Machine Learning in Laboratory Medicine: Recommendations of the IFCC Working Group.

BACKGROUND: Machine learning (ML) has been applied to an increasing number of predictive problems in laboratory medicine, and published work to date suggests that it has tremendous potential for clinical applications. However, a number of groups have noted the potential pitfalls associated with this work, particularly if certain details of the development and validation pipelines are not carefully controlled. METHODS: To address these pitfalls and other specific challenges when applying machine learning in a laboratory medicine setting, a working group of the International Federation for Clinical Chemistry and Laboratory Medicine was convened to provide a guidance document for this domain. RESULTS: This manuscript represents consensus recommendations for best practices from that committee, with the goal of improving the quality of developed and published ML models designed for use in clinical laboratories. CONCLUSIONS: The committee believes that implementation of these best practices will improve the quality and reproducibility of machine learning utilized in laboratory medicine. SUMMARY: We have provided our consensus assessment of a number of important practices that are required to ensure that valid, reproducible machine learning (ML) models can be applied to address operational and diagnostic questions in the clinical laboratory. These practices span all phases of model development, from problem formulation through predictive implementation. Although it is not possible to exhaustively discuss every potential pitfall in ML workflows, we believe that our current guidelines capture best practices for avoiding the most common and potentially dangerous errors in this important emerging field.

Contribution of symmetric dimethylarginine to GFR decline in pediatric chronic kidney disease.

BACKGROUND: In pediatric chronic kidney disease (pCKD), traditional factors (proteinuria, etiology, and race) do not fully explain disease progression. The levels of methylated arginine derivatives (MADs: asymmetric and symmetric dimethylarginine, respectively) rise in CKD and increase with CKD progression. The impact of MADs on glomerular filtration rate (GFR) decline has not been examined in pCKD. The aim of this study was to examine the additive impact of baseline (BL) levels of MADs on directly measured GFR (mGFR) decline per year (ml/min/1.73 m2/year) for a period of up to 4 years. METHODS: Plasma and data, including mGFR by plasma iohexol clearance, were provided by the prospective, observational Chronic Kidney Disease in Children study. BL MADs were analyzed by high-performance liquid chromatography-tandem mass spectrometry. RESULTS: For 352 pCKD subjects, the median [interquartile range] BL mGFR was 45 [35, 57] ml/min/1.73 m2. The levels of BL MADs were inversely related to the initial mGFR and its decline over time (p < 0.0005) but not to the rate of decline. Covariates, non-glomerulopathy and Tanner stage of ≥ 3 demonstrated weaker relationships between BL levels and beginning mGFR (p = 0.004 and p = 0.002, respectively). CONCLUSIONS: In pCKD, higher concentrations of BL MADs were inversely related to BL mGFR. MADs did not affect the CKD progression rate. Quantification of this relationship is novel to the pCKD literature.

Proceedings of the Association for Pathology Informatics Bootcamp 2022.

The Pathology Informatics Bootcamp, held annually at the Pathology Informatics Summit, provides pathology trainees with essential knowledge in the rapidly evolving field of Pathology Informatics. With a focus on data analytics, data science, and data management in 2022, the bootcamp addressed the growing importance of data analysis in pathology and laboratory medicine practice. The expansion of data-related subjects in Pathology Informatics Essentials for Residents (PIER) and the Clinical Informatics fellowship examinations highlights the increasing significance of these skills in pathology practice in particular and medicine in general. The curriculum included lectures on databases, programming, analytics, machine learning basics, and specialized topics like anatomic pathology data analysis and dashboarding.

Predictive power of serum cystatin C to detect acute kidney injury and pediatric-modified RIFLE class in children undergoing cardiac surgery.

