RESUMO
BACKGROUND: In clinical laboratories, the precision and sensitivity of autoverification technologies are crucial for ensuring reliable diagnostics. Conventional methods have limited sensitivity and applicability, making error detection challenging and reducing laboratory efficiency. This study introduces a machine learning (ML)-based autoverification technology to enhance tumor marker test error detection. METHODS: The effectiveness of various ML models was evaluated by analyzing a large data set of 397 751 for model training and internal validation and 215 339 for external validation. Sample misidentification was simulated by random shuffling error-free test results with a 1% error rate to achieve a real-world approximation. The ML models were developed with Bayesian optimization for tuning. Model validation was performed internally at the primary institution and externally at other institutions, comparing the ML models' performance with conventional delta check methods. RESULTS: Deep neural networks and extreme gradient boosting achieved an area under the receiver operating characteristic curve of 0.834 to 0.903, outperforming that of conventional methods (0.705 to 0.816). External validation by 3 independent laboratories showed that the balanced accuracy of the ML model ranged from 0.760 to 0.836, outperforming the balanced accuracy of 0.670 to 0.773 of the conventional models. CONCLUSIONS: This study addresses limitations regarding the sensitivity of current delta check methods for detection of sample misidentification errors and provides versatile models that mitigate the operational challenges faced by smaller laboratories. Our findings offer a pathway toward more efficient and reliable clinical laboratory testing.
RESUMO
Background: Healthcare 4.0. refers to the integration of advanced technologies, such as artificial intelligence (AI) and big data analysis, into the healthcare sector. Recognizing the impact of Healthcare 4.0 technologies in laboratory medicine (LM), we seek to assess the overall awareness and implementation of Healthcare 4.0 among members of the Korean Society for Laboratory Medicine (KSLM). Methods: A web-based survey was conducted using an anonymous questionnaire. The survey comprised 36 questions covering demographic information (seven questions), big data (10 questions), and AI (19 questions). Results: In total, 182 (17.9%) of 1,017 KSLM members participated in the survey. Thirty-two percent of respondents considered AI to be the most important technology in LM in the era of Healthcare 4.0, closely followed by 31% who favored big data. Approximately 80% of respondents were familiar with big data but had not conducted research using it, and 71% were willing to participate in future big data research conducted by the KSLM. Respondents viewed AI as the most valuable tool in molecular genetics within various divisions. More than half of the respondents were open to the notion of using AI as assistance rather than a complete replacement for their roles. Conclusions: This survey highlighted KSLM members' awareness of the potential applications and implications of big data and AI. We emphasize the complexity of AI integration in healthcare, citing technical and ethical challenges leading to diverse opinions on its impact on employment and training. This highlights the need for a holistic approach to adopting new technologies.
RESUMO
BACKGROUND: This study aimed to determine practical delta check limits (DCLs) for thyroid function tests (TFTs) to detect sample misidentifications across various clinical settings. METHODS: Between 2020 and 2022, 610,437 paired TFT results were collected from six university hospitals. The absolute DCL (absDCL) was determined using the 95th percentile for each clinical setting from a random 60 % of the total data. These absDCLs were then tested within and across different settings using the remaining 40 % of the data, alongside mix-up datasets for result and sample comparisons. The sensitivities of absDCL were calculated within and across groups in the mix-up datasets. RESULTS: Health screening absDCLs were notably lower than in other settings (2.58 vs. 5.93-7.08 for thyroid-stimulating hormone; 4.12 vs. 8.24-10.04 for free thyroxine; 0.49 vs. 0.82-0.91 for total triiodothyronine). The proportion of results exceeding absDCL of health screening differed from those of other clinical settings. Furthermore, sensitivity between health screening and other clinical settings was significantly different in both the result mix-up and sample mix-up datasets. CONCLUSIONS: This study determined practical DCLs for TFTs and highlighted differences in absDCLs between health screening and other settings. These findings emphasize the importance of tailored DCLs in improving the accurate reporting of TFTs.
