Series Combination Research Articles

Studying human-AI collaboration protocols: the case of the Kasparov\u2019s law in radiological double reading

PurposeThe integration of Artificial Intelligence into medical practices has recently been advocated for the promise to bring increased efficiency and effectiveness to these practices. Nonetheless, little research has so far been aimed at understanding the best human-AI interaction protocols in collaborative tasks, even in currently more viable settings, like independent double-reading screening tasks.MethodsTo this aim, we report about a retrospective case–control study, involving 12 board-certified radiologists, in the detection of knee lesions by means of Magnetic Resonance Imaging, in which we simulated the serial combination of two Deep Learning models with humans in eight double-reading protocols. Inspired by the so-called Kasparov’s Laws, we investigate whether the combination of humans and AI models could achieve better performance than AI models alone, and whether weak reader, when supported by fit-for-use interaction protocols, could out-perform stronger readers.ResultsWe discuss two main findings: groups of humans who perform significantly worse than a state-of-the-art AI can significantly outperform it if their judgements are aggregated by majority voting (in concordance with the first part of the Kasparov’s law); small ensembles of significantly weaker readers can significantly outperform teams of stronger readers, supported by the same computational tool, when the judgments of the former ones are combined within “fit-for-use” protocols (in concordance with the second part of the Kasparov’s law).ConclusionOur study shows that good interaction protocols can guarantee improved decision performance that easily surpasses the performance of individual agents, even of realistic super-human AI systems. This finding highlights the importance of focusing on how to guarantee better co-operation within human-AI teams, so to enable safer and more human sustainable care practices.

Development and validation of a novel index for the differential diagnosis of corticotropin-dependent Cushing syndrome.

To develop an index for the differential diagnosis of corticotropin-dependent Cushing syndrome (CS). The development cohort included 112 consecutive patients with clinicopathologically confirmed corticotropin-dependent CS at the Department of Endocrinology and Metabolism, Tianjin Medical University General Hospital, from December 2004 to May 2020, and data of 126 patients from studies published from 2016 to August 2020, identified through search in PubMed, Embase and the Cochrane Library, was extracted for external validation. The index was calculated as the product of plasma adrenocorticotropic hormone (ACTH, pmol/L) and urinary free cortisol (UFC, nmol/24h) divided by 10,000. The discriminative ability was tested using receiver operating characteristics (ROC) curve analysis. In development cohort, area under curve of ROC analysis of the ACTH-UFC index in identifying Cushing disease (CD) was 0.977. The diagnostic accuracy of ACTH-UFC index ≤ 11 was comparable to that of 48h 8mg/d high-dose dexamethasone test (HDDST) in identifying CD, with sensitivity, specificity, positive and negative likelihood ratios of 96.6%, 87.5%, 7.73, and 0.04, respectively. The sensitivity of ACTH-UFC index ≤ 11 in parallel combination with pituitary magnetic resonance imaging (MRI) was 100% for identifying CD. The performance of the ACTH-UFC index in parallel or serial combination with pituitary MRI was similar in the validation cohort. ACTH-UFC index provides a rapid, convenient and non-invasive adjunctive approach for the differential diagnosis of corticotropin-dependent CS, with no risk of aggravating metabolic disturbances. Investigations for ectopic causes of corticotropin-dependent CS should be performed with ACTH-UFC index > 11 and negative contrasted pituitary MRI.