OBJECTIVE: Acute kidney injury is a frequent and serious complication of cardiopulmonary bypass. In current clinical practice, serum creatinine is used to detect acute kidney injury. Cystatin C is a novel biomarker for kidney function that has been shown to be superior to serum creatinine in predicting acute kidney injury in adults after cardiopulmonary bypass. The aim of this study was to determine whether early cystatin C levels predict acute kidney injury associated with cardiopulmonary bypass in pediatric patients undergoing cardiac surgery and if cystatin C could predict pediatric-modified RIFLE (Risk, Injury, Failure, Loss, End-stage kidney disease) class and renal injury as determined by estimated glomerular filtration rate. We also investigated whether ultrafiltration during cardiopulmonary bypass affects cystatin C levels. DESIGN: Prospective, observational cohort study. SETTING: Cardiac intensive care unit in a tertiary, academic pediatric hospital. PATIENTS: One hundred pediatric patients who underwent cardiac surgery involving cardiopulmonary bypass. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Acute kidney injury was defined as a 50% increase in serum creatinine from a preoperative baseline anytime through postoperative day 4. Severity of acute kidney injury was determined by pediatric RIFLE class using estimated glomerular filtration rate criteria only. Renal injury was also determined by an absolute estimated glomerular filtration rate <80 mL/min/1.73 m. Cystatin C levels were measured before and after ultrafiltration. Twenty-eight patients (28%) developed acute kidney injury. Cystatin C predicted acute kidney injury as early as 8 hrs after surgery. When applying pediatric RIFLE criteria to the entire study, 30 patients reached "risk" and five developed "injury." Cystatin C was a good predictor of the development of "injury" (under the receiver operating characteristic curve, 0.834-0.875) and of renal injury by estimated glomerular filtration rate (under the receiver operating characteristic curve, 0.717-0.835) (all p < .05). Cystatin C levels decreased perioperatively and correlated with volume of fluid removed by ultrafiltration. CONCLUSIONS: Cystatin C is an early predictor of acute kidney injury in children after cardiopulmonary bypass. Cystatin C is a good predictor of pediatric RIFLE classification and of decreased estimated glomerular filtration rate after cardiopulmonary bypass. Serum cystatin C may be cleared by ultrafiltration.

Evaluation of asymmetric dimethylarginine, arginine, and carnitine metabolism in pediatric sepsis.

OBJECTIVE: Increased plasma concentrations of the endogenous nitric oxide synthase inhibitor, asymmetric dimethylarginine, decreased arginine bioavailability, and mitochondrial dysfunction have been reported in adult sepsis. We studied whether asymmetric dimethylarginine, arginine, and carnitine metabolism (a measure of mitochondrial dysfunction) are altered in pediatric sepsis and whether these are clinically useful biomarkers. DESIGN: : Prospective, observational study. SETTING: Pediatric intensive care unit at an academic medical center. PATIENTS: : Ninety patients ≤ 18 yrs old, 30 with severe sepsis or septic shock, compared with 30 age-matched febrile and 30 age-matched healthy control subjects. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Plasma asymmetric dimethylarginine and whole blood arginine, citrulline, ornithine, and acylcarnitine:free carnitine ratio were measured daily for septic patients and once for control subjects using tandem mass spectrometry. Plasma asymmetric dimethylarginine concentration (median; interquartile range µmol/L) on day 1 was lower in severe sepsis and septic shock (0.38; 0.30-0.56) compared with febrile (0.45; 0.40-0.59) and healthy (0.60; 0.54-0.67) control subjects (p < .001), although decreased asymmetric dimethylarginine was predominantly found in neutropenic patients. Day 1 arginine was lower in septic (10; interquartile range, 7-20 µmol/L) compared with healthy patients (32; interquartile range, 23-40; p < .001), and the arginine:ornithine ratio was decreased in sepsis, indicating increased arginase activity (an alternative pathway for arginine metabolism). The arginine:asymmetric dimethylarginine and acylcarnitine:free carnitine ratios did not differ between septic and control patients. Asymmetric dimethylarginine was inversely correlated with organ dysfunction by Pediatric Logistic Organ Dysfunction score (r = -0.50, p = .009), interleukin-6 (r = -0.55, p = .01), and interleukin-8 (r = -0.52, p = .03) on admission. Arginine, arginine:asymmetric dimethylarginine, and acylcarnitine:free carnitine were not associated with organ dysfunction or outcomes. CONCLUSIONS: Asymmetric dimethylarginine was decreased in pediatric sepsis and was inversely associated with inflammation and organ dysfunction. This suggests that inhibition of nitric oxide synthase by asymmetric dimethylarginine accumulation is unlikely to impact sepsis pathophysiology in septic children despite decreased arginine bioavailability. We did not find an association of asymmetric dimethylarginine with altered carnitine metabolism nor were asymmetric dimethylarginine, arginine, and acylcarnitine:free carnitine useful as clinical biomarkers.

