cerebrospinal fluid, etc., which differ from actual patient samples. This difference is known as the exchangeability problem, referring to the inability of non-human-derived materials to fully simulate human samples in a testing system, and thus, to completely mimic the complexity of human samples. Therefore, the results from these QC materials sometimes fail to accurately reflect the performance of the testing system under real-world conditions and may overlook potential errors or anomalies during the testing process.

[0004] Furthermore, traditional quality control processes rely on manually set target values. However, when the testing system experiences systematic deviations, these set target values ​​may fail to effectively reflect changes in testing performance. This can lead to a situation where quality control still shows normal results even if there are problems with the testing system, but in reality, the performance of the testing system has deviated from expectations. Such situations can affect the authenticity and reliability of the test results and raise questions.

[0005] To more accurately reflect actual testing conditions, some studies have proposed using patient samples directly for quality control. Using patient samples as quality control material allows for a more accurate description of a laboratory's testing performance by calculating the average of all patient test results monthly. This method helps identify potential anomalies or errors in the data, thereby improving the accuracy and reliability of test results.

[0006] However, while using patient samples for quality control has its advantages, challenges remain due to the variability among patient samples. This variability can affect the stability and consistency of test results, especially when processing large numbers of samples. Therefore, combining traditional quality control methods with the practical application of patient samples is a key issue in improving quality management in medical laboratories. Summary of the Invention

[0007] In view of the above problems, the purpose of this invention is to provide a clinical decision support system for quality control in order to solve the problem that the variability between patient samples affects the stability and consistency of test results.

[0008] Provides a patient data monitoring internal quality control system, comprising: A server, comprising: A storage unit stores multiple patient records; A processing unit is electrically connected to the storage unit to access the patient data, calculate a skewness value based on the patient data, and further determine whether the skewness value exceeds a positive skewness threshold. When the skewness value exceeds the positive skewness threshold, the processing unit further determines whether the skewness value exceeds a negative skewness threshold. When the skewness value exceeds the negative skewness threshold, the processing unit performs a negative skewness distribution calculation based on the patient data to calculate a negative skewness decision point, and confirms the decision result using the negative skewness decision point; When the skewness value exceeds the positive skewness threshold but does not exceed the negative skewness threshold, the processing unit performs a normal distribution calculation based on the patient data to calculate a normal decision point, and confirms a decision result using the normal decision point; Wherein, when the skewness value does not exceed the positive skewness threshold, the processing unit performs a positive skewness distribution calculation based on the patient data to calculate a positive skewness decision point, and confirms the decision result using the positive skewness decision point; A display unit, electrically connected to the processing unit, displays the decision result; Specifically, when the processing unit performs the negative skewed distribution calculation based on the patient data, the processing unit calculates the complex negative skewed quantiles of the patient data and calculates the arithmetic mean of the negative skewed distribution of the negative skewed quantiles, and the processing unit uses the arithmetic mean of the negative skewed distribution as the negative skewed decision point; Among them, these negatively skewed quantiles are greater than the median of these patient data; Specifically, when the processing unit performs the normal distribution calculation based on the patient data, the processing unit calculates the complex normal quantiles of the patient data and calculates the normal distribution arithmetic mean of the normal quantiles, and the processing unit uses the normal distribution arithmetic mean as the normal decision point; These normal quantiles are symmetrical about the median of these patient data; Specifically, when the processing unit performs the positively skewed distribution calculation based on the patient data, the processing unit calculates the complex positively skewed quantiles of the patient data and calculates the arithmetic mean of the positively skewed quantiles, and the processing unit uses the arithmetic mean of the positively skewed distribution as the positively skewed decision point; Among them, these positively skewed quantiles are smaller than the median of these patient data.

[0009] This invention provides a clinical decision support system for laboratory quality control. By analyzing patient test data, it offers real-time quality control and decision support. Furthermore, when the test data is detected to be positively skewed, it calculates positively skewed decision points based on the skewed distribution. This avoids the impact of patient sample variability on the stability and consistency of test results, thereby improving laboratory operational efficiency and the accuracy of test results. It also avoids the reliance on manually set target values ​​in traditional quality control processes, which can lead to deviations in the testing system's performance from expected results. Simple Explanation of the Diagram

[0010] Figure 1: Block diagram of the clinical decision support system for quality control of the present invention. Figure 2: Schematic diagram of patient data in the clinical decision support system for quality control of the present invention. Figure 3: Flowchart of the clinical decision support system for quality control of the present invention. Figure 4: Schematic diagram of the experimental results of the clinical decision support system for quality control of the present invention. Figure 5: Schematic diagram of the experimental results of the clinical decision support system for quality control of the present invention. Implementation

[0011] Please refer to Figures 1 and 2. The clinical decision support system for quality control of the present invention includes a server 10. The server includes a storage unit 11, a processing unit 12, and a display unit 13. The storage unit 11 stores a plurality of patient data.

