Application role of an AI-patient based real time quality control intelligent monitoring platform for risk quality management of serum tumor markers

doi:10.3969/j.issn.1673-8640.2025.03.011

Abstract

Abstract:

Objective To investigate the application role of an artificial intelliqence（AI）-patient based real time quality control（PBRTQC） intelligent monitoring platform for risk quality management of 9 serum tumor markers. Methods The AI-PBRTQC intelligent monitoring platform was used to automatically collect the determination results of 106 443 samples of 9 serum tumor markers from Guangzhou Huayin Medical Laboratory Center. from January 1，2023 to December 31，2023. The AI-PBRTQC intelligent software was used for the normal distribution test，parameter setting，program establishment，performance verification，quality control efficiency evaluation and real-time operation of patient big data. According to the intelligent quality control rules based on AI algorithm，the application role of AI-PBRTQC intelligent monitoring platform for serum tumor marker determination quality risk monitoring was evaluated. Results The AI-PBRTQC intelligent monitoring platform selected the optimal program parameters，the patient data exponential weighted moving average（EWMA） quality control chart showed a normal distribution when the performance was stable，and the trend was consistent. In the determination of 9 serum tumor markers，AI-PBRTQC intelligent monitoring platform can sensitively，timely and correctly identify and issue warnings when changing the new batch of detemination products and using new batch of detection reagents cause changes in the accuracy performance of determination system. The optimal weighting coefficient of alpha-fetoprotein（AFP） was 0.03，and the optimal weighting coefficient of the other 8 serum tumor markers was 0.02. The actual cumulative precision（CV） of EWMA for 9 items in 1 year was all lower than the quality target. The AI-PBRTQC intelligent monitoring platform gave 8 warnings in 9 items，all of which were true alarms. Conclusions AI-PBRTQC intelligent monitoring platform can monitor the quality risk of serum tumor marker determinations in real time and accurately identify systemic errors.

Key words: Tumor marker, Patient based real time quality control, Exponential weighted moving average, Intelligent quality risk monitoring, Artificial intelligence

CLC Number:

R446.1

YU Xue, QI Aihong, LI Jinhui, QUAN Zhihui, LI Kui, QIU Yurong. Application role of an AI-patient based real time quality control intelligent monitoring platform for risk quality management of serum tumor markers[J]. Laboratory Medicine, 2025, 40(3): 264-270.

Figures/Tables 8

References 16

[1]	徐菱遥, 江南, 赵俊波, 等. 血清HSP90α联合lncRNA SNHG5与常规血清肿瘤标志物诊断早期结直肠癌的价值[J]. 海南医学, 2024, 35(19):2792-2797.
[2]	ZHOU H, WANG X C, YUAN B B, et al. Clinical value of combining serum tumor marker detection with fecal occult blood testing in diagnosing colorectal cancer[J]. J Physiol Pharmacol, 2022, 73(3):443-449.
[3]	ZHANG Y, WANG H L, XIE Y H, et al. Practical application of the patient data-based quality control method:the potassium example[J]. Biochem Med, 2024, 34(1):010901.
[4]	LORDE N, MAHAPATRA S, KALARIA T. Machine learning for patient-based real-time quality control(PBRTQC),analytical and preanalytical error detection in clinical laboratory[J]. Diagnostics(Basel), 2024, 14(16):1808.
[5]	BADRICK T, BIETENBECK A, KATAYEV A, et al. Implementation of patient-based real-time quality control[J]. Crit Rev Clin Lab Sci, 2020, 57(8):532-547.
[6]	YANG X, CHEN Q, PAN Z, et al. Application of patient-based real-time quality control based on artificial intelligence monitoring platform in continuously quality risk monitoring of down syndrome serum screening[J]. J Clin Lab Anal, 2024, 38(5):e25019.
[7]	LOH T P, BIETENBECK A, CERVINSKI M A, et al. Recommendation for performance verification of patient-based real-time quality control[J]. Clin Chem Lab Med, 2020, 58(8):1205-1213. DOI PMID
[8]	NG D, POLITO F A, CERVINSKI M A. Optimization of a moving averages program using a simulated annealing algorithm:the goal is to monitor the process not the patients[J]. Clin Chem, 2016, 62(10):1361-1371.
[9]	ROSSUM H H V, KEMPERMAN H. Implementation and application of moving average as continuous analytical quality control instrument demonstrated for 24 routine chemistry assays[J]. Clin Chem Lab Med, 2017, 55(8):1142-1151. DOI PMID
[10]	温冬梅, 郝晓柯. 基于患者数据的实时质量控制建立原则及研究进展[J]. 中华检验医学杂志, 2022, 45(1): 82-86.
[11]	李萍, 李园园, 黄亨建, 等. 患者数据质控在定义室内质控分析批长度中的应用[J]. 临床检验杂志, 2009, 27(3):228-232.
[12]	杨帆, 董丹凤, 陆怡德. 患者数据指数加权移动平均法在血清离子项目室内质量控制中的应用[J]. 检验医学, 2021, 36(9):962-968. DOI
[13]	LOH T P, CERVINSKI M A, KATAYEV A, et al. Recommendations for laboratory informatics specifications needed for the application of patient-based real time quality control[J]. Clin Chim Acta, 2019, 495:625-629. DOI PMID
[14]	中华人民共和国国家卫生健康委员会. WS/T 641—2018 临床检验定量测定室内质量控制[S]. 北京: 中华人民共和国国家卫生健康委员会, 2018.
[15]	LU Y, YANG F, WEN D, et al. Assessment of patient based real-time quality control on comparative assays for common clinical analytes[J]. J Clin Lab Anal, 2022, 36(9):e24651.
[16]	徐黎明, 王瑾, 袁梦娇, 等. 患者数据指数加权移动平均法在区域检验中心MCV、MCH、MCHC项目稳定性和可比性监测中的应用[J]. 检验医学, 2023, 38(11):1098-1100. DOI