Serial Combination Optimization Method of Packet Transport Network

Packet transport network (PTN) has problems such as waste of resources and low network stability due to the excessive complexity of the existing network or improper network architecture design. The optimization of the transport networks can not only make the network structure more reasonable but also reduce all kinds of unexpected scenarios in the network operation, improving the network efficiency and reducing the failure rate. This research will be optimized from three aspects. (1) In order to solve the problem of the same active and standby routing in the existing network, an optimization algorithm for the same active and standby routing of LSP is proposed. The essence of the optimization algorithm is to search the existing routing using the K‐shortest path (KSP) between two network nodes as protection routing for LSP protection. (2) Aiming at the link with a high CIR bandwidth occupancy rate, a method is completed without adding optical fibers and other physical resources; an optimization method for the committed information rate bandwidth occupancy rate based on the KSP algorithm is proposed. (3) When the PTN ring formation rate is low, the security of the PTN is seriously reduced. In order to solve the problem of low ring formation rate in the network, this paper proposes a ring formation rate optimization scheme for PTN access layer equipment based on network elements accounting income. Through the experimental verification on the mobile PTN in one city, Hubei Province, the combination optimization method can improve the network LSP protection rate by 24%, the CIR bandwidth occupancy rate is reduced by 13.82%, and the nonring forming rate was reduced by 17.9%. This method improved network stability, reducing the risk of failure in service transportation effectively.

A Sequential Algorithm Combining ADAPT and Liver Stiffness Can Stage Metabolic-Associated Fatty Liver Disease in Hospital-Based and Primary Care Patients.

Metabolic-associated fatty liver disease is common, with fibrosis the major determinant of adverse outcomes. Population-based screening tools with high diagnostic accuracy for the staging of fibrosis are lacking. Three independent cohorts, 2 with both liver biopsy and liver stiffness measurements (LSMs, n = 254 and 65) and a population sample (n = 713), were studied. The performance of a recently developed noninvasive algorithm (ADAPT [age, diabetes, PRO-C3 and platelets panel]) as well as aspartate aminotransferase-to-platelet ratio index, fibrosis-4, nonalcoholic fatty liver disease fibrosis score, and LSM was used to stage patients for significant (≥F2) and advanced (≥F3) fibrosis. In the hospital-based cohorts, the N-terminal propeptide of type 3 collagen (Pro-C3) increased with fibrosis stage (P < 0.0001) and independently associated with advanced fibrosis (odds ratio = 1.091, 95% confidence interval [CI]: 1.053-1.113, P = 0.0001). ADAPT showed areas under the receiver operating characteristics curve of 0.831 (95% CI: 0.779-0.875) in the derivation and 0.879 (95% CI: 0.774-0.946) in the validation cohort for advanced fibrosis. This was superior to the existing fibrosis scores, aspartate aminotransferase-to-platelet ratio index, fibrosis-4, BARD (BMI, aspartate aminotransferase to alanine aminotransferase ratio [AAR], diabetes), and nonalcoholic fatty liver disease fibrosis score in most comparisons and comparable with LSM. Serial use of ADAPT and LSM had diagnostic accuracy of 92.5%, with 98% and 100% negative predictive value in the derivation and validation cohorts, respectively. In the population cohort, PRO-C3 associated with advanced fibrosis (P = 0.04), while ADAPT had a negative predictive value of 98% for excluding advanced fibrosis. PRO-C3 and ADAPT reliably exclude advanced fibrosis in low-risk populations. The serial combination of ADAPT with LSM has high diagnostic accuracy with a low requirement for liver biopsy. The proposed algorithm would help stratify those who need biopsies and narrow down those patients who would need to be referred to specialty clinics.

Combination of multiple classifiers for automatic recognition of diseases and damages on plant leaves

In this paper, we present an automatic recognition system of diseases and damages on plant leaves. The proposed system is based on classifiers combination technique, in which we have two variants of combination: serial combination of two classifiers, and hybrid combination of three classifiers including a serial combination of two classifiers in parallel with an individual classifier. Three types of features are adopted including color, texture and shape. The tests of this study to evaluate the three variants of combination are carried out on a database of 600 images of six classes (Leaf miners, Tuta absoluta and Thrips, Early blight, Late blight and powdery mildew). The comparison of results between the two methods serial and hybrid of the proposed system indicates that significant performances were obtained by applying the hybrid method for the recognition of diseases and damages on plant leaves.