[0012] The processing unit 12 is electrically connected to the storage unit 11 to access the patient data and calculate a skewness value based on the patient data. The processing unit 12 further determines whether the skewness value exceeds a positive skewness threshold. When the skewness value exceeds the positive skewness threshold, the processing unit 12 performs a normal distribution calculation based on the patient data to calculate a normal decision point and confirms a decision result using the normal decision point. When the skewness value does not exceed the positive skewness threshold, the processing unit 12 performs a positive skewness distribution calculation based on the patient data to calculate a positive skewness decision point and confirms the decision result using the positive skewness decision point.

[0013] The display unit 13 is electrically connected to the processing unit 12 to receive the decision result transmitted by the processing unit 12 and to display the decision result.

[0014] In medical laboratory quality control without standard control solutions, when patient data exhibits random testing bias due to changes in patient population or systematic testing bias due to instruments, reagents, or corrections, the data may not form a normal distribution. If evaluation is still based on a normal distribution, misjudgments are likely. Therefore, the clinical decision support system for quality control in this invention, upon determining that the patient data is skewed, calculates a positively skewed decision point based on the skewed distribution and confirms the decision result based on this positively skewed decision point. This avoids the impact of variability among patient samples on the stability and consistency of test results, thereby improving laboratory operational efficiency and the accuracy of test results.

[0015] Furthermore, the clinical decision support system for quality control may also include a hospital server 20, which is electrically connected to the server 10 via a communication network 30, and the hospital server 20 stores the plurality of patient data. The processing unit 12 downloads the patient data from the hospital server 20 via the communication network 30 and stores it in the storage unit 11.

[0016] In this embodiment, the processing unit 12 calculates a median of the patient data and an arithmetic mean of the patient data, and the processing unit 12 divides the median by the arithmetic mean to obtain the skewed value.

[0017] For example, the processing unit 12 first sorts the patient data in ascending or descending order to rank the patient data according to their numerical values. Then, the processing unit 12 calculates the median of the patient data and the arithmetic mean of the patient data. Finally, the processing unit 12 divides the median of the patient data by the arithmetic mean of the patient data to obtain the skewness value.

[0018] Furthermore, when calculating the median of these patient data, if the total number of patient data is odd, the median is directly obtained by taking the middle value of the data after sorting by numerical value. If the total number of patient data is even, the median is obtained by taking the two middle values ​​of the data after sorting by numerical value and calculating their arithmetic mean. In this embodiment, the median is calculated using the "MEDIAN" function in Microsoft Excel.

[0019] Furthermore, when the processing unit 12 performs the normal distribution calculation based on the patient data, the processing unit 12 calculates the complex normal quantiles of the patient data and calculates the normal distribution arithmetic mean of the normal quantiles, and the processing unit 12 uses the normal distribution arithmetic mean as the normal decision point. The normal quantiles 12 are symmetrical about the median of the patient data.

[0020] In this embodiment, the normal quantiles include a first normal quantile, a second normal quantile, and a third normal quantile of the patient data. The first normal quantile is the first quartile of the patient data, the second normal quantile is the median of the patient data, and the third normal quantile is the third quartile of the patient data.

[0021] For example, if the total number of patient data is 6, and the ascending order is 1, 3, 5, 6, 7, 7, then the value of the first quartile at this position is 1.75. This means that the first quartile lies between the value of the first position and the value of the second position, and this first quartile can be calculated using interpolation; for example, the value of the first quartile is 2.5.

[0022] In this embodiment, the first quartile and the third quartile are calculated using the "QUARTILE.INC" function in Microsoft Excel.

[0023] Furthermore, for example, the positive skewness threshold can be set to 0.97. As shown in Figure 2, for example, in a potassium (K) test item of a patient test item in these patient data, the processing unit 12 calculates the skewness value of the potassium (K) test item to be 0.99, which is greater than 0.97, indicating that these patient data of the potassium (K) test item still belong to the normal distribution. Therefore, the processing unit 12 performs the normal distribution calculation, for example, the calculation result of CDSS (3) in Figure 2, and calculates the first decision value to be 3.9.

[0024] Furthermore, when the processing unit 12 performs the positively skewed distribution calculation based on the patient data, the processing unit 12 calculates the complex positively skewed quantiles of the patient data and calculates the arithmetic mean of the positively skewed quantiles. The processing unit 12 uses the arithmetic mean of the positively skewed distribution as the positively skewed decision point. These positively skewed quantiles are less than the median of the patient data.