项目	加权系数	截断范围	样本数/例	EWMA质控图控制限			EWMA质控实际累积结果			质量目标（CV）/ %
项目	加权系数	截断范围	样本数/例	靶值	标准差	CV/%	累积均值	累积标准差	累积CV/%	质量目标（CV）/ %
CA19-9	0.02	2.51~12.28 U·mL^-1	9 597	6.28 U·mL^-1	0.52 U·mL^-1	8.33	8.26 U·mL^-1	0.29 U·mL^-1	4.56	8.33
CEA	0.02	1.77~6.45 ng·mL^-1	24 129	3.07 ng·mL^-1	0.26 ng·mL^-1	8.33	3.07ng·mL^-1	1.23 ng·mL^-1	4.27	8.33
CA125	0.02	10.0~40.0 U·mL^-1	12 762	18.67 U·mL^-1	1.55 U·mL^-1	8.33	18.67 U·mL^-1	0.84 U·mL^-1	4.52	8.33
AFP	0.03	2.06~6.13 ng·mL^-1	26 667	3.32 ng·mL^-1	0.28 ng·mL^-1	8.33	3.32 ng·mL^-1	0.13 ng·mL^-1	4.04	8.33
CA15-3	0.02	2.40~19.2 U·mL^-1	11 896	8.71 U·mL^-1	0.73 U·mL^-1	8.33	9.17 U·mL^-1	0.47 U·mL^-1	5.14	8.33
t-PSA	0.02	0.50~1.50 ng·mL^-1	4 490	0.82 ng·mL^-1	0.07 ng·mL^-1	8.33	0.82 ng·mL^-1	0.02 ng·mL^-1	2.74	8.33
绝经前HE4	0.02	20.0~70.0 pmol·L^-1	10 117	34.69 pmol·L^-1	2.89 pmol·L^-1	8.33	34.69 pmol·L^-1	2.38 pmol·L^-1	6.87	8.33
绝经后HE4	0.02	20.0~140.0 pmol·L^-1	2 829	42.79 pmol·L^-1	3.56 pmol·L^-1	8.33	42.79 pmol·L^-1	2.89 pmol·L^-1	6.76	8.33
f-PSA	0.02	0.10~0.50 ng·mL^-1	3 956	0.25 ng·mL^-1	0.02 ng·mL^-1	8.33	0.25 ng·mL^-1	0.01 ng·mL^-1	4.15	8.33

