Maternal and child life years gained by transfusing low titer group O whole blood in trauma: A computer simulation.

BACKGROUND: Using low titer group O whole blood (LTOWB) is increasingly popular for resuscitating trauma patients. LTOWB is often RhD-positive, which might cause D-alloimmunization and hemolytic disease of the fetus and newborn (HDFN) if transfused to RhD-negative females of childbearing potential (FCP). This simulation determined the number of life years gained by the FCP and her future children if she was resuscitated with LTOWB compared with conventional component therapy (CCT). METHODS: The model simulated 500,000 injured FCPs of each age between 0 and 49 years with LTOWB mortality relative reductions (MRRs) compared with components between 0.1% and 25%. For each surviving FCP, number of life years gained was calculated using her age at injury and average life expectancy for American women. The number of expected future pregnancies for FCPs that did not survive was also based on her age at injury; each future child was assigned the maximum lifespan unless they suffered perinatal mortality or serious neurological events from HDFN. RESULTS: The LTOWB group with an MRR 25% compared with CCT had the largest total life years gained. The point of equivalence for RhD-positive LTOWB compared to CCT, where life years lost due to severe HDFN was equivalent to life years gained due to FCP survival/future childbearing, occurred at an MRR of approximately 0.1%. CONCLUSION: In this model, RhD-positive LTOWB resulted in substantial gains in maternal and child life years compared with CCT. A >0.1% relative mortality reduction from LTOWB offset the life years lost to HDFN mortality and severe neurological events.

Dual T-cell constant ß chain (TRBC)1 and TRBC2 staining for the identification of T-cell neoplasms by flow cytometry.

The diagnosis of leukemic T-cell malignancies is often challenging, due to overlapping features with reactive T-cells and limitations of currently available T-cell clonality assays. Recently developed therapeutic antibodies specific for the mutually exclusive T-cell receptor constant ß chain (TRBC)1 and TRBC2 isoforms provide a unique opportunity to assess for TRBC-restriction as a surrogate of clonality in the flow cytometric analysis of T-cell neoplasms. To demonstrate the diagnostic utility of this approach, we studied 164 clinical specimens with (60) or without (104) T-cell neoplasia, in addition to 39 blood samples from healthy donors. Dual TRBC1 and TRBC2 expression was studied within a comprehensive T-cell panel, in a fashion similar to the routine evaluation of kappa and lambda immunoglobulin light chains for the detection of clonal B-cells. Polytypic TRBC expression was demonstrated on total, CD4+ and CD8+ T-cells from all healthy donors; and by intracellular staining on benign T-cell precursors. All neoplastic T-cells were TRBC-restricted, except for 8 cases (13%) lacking TRBC expression. T-cell clones of uncertain significance were identified in 17 samples without T-cell malignancy (13%) and accounted for smaller subsets than neoplastic clones (median: 4.7 vs. 69% of lymphocytes, p < 0.0001). Single staining for TRBC1 produced spurious TRBC1-dim subsets in 24 clinical specimens (15%), all of which resolved with dual TRBC1/2 staining. Assessment of TRBC restriction by flow cytometry provides a rapid diagnostic method to detect clonal T-cells, and to accurately determine the targetable TRBC isoform expressed by T-cell malignancies.

Computational Flow Cytometry Accurately Identifies Sezary Cells Based on Simplified Aberrancy and Clonality Features.

Flow cytometric identification of circulating neoplastic cells (Sezary cells) in patients with mycosis fungoides and Sezary syndrome is essential for diagnosis, staging, and prognosis. Although recent advances have improved the performance of this laboratory assay, the complex immunophenotype of Sezary cells and overlap with reactive T cells demand a high level of analytic expertise. We utilized machine learning to simplify this analysis using only 2 predefined Sezary cell-gating plots. We studied 114 samples from 59 patients with Sezary syndrome/mycosis fungoides and 66 samples from unique patients with inflammatory dermatoses. A single dimensionality reduction plot highlighted all TCR constant ß chain-restricted (clonal) CD3+/CD4+ T cells detected by expert analysis. On receiver operator curve analysis, an aberrancy scale feature computed by comparison with controls (area under the curve = 0.98) outperformed loss of CD2 (0.76), CD3 (0.83), CD7 (0.77), and CD26 (0.82) in discriminating Sezary cells from reactive CD4+ T cells. Our results closely mirrored those obtained by exhaustive expert analysis for event classification (positive percentage agreement = 100%, negative percentage agreement = 99%) and Sezary cell quantitation (regression slope = 1.003, R squared = 0.9996). We demonstrate the potential of machine learning to simplify the accurate identification of Sezary cells.