Assuntos
Testes de Função Tireóidea , Humanos , Testes de Função Tireóidea/normas , Tireotropina/sangue , Tireotropina/análise , Tiroxina/sangue , Tiroxina/análise , Masculino , Feminino , Adulto , Tri-Iodotironina/sangue , Tri-Iodotironina/análise , Pessoa de Meia-Idade , Glândula Tireoide/fisiologiaRESUMO
Background: In recent decades, the analytical quality of clinical laboratory results has substantially increased because of collaborative efforts. To effectively utilize laboratory results in applications, such as machine learning through big data, understanding the level of harmonization for each test would be beneficial. We aimed to develop a quantitative harmonization index that reflects the harmonization status of real-world laboratory tests. Methods: We collected 2021-2022 external quality assessment (EQA) results for eight tests (HbA1c, creatinine, total cholesterol, HDL-cholesterol, triglyceride, alpha-fetoprotein [AFP], carcinoembryonic antigen [CEA], and prostate-specific antigen [PSA]). This EQA was conducted by the Korean Association of External Quality Assessment Service, using commutable materials. The total analytical error of each test was determined according to the bias% and CV% within peer groups. The values were divided by the total allowable error from biological variation (minimum, desirable, and optimal) to establish a real-world harmonization index (RWHI) at each level (minimum, desirable, and optimal). Good harmonization was arbitrarily defined as an RWHI value ≤ 1 for the three levels. Results: Total cholesterol, triglyceride, and CEA had an optimal RWHI of ≤ 1, indicating an optimal harmonization level. Tests with a desirable harmonization level included HDL-cholesterol, AFP, and PSA. Creatinine had a minimum harmonization level, and HbA1c did not reach the minimum harmonization level. Conclusions: We developed a quantitative RWHI using regional EQA data. This index may help reflect the actual harmonization level of laboratory tests in the field.
RESUMO
Patient-based real-time QC (PBRTQC) uses patient-derived data to assess assay performance. PBRTQC algorithms have advanced in parallel with developments in computer science and the increased availability of more powerful computers. The uptake of Artificial Intelligence in PBRTQC has been rapid, with many stated advantages over conventional approaches. However, until this review, there has been no critical comparison of these. The PBRTQC algorithms based on moving averages, regression-adjusted real-time QC, neural networks and anomaly detection are described and contrasted. As Artificial Intelligence tools become more available to laboratories, user-friendly and computationally efficient, the major disadvantages, such as complexity and the need for high computing resources, are reduced and become attractive to implement in PBRTQC applications.
Assuntos
Algoritmos , Controle de Qualidade , Humanos , Redes Neurais de Computação , Inteligência Artificial , Laboratórios Clínicos/normasRESUMO
BACKGROUND & AIMS: This study aims to reevaluate upper reference limit (URL) for alanine aminotransferase (ALT) by considering the changing epidemiology of major liver diseases. We employed histological and metabolic parameters in Asian living liver donors. METHODS: We performed a retrospective analysis of 5455 potential living liver donors from 2005 to 2019. Participants were screened for hepatitis B, C, HIV, and alcohol use. Histologically and metabolically healthy participants were assessed using the Prati criteria (body mass index <23 kg/m2, triglyceride ≤200 mg/dL, fasting glucose ≤105 mg/dL, total cholesterol ≤220 mg/dL). The updated ALT-URL was determined as the 95th percentile among participants without hepatic steatosis and who met the Prati criteria. RESULTS: The median age was 30 years, with a male predominance (66.2%). Among 5455 participants, 3162 (58.0%) showed no hepatic steatosis, with 1553 (49.1%) meeting both the criteria for no steatosis and the Prati criteria for metabolic health. The updated URL for ALT in these participants was 34 U/L for males and 22 U/L for females, which was significantly lower than conventionally accepted values. Using this revised ALT-URL, 72.8% of males with ALT levels ≥34 U/L and 55.0% of females with ALT levels ≥22 U/L showed signs of steatosis, whereas 32.7% of males and 22.2% of females met the criteria for metabolic syndrome. CONCLUSIONS: Our study provided the newly established reference intervals for ALT levels in a metabolically and histologically verified Asian population. The proposed URL for ALT are 34 U/L and 22 U/L for males and females, respectively.