System Performance Monitoring in Clinical Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS).

Quality assurance (QA) activities enable continuous improvement through ongoing post-implementation monitoring to identify, evaluate, and correct problems. QA for clinical liquid chromatography tandem mass spectrometry (LC-MS/MS) assays should include specific components that address the unique aspects of these methods. This chapter briefly describes approaches for clinical LC-MS/MS system performance monitoring using batch and peak review metrics, largely following CLSI-C62A guidance. Though routine checks ensure the quality of results reported for each run, there is also a need to evaluate metrics between runs over time. Post-implementation performance monitoring of LC-MS/MS methods is typically focused on calibration curves, retention times, peak intensities, and ion ratios.

Create laboratory business intelligence dashboards for free using R: A tutorial using the flexdashboard package.

This tutorial described use of the R flexdashboard package and its possibilities for customization to develop a dashboard to monitor critical result reporting in a clinical laboratory. Several examples of interactive components were demonstrated. Sample code and exercises were provided to facilitate learning.

The Impact of COVID-19 on Laboratory Test Utilization at a Pediatric Medical Center.

BACKGROUND: The epidemiology and clinical manifestation of coronavirus disease 2019 (COVID-19) in the pediatric population is different from the adult population. The purpose of this study is to identify effects of the COVID-19 pandemic on laboratory test utilization in a pediatric hospital. METHODS: We performed retrospective analysis on test utilization data from Ann & Robert H. Lurie Children's Hospital of Chicago, an academic pediatric medical center. Data between two 100-day periods prior to (prepandemic) and during the pandemic (mid-pandemic) were analyzed to evaluate changes in test volume, lab utilization, and test positivity rate. We also evaluated these metrics based on in- vs outpatient testing and performed modeling to determine what variables significantly impact the test positivity rate. RESULTS: During the pandemic period, there was an expected surge in COVID-19 testing, while over 84% of lab tests studied decreased in ordering volume. The average number of tests ordered per patient was not significantly different during the pandemic for any of the laboratories (adjusted P value > 0.05). Thirty-three studied tests showed significant change in positivity rate during the pandemic. Linear modeling revealed test volume and inpatient status as the key variables associated with change in test positivity rate. CONCLUSIONS: Excluding severe acute respiratory syndrome coronavirus 2 tests, the COVID-19 pandemic has generally led to decreased test ordering volume and laboratory utilization. However, at this pediatric hospital, the average number of tests performed per patient and test positivity rates were comparable between pre- and mid-pandemic periods. These results suggest that, overall, clinical test utilization at this site remained consistent during the pandemic.

A rapid LC-MS/MS assay for detection and monitoring of underivatized branched-chain amino acids in maple syrup urine disease.