项目	加权系数	截断范围	样本数/例	EWMA质控图控制限			EWMA质控实际累积结果			质量目标（CV）/ %
项目	加权系数	截断范围	样本数/例	靶值	标准差	CV/%	累积均值	累积标准差	累积CV/%	质量目标（CV）/ %
CA19-9	0.02	2.51~12.28 U·mL^-1	9 597	6.28 U·mL^-1	0.52 U·mL^-1	8.33	8.26 U·mL^-1	0.29 U·mL^-1	4.56	8.33
CEA	0.02	1.77~6.45 ng·mL^-1	24 129	3.07 ng·mL^-1	0.26 ng·mL^-1	8.33	3.07ng·mL^-1	1.23 ng·mL^-1	4.27	8.33
CA125	0.02	10.0~40.0 U·mL^-1	12 762	18.67 U·mL^-1	1.55 U·mL^-1	8.33	18.67 U·mL^-1	0.84 U·mL^-1	4.52	8.33
AFP	0.03	2.06~6.13 ng·mL^-1	26 667	3.32 ng·mL^-1	0.28 ng·mL^-1	8.33	3.32 ng·mL^-1	0.13 ng·mL^-1	4.04	8.33
CA15-3	0.02	2.40~19.2 U·mL^-1	11 896	8.71 U·mL^-1	0.73 U·mL^-1	8.33	9.17 U·mL^-1	0.47 U·mL^-1	5.14	8.33
t-PSA	0.02	0.50~1.50 ng·mL^-1	4 490	0.82 ng·mL^-1	0.07 ng·mL^-1	8.33	0.82 ng·mL^-1	0.02 ng·mL^-1	2.74	8.33
绝经前HE4	0.02	20.0~70.0 pmol·L^-1	10 117	34.69 pmol·L^-1	2.89 pmol·L^-1	8.33	34.69 pmol·L^-1	2.38 pmol·L^-1	6.87	8.33
绝经后HE4	0.02	20.0~140.0 pmol·L^-1	2 829	42.79 pmol·L^-1	3.56 pmol·L^-1	8.33	42.79 pmol·L^-1	2.89 pmol·L^-1	6.76	8.33
f-PSA	0.02	0.10~0.50 ng·mL^-1	3 956	0.25 ng·mL^-1	0.02 ng·mL^-1	8.33	0.25 ng·mL^-1	0.01 ng·mL^-1	4.15	8.33

项目	加权系数	截断范围	样本数/例	最优参数和性能评价结果					临床应用中的报警结果和原因
项目	加权系数	截断范围	样本数/例	Ped/%	Pfr/%	ANPed	FPR/%	FNR/%	预警次数/次	预警类型	预警原因
CA19-9	0.02	2.51~12.28 U·mL^-1	9 597	100	0	12	0.5	3.5	0
CEA	0.02	1.77~6.45 ng·mL^-1	24 129	100	0	8	1.5	0.2	0
CA125	0.02	10.0~40.0 U·mL^-1	12 762	100	0	10	4.0	0.9	0
AFP	0.03	2.06~6.13 ng·mL^-1	26 667	100	0	10	3.0	4.5	0
CA15-3	0.02	2.40~19.2 U·mL^-1	11 896	100	0	7	2.5	0.4	4	真报警	更换新批号校准品
t-PSA	0.02	0.50~1.50 ng·mL^-1	4 490	100	0	7	1.0	0.7	0
绝经前HE4	0.02	20.0~70.0 pmol·L^-1	10 117	100	0	11	2.0	4.0	1	真报警	使用新批次检测试剂
绝经后HE4	0.02	20.0~140 pmol·L^-1	2 829	100	0	7	0.1	0.8	3	真报警	使用新批次检测试剂
f-PSA	0.02	0.10~0.50 ng·mL^-1	3 956	100	0	9	0.3	1.0	0

项目	加权系数	截断范围	样本数/例	最优参数和性能评价结果					临床应用中的报警结果和原因
项目	加权系数	截断范围	样本数/例	Ped/%	Pfr/%	ANPed	FPR/%	FNR/%	预警次数/次	预警类型	预警原因
CA19-9	0.02	2.51~12.28 U·mL^-1	9 597	100	0	12	0.5	3.5	0
CEA	0.02	1.77~6.45 ng·mL^-1	24 129	100	0	8	1.5	0.2	0
CA125	0.02	10.0~40.0 U·mL^-1	12 762	100	0	10	4.0	0.9	0
AFP	0.03	2.06~6.13 ng·mL^-1	26 667	100	0	10	3.0	4.5	0
CA15-3	0.02	2.40~19.2 U·mL^-1	11 896	100	0	7	2.5	0.4	4	真报警	更换新批号校准品
t-PSA	0.02	0.50~1.50 ng·mL^-1	4 490	100	0	7	1.0	0.7	0
绝经前HE4	0.02	20.0~70.0 pmol·L^-1	10 117	100	0	11	2.0	4.0	1	真报警	使用新批次检测试剂
绝经后HE4	0.02	20.0~140 pmol·L^-1	2 829	100	0	7	0.1	0.8	3	真报警	使用新批次检测试剂
f-PSA	0.02	0.10~0.50 ng·mL^-1	3 956	100	0	9	0.3	1.0	0