[0025] Since a skewness value less than the positive skewness threshold indicates a positively skewed distribution of patient data, the processing unit 12 performs positive skewness distribution calculations to generate positively skewed decision points. For example, these positively skewed quantiles include a first positively skewed quantile, a second positively skewed quantile, a third positively skewed quantile, a fourth positively skewed quantile, and a fifth positively skewed quantile. The first positively skewed quantile is the 5th percentile of the patient data, the second positively skewed quantile is the 10th percentile, the third positively skewed quantile is the 15th percentile, the fourth positively skewed quantile is the 20th percentile, and the fifth positively skewed quantile is the 30th percentile.

[0026] Similarly, the skewed distribution calculation first sorts the patient data in ascending or descending order to rank them according to their numerical values. Next, the processing unit 12 calculates the 5th, 10th, 15th, 20th, and 30th percentiles of the patient data, and calculates the arithmetic mean of these percentiles as the arithmetic mean of the positively skewed distribution. In this embodiment, the median is calculated using the "PERCENTILE.INC" function in Microsoft Excel.

[0027] For example, as shown in Figure 2, in the blood urea nitrogen (BUN) test item of the patient test items, the processing unit 12 calculates the skewness value of the BUN test item to be 0.78, which is less than 0.97, indicating that the patient data of the BUN test item belongs to a positively skewed distribution. Therefore, the processing unit 12 performs the positively skewed distribution calculation, such as the calculation result of CDSS(4) in Figure 2, and calculates the second decision value to be 12. In contrast, if the normal distribution is still used to calculate the BUN test item, the first decision value obtained will be 19, which has a large deviation and is prone to misjudgment.

[0028] Furthermore, when the skewness value exceeds the skewness threshold, the processing unit 12 further determines whether the skewness value exceeds a negative skewness threshold. When the skewness value does not exceed the second negative skewness threshold, the processing unit 12 performs the normal distribution calculation based on the patient data to calculate the normal decision point. When the skewness value exceeds the negative skewness threshold, the processing unit performs a negative skewness distribution calculation based on the patient data to calculate a negative skewness decision point.

[0029] Since when the skewness value exceeds the negative skewness threshold, it means that the patient data is a negatively skewed distribution. Therefore, the processing unit 12 calculates the negatively skewed decision point by calculating the negatively skewed distribution and confirms the decision result with the negatively skewed decision point.

[0030] Furthermore, when the processing unit 12 performs the negative skewed distribution calculation based on the patient data, the processing unit 12 calculates the complex negative skewed quantiles of the patient data and calculates the arithmetic mean of the negative skewed distribution of these negative skewed quantiles. The processing unit 12 uses the arithmetic mean of the negative skewed distribution as the negative skewed decision point. These negative skewed quantiles are greater than a median of the patient data.

[0031] For example, these negatively skewed quantiles include a first negatively skewed quantile, a second negatively skewed quantile, a third negatively skewed quantile, a fourth negatively skewed quantile, and a fifth negatively skewed quantile. The first negatively skewed quantile is the 95th percentile of these patient data, the second negatively skewed quantile is the 90th percentile of these patient data, the third negatively skewed quantile is the 85th percentile of these patient data, the fourth negatively skewed quantile is the 80th percentile of these patient data, and the fifth negatively skewed quantile is the 70th percentile of these patient data.

[0032] In this embodiment, the negative skewness threshold can be set to 1.03.

[0033] Please refer to Figure 3. A clinical decision support method for quality control in laboratory testing is executed by this processing unit 12 and includes the following steps: Step S301: Read the patient data stored in the storage unit 11; Step S302: Calculate the skewness value of the patient data based on the patient data; Step S303: Determine whether the skewed value of these patient data exceeds the positive skewness threshold; Step S304: When the skewness value exceeds the positive skewness threshold, determine whether the skewness value exceeds the negative skewness threshold; Step S305: When the skewness value does not exceed the negative skewness threshold, the normal distribution is calculated based on the patient data to calculate the normal decision point, and the decision result is confirmed using the normal decision point; Step S306: When the skewness value exceeds the negative skewness threshold, the negative skewness distribution is calculated based on the patient data to calculate the negative skewness decision point, and the decision result is confirmed using the negative skewness decision point; Step S307: When the skewness value does not exceed the positive skewness threshold, the positive skewness distribution is calculated based on the patient data to calculate the positive skewness decision point, and the decision result is confirmed using the positive skewness decision point; Step S308: Display the decision result through the display unit.

[0034] Please refer to Figures 4 and 5. Display unit 13 displays the decision point results. These results can be, for example, a Levey-Jennings control chart or a Medical Decision chart, capable of dynamically monitoring data bias, error causes, and instrument status. The Levey-Jennings control chart displays the daily test values ​​for the patient's laboratory items. The Medical Decision chart displays the standard deviation of decision points based on monthly patient data, using a Westgard Sigma rule to evaluate and ensure the accuracy and reliability of the system.