Effect of residual platelets in frozen-thawed plasma on results of dilute Russell's viper venom time assay for lupus anticoagulant testing.

OBJECTIVES: To determine the impact of residual platelets on dilute Russell's viper venom time (DRVVT) assay in frozen-thawed plasma submitted for lupus anticoagulant (LAC) testing. METHODS: We measured platelet counts in frozen-thawed samples submitted for LAC testing and evaluated the association between platelet count and the DRVVT screening time and ratios. We also spiked platelets into a LAC-positive sample to observe the effect on the DRVVT. RESULTS: Progressive increase in platelet count resulted in a statistically significant shortening of the DRVVT assay results on plasma after 1 freeze-thaw cycle. A similar effect was noted on the LAC-positive sample. CONCLUSIONS: Residual platelets in plasma samples result in shortening of DRVVT assay after 1 freeze-thaw cycle. This may result in a false-negative LAC test result.

The Relationship between Nociceptive Detection Thresholds and Pressure- and Electrical Pain Thresholds: An Explorative Study in Rheumatoid Arthritis Patients.

Recently, methods have been developed enabling the characterization of the nociceptive function at the detection threshold level by measuring nociceptive detection thresholds (NDTs), rather than at the level of the pain threshold via pain threshold (PT) measurements. Both NDT and PT measurements aim to characterize (parts of) the nociceptive system. To date it is unclear if, and if so to what extent, the two outcomes relate to one another. In this study, the primary aim is to explore the relationship between the two measures in patients with rheumatoid arthritis (RA). As secondary aim, we explore differences in NDT between these RA patients with age- and sex-matched healthy controls (HC) from a readily existing dataset. In total 46 RA patients have been recruited, whereby the pressure- (PPT; bilaterally at two locations) and electrical (EPT) pain threshold were evaluated, as well as the NDTs. Significant, positive correlations were found between the EPT and PPT (R=0.54-0.60), but not with the NDTs (R≤0.25). As compared to HC, higher NDTs were found in the RA group. As the presence of a statistically significant weak relationship can only be evaluated using a larger sample size, our results indicate that there is no moderate or stronger relation between PT and NDT outcomes. This implicates that the two outcomes are not strongly driven by the same (nociceptive) mechanism(s). Future research into NDTs and what factors and/or mechanisms affect the outcome, could yield relevant insights into how to use and interpret the results of this relatively new method.Clinical Relevance - The evaluation of nociceptive detection thresholds, in isolation or together with conventionally evaluated pain thresholds, might provide valuable and complementary insights into nociceptive (dis)function in man.

Recommendations for Performance Evaluation of Machine Learning in Pathology: A Concept Paper From the College of American Pathologists.

CONTEXT.­: Machine learning applications in the pathology clinical domain are emerging rapidly. As decision support systems continue to mature, laboratories will increasingly need guidance to evaluate their performance in clinical practice. Currently there are no formal guidelines to assist pathology laboratories in verification and/or validation of such systems. These recommendations are being proposed for the evaluation of machine learning systems in the clinical practice of pathology. OBJECTIVE.­: To propose recommendations for performance evaluation of in vitro diagnostic tests on patient samples that incorporate machine learning as part of the preanalytical, analytical, or postanalytical phases of the laboratory workflow. Topics described include considerations for machine learning model evaluation including risk assessment, predeployment requirements, data sourcing and curation, verification and validation, change control management, human-computer interaction, practitioner training, and competency evaluation. DATA SOURCES.­: An expert panel performed a review of the literature, Clinical and Laboratory Standards Institute guidance, and laboratory and government regulatory frameworks. CONCLUSIONS.­: Review of the literature and existing documents enabled the development of proposed recommendations. This white paper pertains to performance evaluation of machine learning systems intended to be implemented for clinical patient testing. Further studies with real-world clinical data are encouraged to support these proposed recommendations. Performance evaluation of machine learning models is critical to verification and/or validation of in vitro diagnostic tests using machine learning intended for clinical practice.

Establishing NK-Cell Receptor Restriction by Flow Cytometry and Detecting Potential NK-Cell Clones of Uncertain Significance.