RESUMO
OBJECTIVES: To evaluate the performance of the Academia-Government Collaboration for Laboratory Medicine Standardization in Korea (KR-STDZN) based on data from KR-STDZN proficiency testing (KR-STDZN-PT) for creatinine over eight years (2015-2022). METHODS: We used KR-STDZN-PT data of creatinine tests from 2015 to 2022. Acceptance of the participating institutions' test results was assessed by calculating the acceptance performance as absolute bias (absBias%), total coefficient of variance (tCV%), and total error (TE%) for each sample using six measurements from each institution and true values of each reference material. The test result was considered acceptable when absBias%, tCV%, and TE% were <5.10, <3.20, and <11.40â¯%, respectively. The proportion of acceptable institutions among all participating institutions in each round was defined as the acceptance rate. Improvements in absBias%, tCV%, and TE% were analyzed using creatinine concentration ranges in samples. RESULTS: The number of participating institutions increased from 2015 to 2017 but remained consistent since 2018. The acceptance rates for absBias% and TE% increased from 52.2 and 77.6â¯%, in 2015 and to 90.7 and 96.3â¯%, in 2022, respectively. The acceptance rate for tCV% remained in the 90â¯% range for eight years. When creatinine <3â¯mg/dL, mean absBias%, and mean TE% improved significantly in 2021-2022 compared to 2015-2016 (p<0.05). When creatinine >3â¯mg/dL, acceptance performance did not improve. Mean tCV% remained consistent annually regardless of creatinine concentration. No significant variations in test methods were observed. CONCLUSIONS: The collaboration between academia and the government improved creatinine testing quality. Nevertheless, KR-STDZN must be expanded and refined.
Assuntos
Academia , Ensaio de Proficiência Laboratorial , Humanos , Creatinina , Padrões de Referência , GovernoRESUMO
OBJECTIVES: Misidentification errors in tumor marker tests can lead to serious diagnostic and treatment errors. This study aims to develop a method for detecting these errors using a machine learning (ML)-based delta check approach, overcoming limitations of conventional methods. METHODS: We analyzed five tumor marker test results: alpha-fetoprotein (AFP), cancer antigen 19-9 (CA19-9), cancer antigen 125 (CA125), carcinoembryonic antigen (CEA), and prostate-specific antigen (PSA). A total of 246,261 records were used in the analysis. Of these, 179,929 records were used for model training and 66,332 records for performance evaluation. We developed a misidentification error detection model based on the random forest (RF) and deep neural network (DNN) methods. We performed an in silico simulation with 1â¯% random sample shuffling. The performance of the developed models was evaluated and compared to conventional delta check methods such as delta percent change (DPC), absolute DPC (absDPC), and reference change values (RCV). RESULTS: The DNN model outperformed the RF, DPC, absDPC, and RCV methods in detecting sample misidentification errors. It achieved balanced accuracies of 0.828, 0.842, 0.792, 0.818, and 0.833 for AFP, CA19-9, CA125, CEA, and PSA, respectively. Although the RF method performed better than DPC and absDPC, it showed similar or lower performance compared to RCV. CONCLUSIONS: Our research results demonstrate that an ML-based delta check method can more effectively detect sample misidentification errors compared to conventional delta check methods. In particular, the DNN model demonstrated superior and stable detection performance compared to the RF, DPC, absDPC, and RCV methods.
RESUMO
BACKGROUND: Clinical laboratory tests are inevitably affected by various factors. Therefore, when comparing consecutive test results, it is crucial to consider the inherent uncertainty of the test. Clinical laboratories use reference change value (RCV) to determine a significant change between 2 results. Whereas the criteria for the interpretation of consecutive results by clinicians are not well known. We investigated the clinician's interpretation of a clinically significant change in consecutive laboratory test results and compared them to RCV. METHODS: We performed a questionnaire survey on clinicians, which comprised 2 scenarios with 22 laboratory test items suggesting initial test results. Clinicians were asked to choose a result showing clinically significant change. RCV of the analytes from EFLM database were collected. RESULTS: We received 290 valid questionnaire responses. Clinicians' opinions on clinically significant change was inconsistent between clinicians and scenarios, and was generally larger than RCV. Clinicians commented that they were not familiar with the variability of the laboratory tests. CONCLUSIONS: Clinicians' opinions on clinically significant changes were more prominent than RCV. Meanwhile, they tended to neglect the analytical and biological variation. Laboratories should properly guide clinicians on the RCV of tests for better decision-making on patients' clinical states.