Introduction: Quantitation of the isomeric branched-chain amino acids (BCAA; valine, alloisoleucine, isoleucine, leucine) is a challenging task that typically requires derivatization steps or long runtimes if a traditional chromatographic method involving a ninhydrin ion pairing reagent is used. Objectives: To develop and perform clinical validation of a rapid, LC-MS/MS-based targeted metabolomics assay for detection and monitoring of underivatized BCAA in human plasma. Methods: Various columns and modes of chromatography were tested. The final optimized method utilized mixed mode chromatography with an Intrada column under isocratic condition. Sample preparation utilized the 96-well format. Briefly, extraction solvent containing the internal standard is added to 20 uL of sample, followed by shaking and positive pressure filtering, and the resulting extracted sample is analyzed. The assay was validated based on accepted quality standards (e.g., CLIA and CLSI) for clinical assays. Results: The method is linear over a wide range of concentrations, 2.0-1500 µM, with LOD of 0.60 µM and LOQ of 2.0 µM. The precision of the assay was 4-10% across analytes. The method was also validated against reference laboratories via blinded split-sample analysis and demonstrated good agreement with accuracy: 89-95% relative to the external group mean. Conclusion: We have developed a method that is accurate, rapid, and reliable for routine clinical testing of patient sample BCAA, which is used in the diagnosis and management of maple syrup urine disease (MSUD). The assay also has desirable characteristics, such as short run time, small sample volume requirement, simple sample preparation without the need for derivatization, and high throughput.

Quantification of Branched-Chain Amino Acids in Plasma by High-Performance Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS).

Branched-chain amino acids (BCAA), including valine, alloisoleucine, isoleucine, and leucine, play significant roles in a number of metabolic pathways in the body. Deficiency in branched-chain ketoacid dehydrogenase complex, an enzyme required for metabolism of those amino acids, will lead to elevation and accumulation of BCAA and ketoacids in bodily fluids. This results in maple syrup urine disease (MSUD), a condition estimated to affect 1 in 100,000-300,000 births. If MSUD is not diagnosed in the first few days of life, progression of this disease can lead to intellectual disability, coma, irreversible brain damage, seizures, or even death. If diagnosed early, MSUD can be managed by monitoring the blood concentrations of BCAA and adjusting the patient's dietary intake accordingly. Therefore, it is critical to have a rapid, accurate, and reliable BCAA assay for confirmation of MSUD in newborns as well as routine monitoring of MSUD patients. Here, we describe a high-performance liquid chromatography tandem mass spectrometry (LC-MS/MS) method for BCAA measurement which requires only 20 µL of plasma. The sample preparation does not require derivatization and only involves protein precipitation with LC/MS-grade methanol, which contains leucine(13C6;15N), isoleucine(13C6;15N), and valine(13C5;15N) as the internal standards. The final sample extracts do not require dry-down and reconstitution and are readily compatible with the liquid chromatography (LC) method. BCAA are separated using the isocratic gradient method on a mixed-mode Intrada column. Multiple-reaction monitoring (MRM) mode is used for MS/MS detection to monitor the parent-to-daughter transitions m/z 132.2 to 86.4 for leucine, isoleucine, and alloisoleucine; m/z 118.2 to 72.4 for valine; m/z 139.2 to 92.4 for leucine(13C6;15N) and isoleucine(13C6;15N); and m/z 124.2 to 77.4 for valine(13C5;15N).

Rise of the Machines: Artificial Intelligence and the Clinical Laboratory.

BACKGROUND: Artificial intelligence (AI) is rapidly being developed and implemented to augment and automate decision-making across healthcare systems. Being an essential part of these systems, laboratories will see significant growth in AI applications for the foreseeable future. CONTENT: In laboratory medicine, AI can be used for operational decision-making and automating or augmenting human-based workflows. Specific applications include instrument automation, error detection, forecasting, result interpretation, test utilization, genomics, and image analysis. If not doing so today, clinical laboratories will be using AI routinely in the future, therefore, laboratory experts should understand their potential role in this new area and the opportunities for AI technologies. The roles of laboratorians range from passive provision of data to fuel algorithms to developing entirely new algorithms, with subject matter expertise as a perfect fit in the middle. The technical development of algorithms is only a part of the overall picture, where the type, availability, and quality of data are at least as important. Implementation of AI algorithms also offers technical and usability challenges that need to be understood to be successful. Finally, as AI algorithms continue to become available, it is important to understand how to evaluate their validity and utility in the real world. SUMMARY: This review provides an overview of what AI is, examples of how it is currently being used in laboratory medicine, different ways for laboratorians to get involved in algorithm development, and key considerations for AI algorithm implementation and critical evaluation.