Natural killer (NK) cells develop a complex inhibitory and/or activating NK-cell receptor system, including killer cell immunoglobulin-like receptors (KIRs or CD158) and CD94/NKG2 dimers, which are variably combined to generate the individual's NK-cell receptor repertoire. Establishing NK-cell receptor restriction by flow cytometric immunophenotyping is an important step in diagnosing NK-cell neoplasms, but reference interval (RI) data for interpreting these studies are lacking. Specimens from 145 donors and 63 patients with NK-cell neoplasms were used to identify discriminatory rules based on 95% and 99% nonparametric RIs for CD158a+, CD158b+, CD158e+, KIR-negative, and NKG2A+ NK-cell populations to establish NK-cell receptor restriction. These 99% upper RI limits (NKG2a >88% or CD158a >53% or CD158b >72% or CD158e >54% or KIR-negative >72%) provided optimal discrimination between NK-cell neoplasm cases and healthy donor controls with an accuracy of 100% compared with the clinicopathologic diagnosis. The selected rules were applied to 62 consecutive samples received in our flow cytometry laboratory that were reflexed to an NK-cell panel due to an expanded NK-cell percentage (exceeding 40% of total lymphocytes). Twenty-two (35%) of 62 samples were found to harbor a very small NK-cell population with restricted NK-cell receptor expression based on the rule combination, suggestive of NK-cell clonality. A thorough clinicopathologic evaluation for the 62 patients did not reveal diagnostic features of NK-cell neoplasms; therefore, these potential clonal populations of NK cells were designated as NK-cell clones of uncertain significance (NK-CUS). In this study, we established decision rules for NK-cell receptor restriction from the largest published cohorts of healthy donors and NK-cell neoplasms. The presence of small NK-cell populations with restricted NK-cell receptors does not appear to be an uncommon finding, and its significance requires further exploration.

Development, Evaluation, and Multisite Deployment of a Machine Learning Decision Tree Algorithm To Optimize Urinalysis Parameters for Predicting Urine Culture Positivity.

PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Reflex algorithm training utilized results from 19,511 paired UA and UC cases (26.8% UC positive); the average patient age was 57.4 years, and 70% of samples were from female patients. Receiver operating characteristic (ROC) analysis identified urine white blood cells (WBCs), leukocyte esterase, and bacteria as the best predictors of UC positivity, with areas under the ROC curve of 0.79, 0.78, and 0.77, respectively. Using the held-out test data set (9,773 cases; 26.3% UC positive), the PittUDT algorithm met the prespecified target of a negative predictive value above 90% and resulted in a 30 to 60% total negative proportion (true-negative plus false-negative predictions). These data show that a supervised rule-based machine learning algorithm trained on paired UA and UC data has adequate predictive ability for triaging urine specimens by identifying low-risk urine specimens, which are unlikely to grow pathogenic organisms, with a false-negative proportion under 5%. The decision tree approach also generates human-readable rules that can be easily implemented across multiple hospital sites and settings. Our work demonstrates how a data-driven approach can be used to optimize UA parameters for predicting UC positivity in a reflex protocol, with the intent of improving antimicrobial stewardship and UC utilization, a potential avenue for cost savings.

Proceedings of the 2022 NHLBI and OASH state of the science in transfusion medicine symposium.

BACKGROUND: State of the Science (SoS) meetings are used to define and highlight important unanswered scientific questions. The National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health, and the Office of the Assistant Secretary for Health (OASH), Department of Health and Human Services held a virtual SoS in transfusion medicine (TM) symposium. STUDY DESIGN AND METHODS: In advance of the symposium, six multidisciplinary working groups (WG) convened to define research priorities in the areas of: blood donors and the supply, optimizing transfusion outcomes for recipients, emerging infections, mechanistic aspects of components and transfusion, new computational methods in transfusion science, and impact of health disparities on donors and recipients. The overall objective was to identify key basic, translational, and clinical research questions that will help to increase and diversify the volunteer donor pool, ensure safe and effective transfusion strategies for recipients, and identify which blood products from which donors best meet the clinical needs of specific recipient populations. RESULTS: On August 29-30, 2022, over 400 researchers, clinicians, industry experts, government officials, community members, and patient advocates discussed the research priorities presented by each WG. Dialogue focused on the five highest priority research areas identified by each WG and included the rationale, proposed methodological approaches, feasibility, and barriers for success. DISCUSSION: This report summarizes the key ideas and research priorities identified during the NHLBI/OASH SoS in TM symposium. The report highlights major gaps in our current knowledge and provides a road map for TM research.

Hemostatic In Vitro Properties of Novel Plasma Supernatants Produced from Late-storage Low-titer Type O Whole Blood.