Assuntos
Serviços de Laboratório Clínico , Laboratórios , Humanos , Técnicas de Laboratório Clínico , Incerteza , Valores de ReferênciaRESUMO
Background: To ensure valid results of big data research in the medical field, the input laboratory results need to be of high quality. We aimed to establish a strategy for evaluating the quality of laboratory results suitable for big data research. Methods: We used Korean Association of External Quality Assessment Service (KEQAS) data to retrospectively review multicenter data. Seven measurands were analyzed using commutable materials: HbA1c, creatinine (Cr), total cholesterol (TC), triglyceride (TG), alpha-fetoprotein (AFP), prostate-specific antigen (PSA), and cardiac troponin I (cTnI). These were classified into three groups based on their standardization or harmonization status. HbA1c, Cr, TC, TG, and AFP were analyzed with respect to peer group values. PSA and cTnI were analyzed in separate peer groups according to the calibrator type and manufacturer, respectively. The acceptance rate and absolute percentage bias at the medical decision level were calculated based on biological variation criteria. Results: The acceptance rate (22.5%-100%) varied greatly among the test items, and the mean percentage biases were 0.6%-5.6%, 1.0%-9.6%, and 1.6%-11.3% for all items that satisfied optimum, desirable, and minimum criteria, respectively. Conclusions: The acceptance rate of participants and their external quality assessment (EQA) results exhibited statistically significant differences according to the quality grade for each criterion. Even when they passed the EQA standards, the test results did not guarantee the quality requirements for big data. We suggest that the KEQAS classification can serve as a guide for building big data.
Assuntos
Antígeno Prostático Específico , alfa-Fetoproteínas , Masculino , Humanos , Garantia da Qualidade dos Cuidados de Saúde , Hemoglobinas Glicadas , Big Data , Estudos Retrospectivos , Troponina I , CreatininaRESUMO
Background: The quality of laboratory test results is crucial for accurate clinical diagnosis and treatment. Pre-analytical errors account for approximately 60%-70% of all laboratory test errors. Laboratory test results may be largely impacted by pre-analytical phase management. However, primary care clinics currently do not have pre-analytical quality management audit systems. We aimed to understand the current status of pre-analytical quality management in laboratory medicine in Korean primary care clinics. Methods: Questionnaires were designed to focus on essential components of the pre-analytical process of primary care clinics. An online survey platform was used to administer the survey to internal medicine or family medicine physicians in primary care clinics. Results: A total of 141 physicians provided a complete response to the questionnaire. In 65.2% of the clinics, patient information was hand-labeled rather than barcoded on the specimen bottles; 14.2% of clinics displayed only one piece of patient information (name or identification number), and 19.9% of clinics displayed two pieces of information. Centrifuges were not available in 29.1% of the clinics. Institutions carrying out the National Health Screening Program (NHSP) used more barcode system and had more centrifuges than institutions that did not carrying out the NHSP. Conclusions: Pre-analytical quality management is inadequate in many primary clinics. We suggest implementation of a mandatory management system, allowing for a pre-analytical quality management to be carried out in primary care clinics.
Assuntos
Laboratórios , Humanos , República da CoreiaRESUMO
BACKGROUND: Chryseobacterium indologenes is ubiquitous in nature and rarely causes infections. However, the clinical impact of C. indologenes has increased in recent years, especially in immunocompromised patients, and has resulted in high mortality rates. We aimed to investigate the clinical and microbiological characteristics of C. indologenes bacteremia. MATERIALS AND METHODS: We retrospectively reviewed medical records of a 642-bed university-affiliated hospital in Korea, dating from January 2001 to December 2020, to investigate C. indologenes bacteremia. RESULTS: A total of 22 C. indologenes isolates were identified from blood culture records. All patients were hospitalized at the time of bacteremia, and the most common manifestation was primary bacteremia. A sizable majority of the patients (83.3%) had underlying diseases, and all patients received intensive care unit care during their admission. The 14-day and 28-day mortality rates were 8.3% and 16.7%, respectively. Importantly, all C. indologenes isolates were 100% susceptible to trimethoprim-sulfamethoxazole. CONCLUSION: In our study, most of the infections were hospital-acquired, and the susceptibility pattern of the C. indologenes isolates showed multidrug resistance. However, trimethoprim-sulfamethoxazole is a potentially useful antibiotic for C. indologenes bacteremia treatment. More attention is required to identify C. indologenes as one of the most important nosocomial bacteria with detrimental effects in immunocompromised patients.