Photothermally Modulatable and Structurally Disintegratable Sub-8-nm Au1Ag9 Embedded Nanoblocks for Combination Cancer Therapy Produced by Plug-in Assembly

As well as the exploration of translatable delivery nanosystems for cancer therapeutic agents, the development of automatable continuous-flow manufacturing technology comprising digitally controlled reactions for the on-demand production of pharmaceuticals is an important challenge in anticancer nanomedicine. Most attempts to resolve these issues have involved the development of alternative reactions, formulations, or constructs containing stimulus components aimed at producing multiple approaches for highly efficacious combination cancer therapies. However, there has been no report of a platform based on plug-in execution that enables continuous-flow manufacture in a compact, reconfigurable manner, although an optimal platform technology may be a prerequisite for the timely translation of recently developed nanomedicines. To this end, we describe the development of a platform toward digitizable, continuous manufacture by a serial combination of plug-in reactionwares (heating plates, a spraying cup, and a photochamber) for single-pass flow fabrication. Specifically, we fabricated three different composite nanoblocks consisting of Au1Ag9 (<8 nm; stimulus component), docetaxel (an anticancer drug), and bovine serum albumin (a protective and targeting agent) using our system, with the result of producing nanoblocks with photothermally modulatable and structurally disintegratable properties. These were examined for effectiveness in near-infrared-induced chemothermal cancer therapy and renal excretion of Au1Ag9 particles and exhibited high anticancer efficacy and warrantable biosafety.

Coherency vector formalism for polarimetric transformations

Despite the virtues of Jones and Mueller formalisms for the representation of the polarimetric properties, for some purposes in both Optics and SAR Polarimetry, the concept of coherency vector associated with a nondepolarizing medium has proven to be an useful mathematical structure that inherits certain symmetries underlying the nature of linear polarimetric transformations of the states of polarization of light caused by its interaction with material media. While the Jones and Mueller matrices of a serial combination of devices are given by the respective conventional matrix products, the composition of coherency vectors of such serial combinations requires a specific and unconventional mathematical rule. In this work, a vector product of coherency vectors is presented that satisfies, in a meaningful and consistent manner, the indicated requirements. As a result, a new algebraic formalism is built where the representation of polarization states of electromagnetic waves through Stokes vectors is preserved, while nondepolarizing media are represented by coherency vectors and general media are represented by coherency matrices generated by partially coherent compositions of the coherency vectors of the constituents.

Multi-Source Neural Model for Machine Translation of Agglutinative Language

Benefitting from the rapid development of artificial intelligence (AI) and deep learning, the machine translation task based on neural networks has achieved impressive performance in many high-resource language pairs. However, the neural machine translation (NMT) models still struggle in the translation task on agglutinative languages with complex morphology and limited resources. Inspired by the finding that utilizing the source-side linguistic knowledge can further improve the NMT performance, we propose a multi-source neural model that employs two separate encoders to encode the source word sequence and the linguistic feature sequences. Compared with the standard NMT model, we utilize an additional encoder to incorporate the linguistic features of lemma, part-of-speech (POS) tag, and morphological tag by extending the input embedding layer of the encoder. Moreover, we use a serial combination method to integrate the conditional information from the encoders with the outputs of the decoder, which aims to enhance the neural model to learn a high-quality context representation of the source sentence. Experimental results show that our approach is effective for the agglutinative language translation, which achieves the highest improvements of +2.4 BLEU points on Turkish–English translation task and +0.6 BLEU points on Uyghur–Chinese translation task.