BACKGROUND: The use of low-titer group O whole blood is increasing. To reduce wastage, unused units can be converted to packed red blood cells. Supernatant is currently discarded post-conversion; however, it could be a valuable transfusable product. The aim of this study was to evaluate supernatant prepared from late-storage low-titer group O whole blood being converted to red blood cells, hypothesizing it will have higher hemostatic activity compared to fresh never-frozen liquid plasma. METHODS: Low-titer group O whole blood supernatant (n = 12) prepared on storage day 15 was tested on days 15, 21, and 26 and liquid plasma (n = 12) on 3, 15, 21, and 26. Same-day assays included cell counts, rotational thromboelastometry, and thrombin generation. Centrifuged plasma from units was banked for microparticle characterization, conventional coagulation, clot structure, hemoglobin, and additional thrombin generation assays. RESULTS: Low-titer group O whole blood supernatant contained more residual platelets and microparticles compared to liquid plasma. At day 15, low-titer group O whole blood supernatant elicited a faster intrinsic clotting time compared to liquid plasma (257 ± 41 vs. 299 ± 36 s, P = 0.044), and increased clot firmness (49 ± 9 vs. 28 ± 5 mm, P < 0.0001). Low-titer group O whole blood supernatant showed more significant thrombin generation compared to liquid plasma (day 15 endogenous thrombin potential 1,071 ± 315 vs. 285 ± 221 nM·min, P < 0.0001). Flow cytometry demonstrated low-titer group O whole blood supernatant contained significantly more phosphatidylserine and CD41+ microparticles. However, thrombin generation in isolated plasma suggested residual platelets in low-titer group O whole blood supernatant were a greater contributor than microparticles. Additionally, low-titer group O whole blood supernatant and liquid plasma showed no difference in clot structure, despite higher CD61+ microparticle presence. CONCLUSIONS: Plasma supernatant produced from late-storage low-titer group O whole blood shows comparable, if not enhanced, in vitro hemostatic efficacy to liquid plasma.

Transfusion of ABO-group identical red blood cells following uncrossmatched transfusion does not lead to higher mortality in civilian trauma patients.

BACKGROUND: Questions persist about the safety of switching non-group O recipients of group O uncrossmatched red blood cells (RBC) or low titer group O whole blood (LTOWB) to ABO-identical RBCs during their resuscitation. METHODS: The database of an earlier nine-center study of transfusing incompatible plasma to trauma patients was reanalyzed. The patients were divided into three groups based on 24-h RBC transfusion: (1) group O patients who received group O RBC/LTOWB units (control group, n = 1203), (2) non-group O recipients who received only group O units (n = 646), (3) non-group O recipients who received at least one unit of group O and non-group O units (n = 562). Fixed marginal effect of receipt of non-O RBC units on 6- and 24-h and 30-day mortality was calculated. RESULTS: The non-O patients who received only group O RBCs received fewer RBC/LTOWB units and had slightly but significantly lower injury severity score compared to control group; non-group O patients who received both group O and non-O units received significantly more RBC/LTOWB units and had a slightly but significantly higher injury severity score compared to control group. In the multivariate analysis, the non-O patients who received only group O RBCs had significantly higher mortality at 6-h compared to the controls; the non-group O recipients of O and non-O RBCs did not demonstrate higher mortality. At 24-h and 30-days, there were no differences in survival between the groups. CONCLUSION: Providing non-group O RBCs to non-group O trauma patients who also received group O RBC units is not associated with higher mortality.

Metabolic syndrome and the progression of knee osteoarthritis on MRI.

OBJECTIVE: Metabolic osteoarthritis (OA) is one of the proposed clinical phenotypes defined by the existence of metabolic syndrome (MetS). This study aimed to (1) investigate whether MetS and its components are associated with progression of knee OA magnetic resonance imaging (MRI) features, and (2) to evaluate the interaction of MetS with menopause and progression of MRI features. METHOD: 682 women from the Rotterdam Study who participated in a sub-study with knee MRI data available and 5-year follow-up were included. Tibiofemoral (TF) and patellofemoral (PF) OA features were assessed with the MRI Osteoarthritis Knee Score. MetS was quantified by the MetS severity Z-score. Generalized estimating equations were used to evaluate associations between MetS and menopausal transition and progression of MRI features. RESULTS: MetS severity at baseline was associated with progression of osteophytes in all compartments, bone marrow lesions (BMLs) in the PF compartment, and cartilage defects in the medial TF compartment. Waist circumference was associated with progression of osteophytes in all compartments and cartilage defects in the medial TF compartment. High-density lipoprotein (HDL)-cholesterol levels were associated with progression of osteophytes in the medial and lateral TF compartment and glucose levels with osteophytes in the PF and medial TF compartment. No interactions were found between MetS with menopausal transition and MRI features. CONCLUSION: Women with higher MetS severity at baseline showed progression of osteophytes, BMLs, and cartilage defects, indicating more structural knee OA progression after 5 years. Further studies are required to understand whether targeting MetS components may prevent the progression of structural knee OA in women.