RESUMO
OBJECTIVES: Few studies have reported on delta checks for tumour markers, even though these markers are often evaluated serially. Therefore, this study aimed to establish a practical delta check limit in different clinical settings for five tumour markers: alpha-fetoprotein, cancer antigen 19-9, cancer antigen 125, carcinoembryonic antigen, and prostate-specific antigen. METHODS: Pairs of patients' results (current and previous) for five tumour markers between 2020 and 2021 were retrospectively collected from three university hospitals. The data were classified into three subgroups, namely: health check-up recipient (subgroup H), outpatient (subgroup O), and inpatient (subgroup I) clinics. The check limits of delta percent change (DPC), absolute DPC (absDPC), and reference change value (RCV) for each test were determined using the development set (the first 18 months, n=179,929) and then validated and simulated by applying the validation set (the last 6 months, n=66,332). RESULTS: The check limits of DPC and absDPC for most tests varied significantly among the subgroups. Likewise, the proportions of samples requiring further evaluation, calculated by excluding samples with both current and previous results within the reference intervals, were 0.2-2.9% (lower limit of DPC), 0.2-2.7% (upper limit of DPC), 0.3-5.6% (absDPC), and 0.8-35.3% (RCV99.9%). Furthermore, high negative predictive values >0.99 were observed in all subgroups in the in silico simulation. CONCLUSIONS: Using real-world data, we found that DPC was the most appropriate delta-check method for tumour markers. Moreover, Delta-check limits for tumour markers should be applied based on clinical settings.
Assuntos
Biomarcadores Tumorais , Antígeno Prostático Específico , Masculino , Humanos , Estudos Retrospectivos , Antígeno Carcinoembrionário , Valores de Referência , Antígeno Ca-125RESUMO
Stenotrophomonas maltophilia is a Gram-negative opportunistic pathogen that can cause serious infection. We aimed to analyze the prevalence and susceptibility rates to trimethoprim/sulfamethoxazole of S. maltophilia. We conducted a retrospective study of S. maltophilia isolates from a university hospital from 2001 to 2020. Clinical information, the numbers of isolates and susceptibility rates were analyzed by year. Susceptibility rates and changes in respiratory and non-respiratory samples were compared. 1805 S. maltophilia isolates were identified, of which 81.4% (1469/1805) were from respiratory samples. There was a male predominance and 52% of the isolates were from general wards. The average susceptibility rate was 87.7% and there was no significant annual trend (Pâ =â .519). The susceptibility rate was 88.7% in respiratory samples and 84.1% in non-respiratory samples (Pâ =â .018). Susceptibility analyses using clinical data over long periods can guide the choice of antimicrobials especially for pathogen whose treatment options are limited.
Assuntos
Infecções por Bactérias Gram-Negativas , Stenotrophomonas maltophilia , Combinação Trimetoprima e Sulfametoxazol , Feminino , Humanos , Masculino , Antibacterianos/farmacologia , Antibacterianos/uso terapêutico , Infecções por Bactérias Gram-Negativas/tratamento farmacológico , Infecções por Bactérias Gram-Negativas/epidemiologia , Hospitais Universitários , Testes de Sensibilidade Microbiana , Prevalência , República da Coreia/epidemiologia , Estudos Retrospectivos , Atenção Secundária à Saúde , Stenotrophomonas maltophilia/efeitos dos fármacos , Combinação Trimetoprima e Sulfametoxazol/farmacologia , Combinação Trimetoprima e Sulfametoxazol/uso terapêuticoRESUMO
Background: Chest computed tomography (CT) scans play an important role in the diagnosis of coronavirus disease 2019 (COVID-19). This study aimed to describe the quantitative CT parameters in COVID-19 patients according to disease severity and build decision trees for predicting respiratory outcomes using the quantitative CT parameters. Methods: Patients hospitalized for COVID-19 were classified based on the level of disease severity: (1) no pneumonia or hypoxia, (2) pneumonia without hypoxia, (3) hypoxia without respiratory failure, and (4) respiratory failure. High attenuation area (HAA) was defined as the quantified percentage of imaged lung volume with attenuation values between -600 and -250 Hounsfield units (HU). Decision tree models were built with clinical variables and initial laboratory values (model 1) and including quantitative CT parameters in addition to them (model 2). Results: A total of 387 patients were analyzed. The mean age was 57.8 years, and 50.3% were women. HAA increased as the severity of respiratory outcome increased. HAA showed a moderate correlation with lactate dehydrogenases (LDH) and C-reactive protein (CRP). In the decision tree of model 1, the CRP, fibrinogen, LDH, and gene Ct value were chosen as classifiers whereas LDH, HAA, fibrinogen, vaccination status, and neutrophil (%) were chosen in model 2. For predicting respiratory failure, the decision tree built with quantitative CT parameters showed a greater accuracy than the model without CT parameters. Conclusions: The decision tree could provide higher accuracy for predicting respiratory failure when quantitative CT parameters were considered in addition to clinical characteristics, PCR Ct value, and blood biomarkers.