Ensemble feature selection in high dimension, low sample size datasets: Parallel and serial combination approaches

Feature selection in high dimension, low sample size (HDLSS) data is always an important data pre-processing task. In the literature, the concept of ensemble learning has been applied to improve single feature selection methods, the so-called ensemble feature selection techniques. The most widely used approach is to combine multiple feature selection methods and their selection results via some sort of aggregation function in a parallel manner. Another ensemble strategy is based on the serial combination approach where the selection results of the first feature selection stage are used as input for the second stage of feature selection to produce the final output. The aim of this paper is to fully explore the performance of parallel and serial combination approaches for ensemble feature selection over HDLSS data. In particular, we strive to answer two research questions: whether parallel and serial based ensemble feature selection can outperform single feature selection and which combination approach is the better choice for ensemble feature selection. The experimental results based on comparing nine parallel and nine serial combinations, as well as three single baseline feature selection methods, including principal component analysis (PCA), genetic algorithm (GA), and C4.5 decision tree, show that ensemble feature selection performs better than single feature selection in terms of classification accuracy. However, there are no significant differences in performance between the single best baseline method (i.e. GA) and the top three parallel and serial combinations. On the other hand, the serial combination approach produces the largest feature reduction rate.

Pharmacological Investigation of Serial Anxiety Tests in the Mouse: A Pilot Study

The use of test batteries has been suggested as a means to enhance throughput, to broaden the spectrum of anxiety parameters detectable and to minimise the numbers of experimental animals required in behavioural studies. This study was undertaken to determine the sensitivity of test batteries each of open field, zero mazes and staircase tests to anxiety behaviours in randomised mouse groups. Mice (n = 8) were exposed to these apparatuses serially in that order, thirty minutes following intraperitoneal administration of distilled water, 0.5 and 1.0 mg/kg diazepam. Each mouse was allowed to spend 5 minutes to freely explore each test apparatus. The results showed diazepam-treated mice exhibited significantly (p&lt;0.05) reduced anxiety behaviours compared to the placebo group on most rodent anxiety parameters evaluated. The findings of this study suggest these behavioural test apparatuses, when used in serial combination, are sensitive and reliable to measure murine anxiety-related behaviours and the anxiolytic effects of standard/putative agents.

Clinical Profile and the Role of Rapid Serological Tests: Typhifast IgM and Enterocheck WB in the Diagnosis of Typhoid Fever.

ABSTRACTIntroductionDuring identification and diagnosis, typhoid fever (TF) causes various issues such as nonspecific symptoms and nontyphoidal Salmonella-associated febrile diseases. Accurate identification and diagnosis are still a substantial concern. The current study was undertaken to study the clinical profile of TF and the role of Typhifast IgM and Enterocheck WB in early diagnosis.Materials and methodsClinically suspected TF patients (121) were included in the study. Patients with antibiotic history during the period of febrile illness were excluded. The diagnosis was confirmed with blood culture test. Widal test and two new rapid serological tests: Typhifast IgM and Enterocheck WB were performed. The outcomes were compared with blood culture-confirmed cases to derive the sensitivity and specificity of the diagnostic tests. The clinical characteristics were compared with diagnostic tests using Chi-square test.ResultsThe most common presentations of TF were fever, chills, vomiting, abdominal pain, anorexia, constipation, and diarrhea. Among the 121 clinically suspected TF patients, 67 had positive blood culture tests for Salmonella typhi. The Typhifast IgM and Enterocheck WB showed sensitivity of 97.01% and 68.52%, respectively. Specificity was also more with Typhifast IgM (85.07%) than with Enterocheck WB (62.96%). When tests were used in parallel combination, 97.01% sensitivity was attained, while the specificity dropped to 46.30%. When used in serial combination, sensitivity of 85.07% and specificity of 85.19% were observed.ConclusionStepladder fever, abnormal serum glutamic oxaloacetic transaminase (SGOT), and abnormal albumin are associated with blood culture, Typhifast IgM, and Enterocheck WB tests. Rapid serological tests might assist in accurate and early identification of TF.How to cite this articleBhume RJ, Babaliche P. Clinical Profile and the Role of Rapid Serological Tests: Typhifast IgM and Enterocheck WB in the Diagnosis of Typhoid Fever. Indian J Crit Care Med 2020;24(5):307–312.