Not as "D"eadly as once thought - the risk of D-alloimmunization and hemolytic disease of the fetus and newborn following RhD-positive transfusion in trauma.

The use of blood products to resuscitate injured and massively bleeding patients in the prehospital and early in-hospital phase of the resuscitation is increasing. Using group O red blood cells (RBC) and low titer group O whole blood (LTOWB) avoids an immediate hemolytic reaction from recipient's naturally occurring anti-A and - B, but choosing the RhD type for these products is more nuanced and requires the balancing of product availability and survival benefit against the risk of D-alloimmunization, especially in females of childbearing potential (FCP) due to the possible future occurrence of hemolytic disease of the fetus and newborn (HDFN). Recent models have estimated the risk of fetal/neonatal death from HDFN resulting from D-alloimmunization of an FCP during her trauma resuscitation at between 0-6.5% depending on her age at the time of the transfusion and other societal factors including trauma mortality, her age when she becomes pregnant, frequency of different RHD genotypes in the population, and the probability that the woman will have children with different fathers; this is counterbalanced by an approximately 24% risk of death from hemorrhagic shock. This review will discuss the different models of HDFN outcomes following RhD-positive transfusion as well as the results of recent surveys where the public was asked about their preferences for urgent transfusion in light of the risks of fetal/neonatal adverse events.

Changing to an Opt Out System for Organ Donation-Reflections From England and Netherlands.

Recently England and Netherlands have changed their consent system from Opt In to Opt Out. The reflections shared in this paper give insight and may be helpful for other nation considering likewise. Strong support in England for the change in legislation led to Opt Out being introduced without requiring a vote in parliament in 2019. In Netherlands the bill passed by the smallest possible majority in 2018. Both countries implemented a public campaign to raise awareness. In England registration on the Donor Register is voluntary. Registration was required in Netherlands for all residents 18 years and older. For those not already on the register, letters were sent by the Dutch Government to ask individuals to register. If people did not respond they would be legally registered as having "no objection." After implementation of Opt Out in England 42.3% is registered Opt In, 3.6% Opt Out, and 54.1% has no registration. In contrast in Netherlands the whole population is registered with 45% Opt In, 31% Opt Out and 24% "No Objection." It is too soon to draw conclusions about the impact on the consent rate and number of resulting organ donors. However, the first signs are positive.

Hyperbaric effects on cold stored whole blood following a military dive mission profile.

BACKGROUND: Whole blood (WB) is carried by special operations forces as part of a remote damage control resuscitation strategy. The effects of an underwater mission on the quality and coagulation profile of WB were simulated by exposure to hyperbaric pressures in a chamber. METHODS: WB units collected in CPDA-1 were exposed to three different combinations of hyperbaric pressure and duration of exposure: Group A 153.52 kPa (15.24 msw; 1.52 atm) for 4 h; n = 9, Group B 506.63 kPa (50.29 msw; 5.00 atm) for 1 h; n = 9, Group C 153.52 kPa (15.24 msw; 1.52 atm) for 1 h; n = 7. The following parameters were measured on each unit: prothrombin time/international normalized ratio, activated partial thromboplastin time, thromboelastography and concentration determinations of platelets, lactate, fibrinogen, and lactate dehydrogenase. Each sample underwent baseline, prepressurization, immediate postpressurization, and 6 h postpressurization laboratory testing. RESULTS: Six hours following hyperbaric exposure, the lactate concentration in group C was higher than prepressurization measurement and the platelet concentration in Group A was lower than prepressurization measurement. There were no changes in any of the other analyzed biochemical, coagulation and thromboelastogram parameters following exposure to hyperbaric stress. DISCUSSION: These data suggest that pressurization of WB up to 5 atm did not impact parameters tested. Changes observed in lactate and platelet count need further study, as well as complementary testing of red blood cell integrity. Further investigation of the hyperbaric extremes is necessary to determine if there is a damage inducing pressure to which WB should not be exposed.

Rate of D-alloimmunization in trauma does not depend on the number of RhD-positive units transfused: The BEST collaborative study.