[0035] For example, the higher the sigma value, the smaller the system error and the better the quality control.

[0036] In summary, the clinical decision support system for laboratory quality control of this invention can provide real-time quality control and decision support by analyzing patient test data. This improves laboratory operational efficiency and the accuracy of test results, and avoids the deviation of the testing system's performance from expected results caused by relying on manually set target values ​​in traditional quality control processes.

[0037] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

[0038] 10: Server 11: Storage Unit 12: Processing Unit 13: Display Unit 20: Hospital Server 30: Communication Network S301~S308: Steps

Claims

1. A clinical decision support system for quality control in laboratory testing, comprising: a server, including: a storage unit storing a plurality of patient data records; and a processing unit electrically connected to the storage unit for accessing the patient data, calculating a skewness value based on the patient data, and further determining whether the skewness value exceeds a positive skewness threshold; wherein, When the skewed value exceeds the positive skewness threshold, the processing unit further determines whether the skewed value exceeds a negative skewness threshold; wherein, when the skewed value exceeds the negative skewness threshold, the processing unit performs a negative skewed distribution calculation based on the patient data to calculate a negative skewed decision point, and confirms the decision result using the negative skewed decision point; wherein, when the skewed value exceeds the positive skewness threshold but does not exceed the negative skewness threshold, the processing unit performs a normal distribution calculation based on the patient data to calculate a normal decision point, and confirms the decision result using the normal decision point; wherein, when the skewed value does not exceed the positive skewness threshold, the processing unit performs a positive skewed distribution calculation based on the patient data to calculate a positive skewed decision point, and confirms the decision result using the positive skewed decision point; a display unit, electrically connected to the processing unit, displays the decision result; When the processing unit performs the negative skewed distribution calculation based on the patient data, it calculates the complex negative skewed quantiles of the patient data and the arithmetic mean of the negative skewed distribution of these quantiles, using the arithmetic mean of the negative skewed distribution as the negative skewed decision point; wherein these negative skewed quantiles are greater than the median of the patient data; when the processing unit performs the normal distribution calculation based on the patient data, it calculates the complex normal quantiles of the patient data and the arithmetic mean of the normal distribution of these quantiles, using the arithmetic mean of the normal distribution as the normal decision point; wherein these normal quantiles are symmetrical about the median of the patient data. When the processing unit performs the positively skewed distribution calculation based on the patient data, the processing unit calculates the complex positively skewed quantiles of the patient data and calculates the arithmetic mean of the positively skewed quantiles, and the processing unit uses the arithmetic mean of the positively skewed distribution as the positively skewed decision point; wherein the positively skewed quantiles are less than the median of the patient data.

2. The clinical decision support system for quality control of laboratory tests as described in any of the claims in item 1, further comprising: a hospital server electrically connected to the server via a communication network; wherein, The hospital server stores the patient data; the processing unit downloads the patient data from the hospital server via the communication network and stores it in the storage unit.

3. The clinical decision support system for quality control of laboratory tests as described in claim 1 or 2, wherein, The processing unit calculates a median of the patient data and an arithmetic mean of the patient data, and divides the median by the arithmetic mean to obtain the skewed value.

4. The clinical decision support system for quality control as described in claim 1, wherein, These normal quantiles include a first normal quantile, a second normal quantile, and a third normal quantile of the patient data; wherein the first normal quantile is the first quartile of the patient data, the second normal quantile is the median of the patient data, and the third normal quantile is the third quartile of the patient data.

5. The clinical decision support system for quality control as described in claim 1, wherein, These positively skewed quantiles include a first positively skewed quantile, a second positively skewed quantile, a third positively skewed quantile, a fourth positively skewed quantile, and a fifth positively skewed quantile; wherein the first positively skewed quantile is the 5th percentile of these patient data, the second positively skewed quantile is the 10th percentile of these patient data, the third positively skewed quantile is the 15th percentile of these patient data, the fourth positively skewed quantile is the 20th percentile of these patient data, and the fifth positively skewed quantile is the 30th percentile of these patient data.

6. The clinical decision support system for quality control as described in claim 1, wherein, These negatively skewed quantiles include a first negatively skewed quantile, a second negatively skewed quantile, a third negatively skewed quantile, a fourth negatively skewed quantile, and a fifth negatively skewed quantile; wherein the first negatively skewed quantile is the 95th percentile of these patient data, the second negatively skewed quantile is the 90th percentile of these patient data, the third negatively skewed quantile is the 85th percentile of these patient data, the fourth negatively skewed quantile is the 80th percentile of these patient data, and the fifth negatively skewed quantile is the 70th percentile of these patient data.

Images