Neuro-fuzzy iterative learning control for 4-poster test rig

In this paper, a new control method is presented for the 4-poster test systems. The primary aim of the paper is to improve the convergence speed and decrease the error rate for model-based iterative learning control (ILC), a widely used method as a tracking control. First, the dynamic equations of the system are generated, and the control problem is formulated. Then, an inverse model of the system is established directly through the adaptive neuro-fuzzy inference system (ANFIS) with auxiliary parameter (piston position) as a serial combination of two sub-models. In order to construct a neuro-fuzzy ILC (NFILC) structure, these sub-models are integrated into the neuro-fuzzy inverse controller (NFIC). Because of this new structure, the modified ILC rule has two layers. In the first layer, the controlled parameter, namely, the acceleration is iterated, whereas, in the second layer, the auxiliary parameter is iterated. The outcomes of the proposed control method are scrutinized by testing through a numerical simulation. Finally, it is demonstrated that the modified ILC rule dramatically increase the convergence speed and reduce the final error rate.

Algorithm for the numerical calculation of the serial components of the normal form of depolarizing Mueller matrices.

The normal form of a depolarizing Mueller matrix constitutes an important tool for the phenomenological interpretation of experimental polarimetric data. Due to its structure as a serial combination of three Mueller matrices, namely a canonical depolarizing Mueller matrix sandwiched between two pure (nondepolarizing) Mueller matrices, it overcomes the necessity of making a priori choices on the order of the polarimetric components, as this occurs in other serial decompositions. Because Mueller polarimetry addresses more and more applications in a wide range of areas in science, engineering, medicine, etc., the normal form decomposition has an enormous potential for the analysis of experimentally determined Mueller matrices. However, its systematic use has been limited to some extent because of the lack of numerical procedure for the calculation of each polarimetric component, in particular in the case of Type II Mueller matrices. In this work, an efficient algorithm applicable to the decomposition of both Type II and Type I Mueller matrices is presented.

Construction of a method for the structural functional-cost modeling of a complex hierarchic system

A method has been proposed for the structural functional-cost modeling of a complex hierarchical system. The initial data for carrying out calculations directly based on the functional-cost model have been determined. We have proposed and substantiated the cost description of a complex system and its components by using analytical approximating dependences. An example of the functional-cost algorithm has been given that employs a Lagrange multiplier method for complex systems with a serial combination of its separate parts. The solution to the example is the distribution among the desired probabilities of the effective operation of individual parts in terms of the minimum cost. Deriving such distribution does not require absolute values of the cost of both parts and the entire system. The issues addressed in the cost rationalization include the following: ensuring the predefined level of the functional perfection of a system at its minimum cost; determining the minimum required level of functional excellence in a single link at the known levels of functional excellence of the system and all other links except the one under investigation; determining the required number of parallel operating links for the same purpose; clarification of the required level of the functional perfection of links (information sensors, information processing links, communication channels) that have parallel communication; the structural improvement of a complex system by selecting a link within the system for which the improvement of functional perfection can be realized at minimum cost. We have proposed rules for the structural rationalization of a complex system. The first of them is the rule of the rational structural structure of a complex system. That makes it possible to receive a sufficient benefit from the complex system at minimum cost. The second rule is the expediency of complicating a complex system. According to it, complicating a complex system is advisable only if it improves the functional perfection of the entire complex system. The third rule, a rule of the proper structure, shows that there are no unnecessary links in the complex system, that is, those links that do not perform any activities that are not functionally required by a given system

Enhanced efficacy of anti-VEGFR2/taxane therapy after progression on immune checkpoint inhibition (ICI) in patients (pts) with metastatic gastroesophageal adenocarcinoma (mGEA).