BACKGROUND: Evidence indicates the life-saving benefits of early blood product transfusion in severe trauma resuscitation. Many of these products will be RhD-positive, so understanding the D-alloimmunization rate is important. METHODS: This was a multicenter, retrospective study whereby injured RhD-negative patients between 18-50 years of age who received at least one unit of RhD-positive red blood cells (RBC) or low titer group O whole blood (LTOWB) during their resuscitation between 1 January, 2010 through 31 December, 2019 were identified. If an antibody detection test was performed ≥14 days after the index RhD-positive transfusion then basic demographic information was collected, including whether the patient became D-alloimmunized. The overall D-alloimmunization rate, and the rate stratified by the number of units transfused, were calculated. RESULTS: Data were collected from nine institutions. Five institutions reported fewer than 10 eligible patients each and were excluded. From the remaining four institutions, all from the USA, there were 235 eligible patients; 77 (random effects estimate: 32.7%; 95% CI: 19.1-50.1%) became D-alloimmunized. Three of the institutions reported D-alloimmunization rates ≥38.6%, while the remaining institution's rate was 12.2%. In both random and fixed-effects models, the rate of D-alloimmunization was not significantly different between those who received one RhD-positive unit and those who received multiple RhD-positive units. CONCLUSION: In this large, multicenter study of injured patients, the overall rate of D-alloimmunization fell within the range previously reported. The rate of D-alloimmunization did not increase as the number of transfused RhD-positive units increased. These data can help to inform RhD type selection decisions.

Artificial Intelligence Enhances Diagnostic Flow Cytometry Workflow in the Detection of Minimal Residual Disease of Chronic Lymphocytic Leukemia.

Flow cytometric (FC) immunophenotyping is critical but time-consuming in diagnosing minimal residual disease (MRD). We evaluated whether human-in-the-loop artificial intelligence (AI) could improve the efficiency of clinical laboratories in detecting MRD in chronic lymphocytic leukemia (CLL). We developed deep neural networks (DNN) that were trained on a 10-color CLL MRD panel from treated CLL patients, including DNN trained on the full cohort of 202 patients (F-DNN) and DNN trained on 138 patients with low-event cases (MRD < 1000 events) (L-DNN). A hybrid DNN approach was utilized, with F-DNN and L-DNN applied sequentially to cases. "Ground truth" classification of CLL MRD was confirmed by expert analysis. The hybrid DNN approach demonstrated an overall accuracy of 97.1% (95% CI: 84.7−99.9%) in an independent cohort of 34 unknown samples. When CLL cells were reported as a percentage of total white blood cells, there was excellent correlation between the DNN and expert analysis [r > 0.999; Passing−Bablok slope = 0.997 (95% CI: 0.988−0.999) and intercept = 0.001 (95% CI: 0.000−0.001)]. Gating time was dramatically reduced to 12 s/case by DNN from 15 min/case by the manual process. The proposed DNN demonstrated high accuracy in CLL MRD detection and significantly improved workflow efficiency. Additional clinical validation is needed before it can be fully integrated into the existing clinical laboratory practice.

Real-time estimation of perceptual thresholds based on the electroencephalogram using a deep neural network.

BACKGROUND: Perceptual thresholds are measured in scientific and clinical setting to evaluate performance of the nervous system in essential tasks such as vision, hearing, touch, and registration of pain. Current procedures for estimating perceptual thresholds depend on the analysis of pairs of stimuli and participant responses, relying on the commitment and cognitive ability of subjects to respond accurately and consistently to stimulation. Here, we demonstrate that it is possible to measure the threshold for the perception of nociceptive stimuli based on non-invasively recorded brain activity alone using a deep neural network. NEW METHOD: For each stimulus, a trained deep neural network performed a 2-interval forced choice procedure, in which the network had to choose which of two time intervals in the electroencephalogram represented post-stimulus brain activity. Network responses were used to estimate the perceptual threshold in real-time using a psychophysical method of limits. COMPARISON WITH EXISTING METHODS: Network classification was able to match participants in reporting stimulus perception, resulting in average network-estimated perceptual thresholds that matched perceptual thresholds based on participant reports. RESULTS: The neural network successfully separated trials containing brain responses from trials without and could consistently estimate perceptual thresholds in real-time during a Go-/No-Go procedure and a counting task. CONCLUSION: Deep neural networks monitoring non-invasively recorded brain activity are now able to accurately predict stimulus perception and estimate the perceptual threshold in real-time without any verbal or motor response from the participant.