RESUMO
OBJECTIVES: Recently, the linearity evaluation protocol by the Clinical & Laboratory Standards Institute (CLSI) has been revised from EP6-A to EP6-ED2, with the statistical method of interpreting linearity evaluation data being changed from polynomial regression to weighted least squares linear regression (WLS). We analyzed and compared the analytical measurement range (AMR) verification results according to the present and prior linearity evaluation guidelines. METHODS: The verification of AMR of clinical chemistry tests was performed using five samples with two replicates in three different laboratories. After analyzing the same evaluation data in each laboratory by the polynomial regression analysis and WLS methods, results were compared to determine whether linearity was verified across the five sample concentrations. In addition, whether the 90% confidence interval of deviation from linearity by WLS was included in the allowable deviation from linearity (ADL) was compared. RESULTS: A linearity of 42.3-56.8% of the chemistry items was verified by polynomial regression analysis in three laboratories. For analysis of the same data by WLS, a linearity of 63.5-78.3% of the test items was verified where the deviation from linearity of all five samples was within the ADL criteria, and the cases where the 90% confidence interval of all deviation from linearity overlapped the ADL was 78.8-91.3%. CONCLUSIONS: Interpreting AMR verification data by the WLS method according to the newly revised CLSI document EP6-ED2 could reduce laboratory workload, enabling efficient laboratory practice.
Assuntos
Testes de Química Clínica , Laboratórios , Humanos , Análise dos Mínimos Quadrados , Modelos Lineares , Padrões de ReferênciaRESUMO
Background and Objectives: High-sensitivity cardiac troponin I (hs-TnI) is an important indicator of acute myocardial infarction (AMI) among patients presenting with chest discomfort at the emergency department (ED). We aimed to determine a reliable hs-TnI cut-off by comparing various values for a baseline single measurement and an algorithmic approach. Materials and Methods: We retrospectively reviewed the hs-TnI values of patients who presented to our ED with chest discomfort between June 2019 and June 2020. We evaluated the diagnostic accuracy of AMI with the Beckman Coulter Access hs-TnI assay by comparing the 99th percentile upper reference limits (URLs) based on the manufacturer's claims, the newly designated URLs in the Korean population, and an algorithmic approach. Results: A total of 1296 patients who underwent hs-TnI testing in the ED were reviewed and 155 (12.0%) were diagnosed with AMI. With a single measurement, a baseline hs-TnI cut-off of 18.4 ng/L showed the best performance for the whole population with a sensitivity of 78.7%, specificity of 95.7%, negative predictive value (NPV) of 97.1%, and positive predictive value (PPV) of 71.3%. An algorithm using baseline and 2-3 h hs-TnI values showed an 100% sensitivity, 97.7% specificity, an NPV of 100%, and a PPV of 90.1%. This algorithm used a cut-off of <4 ng/L for a single measurement 3 h after symptom onset or an initial level of <5 ng/L and a change of <5 ng/L to rule a patient out, and a cut-off of ≥50 ng/L for a single measurement or a change of ≥20 ng/L to rule a patient in. Conclusions: The algorithmic approach using serial measurements could help differentiate AMI patients from patients who could be safely discharged from the ED, ensuring that patients were triaged accurately and did not undergo unnecessary testing. The cut-off values from previous studies in different countries were effective in the Korean population.