362 Background: Most pts with mGEA do not respond to ICI or ramucirumab/paclitaxel (RAM/TAX). Emerging data suggest that ICI may potentiate subsequent therapy (Tx). In KN059 we unexpectedly observed durable responses in 2 pts on RAM/TAX after ICI (PMID 29674442). We examined if ICI impacts efficacy of subsequent RAM/TAX in a larger cohort and explored alterations in the tumor microenvironment. Methods: All pts with mGEA at Mayo Clinic (MN, AZ, FL) who received RAM/TAX (2014-19) were included. We compared best objective response rate (ORR: complete [CR] or partial response [PR]) per RECIST1.1 and overall survival (OS) in pts who received RAM/TAX with (Group A) vs without (Group B) immediately preceding PD-1 blockade. Chi square and multivariate logistic and Cox regression were used. We adjusted for total lines of Tx received and the line of Tx in which RAM/TAX was given, to control for potential differences in the biology of heavily treated pts or those receiving RAM/TAX as 2nd vs 3rd line Tx. We also adjusted for age and ECOG PS. Results: In total 87 pts met inclusion criteria. In the 19 pts (Group A) who received RAM/TAX (usually as 3rd line Tx) after progression on ICI, there was a 77% relative risk reduction of death after initiation of RAM/TAX compared to the 68 pts (Group B) who received RAM/TAX (usually as 2nd line) without preceding ICI (median OS 15.0 vs 7.6 mo; HR 0.23; P &lt; .001). The ORR was also significantly higher (58% vs 18%; P &lt; .001) including the CR rate (16% vs 1%; P = .006). Two CRs in Group A are ongoing (10 - 34 mo). Of note, 95% of Group A pts did not respond to ICI, and all had irRECIST progression on ICI. Immunofluorescent analysis from a responder showed 65-fold reduction (post vs pre-Tx) in intratumoral density of immunosuppressive FOXP3+ Tregs, with preserved density of antitumor CD8+ T cells. Conclusions: Higher than expected efficacy was observed on RAM/TAX when immediately preceded by ICI, suggesting ICI may enhance efficacy of subsequent anti-VEGFR/taxane therapy. This serial immunotherapy combination may be a novel option for pts with primary resistance to ICI and will be tested prospectively in a new randomized phase 2 trial (NCT04069273).

DNN-Based Cepstral Excitation Manipulation for Speech Enhancement

This contribution aims at speech model-based speech enhancement by exploiting the source-filter model of human speech production. The proposed method enhances the excitation signal in the cepstral domain by making use of a deep neural network (DNN). We investigate two types of target representations along with the significant effects of their normalization. The new approach exceeds the performance of a formerly introduced classical signal processing-based cepstral excitation manipulation (CEM) method in terms of noise attenuation by about 1.5 dB. We show that this gain also holds true when comparing serial combinations of envelope and excitation enhancement. In the important low-SNR conditions, no significant trade-off for speech component quality or speech intelligibility is induced, while allowing for substantially higher noise attenuation. In total, a traditional purely statistical state-of-the-art speech enhancement system is outperformed by more than 3 dB noise attenuation.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya.

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection in children, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children presenting with fever (axillary temperature ≥ 37.5 0C) to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants' diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to >98%. Conclusions: LM and RDT afford high Se and Sp in symptomatic care-seeking children in this low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.

СЕРЦЕВА НЕДОСТАТНІСТЬ: МІСЦЕ БІОМАРКЕРІВ NT-proBNP ТА sST 2

Heart failure is the most common disease in recent decades. Optimizing the diagnosis and prognosis of various types of heart failure is an important task of modern medicine. The use of specific biomarkers to determine the diagnosis of heart failure, prognosis of the disease is an acute problem. To date, known biomarkers are natriuretic peptides and ST2 and its isoforms. These are highly sensitive biomarkers that allow you to personalize the risk of complications in a patient with heart failure. The article discusses both the single, single use of individual biomarkers, as well as the serial identification and combination of biomarkers to evaluate the prognosis.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection in children, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children presenting with fever (axillary temperature ≥ 37.50C) to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants’ diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to &gt;98%. Conclusions: LM and RDT afford high Se and Sp in a low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection in children, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children presenting with fever (axillary temperature ≥ 37.5 0C) to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants’ diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to >98%. Conclusions: LM and RDT afford high Se and Sp in symptomatic care-seeking children in this low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.