Rapidly adaptable automated interpretation of point-of-care COVID-19 diagnostics

BACKGROUND

Point-of-care diagnostic devices, such as lateral-flow assays, are becoming widely used by the public. However, efforts to ensure correct assay operation and result interpretation rely on hardware that cannot be easily scaled or image processing approaches requiring large training datasets, necessitating large numbers of tests and expert labeling with validated specimens for every new test kit format.

METHODS

We developed a software architecture called AutoAdapt POC that integrates automated membrane extraction, self-supervised learning, and few-shot learning to automate the interpretation of POC diagnostic tests using smartphone cameras in a scalable manner. A base model pre-trained on a single LFA kit is adapted to five different COVID-19 tests (three antigen, two antibody) using just 20 labeled images.

RESULTS

Here we show AutoAdapt POC to yield 99% to 100% accuracy over 726 tests (350 positive, 376 negative). In a COVID-19 drive-through study with 74 untrained users self-testing, 98% found image collection easy, and the rapidly adapted models achieved classification accuracies of 100% on both COVID-19 antigen and antibody test kits. Compared with traditional visual interpretation on 105 test kit results, the algorithm correctly identified 100% of images; without a false negative as interpreted by experts. Finally, compared to a traditional convolutional neural network trained on an HIV test kit, the algorithm showed high accuracy while requiring only 1/50th of the training images.

CONCLUSIONS

The study demonstrates how rapid domain adaptation in machine learning can provide quality assurance, linkage to care, and public health tracking for untrained users across diverse POC diagnostic tests.

OSA Patient Monitoring Based on the Beidou System

This paper presents an OSA patient interactive monitoring system based on the Beidou system. This system allows OSA patients to get timely rescue when they become sleepy outside. Because the Beidou position marker has an interactive function, it can reduce the anxiety of the patient while waiting for the rescue. At the same time, if a friend helps the OSA patients to call the doctor, the friend can also report the patient's condition in time. This system uses the popular IoT framework. At the bottom is the data acquisition layer, which uses wearable sensors to collect vital signs from patients, with a focus on ECG and SpO2 signals. The middle layer is the network layer that transmits the collected physiological signals to the Beidou indicator using the Bluetooth Low Energy (BLE) protocol. The top layer is the application layer, and the application layer uses the mature rescue interactive platform of Beidou. The Beidou system was developed by China itself, the main coverage of the satellite is in Asia, and is equipped with a high-density ground-based augmentation system. Therefore, the Beidou model improves the positioning accuracy and is equipped with a special communication satellite, which increases the short message interaction function. Therefore, patients can report disease progression in time while waiting for a rescue. After our simulation test, the effectiveness of the OSA patient rescue monitoring system based on the Beidou system and the positioning accuracy of OSA patients have been greatly improved. Especially when OSA patients work outdoors, the cell phone base station signal coverage is relatively weak. The satellite signal is well-covered, plus the SMS function of the Beidou indicator. Therefore, the system can be used to provide timely patient progress and provide data support for the medical rescue team to provide a more accurate rescue plan. After a comparative trial, the rescue rate of OSA patients using the detection device of this system was increased by 15 percentage points compared with the rescue rate using only GPS satellite phones.

Extended Kalman Filter Nonlinear Finite Element Method for Nonlinear Soft Tissue Deformation

BACKGROUND AND OBJECTIVE

Soft tissue modelling is crucial to surgery simulation. This paper introduces an innovative approach to realistic simulation of nonlinear deformation behaviours of biological soft tissues in real time.

METHODS

This approach combines the traditional nonlinear finite-element method (NFEM) and nonlinear Kalman filtering to address both physical fidelity and real-time performance for soft tissue modelling. It defines tissue mechanical deformation as a nonlinear filtering process for dynamic estimation of nonlinear deformation behaviours of biological tissues. Tissue mechanical deformation is discretized in space using NFEM in accordance with nonlinear elastic theory and in time using the central difference scheme to establish the nonlinear state-space models for dynamic filtering.

RESULTS

An extended Kalman filter is established to dynamically estimate nonlinear mechanical deformation of biological tissues. Interactive deformation of biological soft tissues with haptic feedback is accomplished as well for surgery simulation.

CONCLUSIONS

The proposed approach conquers the NFEM limitation of step computation but without trading off the modelling accuracy. It not only has a similar level of accuracy as NFEM, but also meets the real-time requirement for soft tissue modelling.

Rapid point-of-care testing methods/devices for meat species identification: A review

The authentication of animal species is an important issue due to an increasing trend of adulteration and mislabeling of animal species in processed meat products. Polymerase chain reaction is the most sensitive and specific technique for nucleic acid-based animal species detection. However, it is a time-consuming technique that requires costly thermocyclers and sophisticated labs. In recent times, there is a need of on-site detection by point-of-care (POC) testing methods and devices under low-resource settings. These POC devices must be affordable, sensitive, specific, user-friendly, rapid and robust, equipment free, and delivered to the end users. POC devices should also confirm the concept of micro total analysis system. This review discusses POC testing methods and devices that have been developed for meat species identification. Recent developments in lateral flow assay-based devices for the identification of animal species in meat products are also reviewed. Advancements in increasing the efficiency of lateral flow detection are also discussed.

Algorithms for immunochromatographic assay: review and impact on future application

Lateral flow immunoassay (LFIA) is a critical choice for applications of point-of-care testing (POCT) in clinical and laboratory environments because of its excellent features and versatility. To obtain authentic values of analyte concentrations and reliable detection results, the relevant research has featured the application of a diversity of methods of mathematical analysis to technical analysis to allow for use with a small quantity of data. Accordingly, a number of signal and image processing strategies have also emerged for the application of gold immunochromatographic and fluorescent strips to improve sensitivity and overcome the limitations of correlative hardware systems. Instead of traditional methods to solve the problem, researchers nowadays are interested in machine learning and its more powerful variant, deep learning technology, for LFIA detection. This review emphasizes different models for the POCT of accurate labels as well as signal processing strategies that use artificial intelligence and machine learning. We focus on the analytical mechanism, procedural flow, and the results of the assay, and conclude by summarizing the advantages and limitations of each algorithm. We also discuss the potential for application of and directions of future research on LFIA technology when combined with Artificial Intelligence and deep learning.

Analysis of linear lung models based on state-space models

BACKGROUND AND OBJECTIVES

Linear parametric respiratory system models have been used in the model-based analysis of the respiratory system. Although there are studies exploring the physiological correctness and fitting accuracy of the models, they are not analysed in terms of interaction between parameters and dynamics of the model. In this study we propose to use state-space modelling to yield the time-varying nature of the system incorporated by the parameters.

METHODS

We tested controllability, observability and stability characteristics of the equation of motion, 2-comp. parallel, 2-comp. series, viscoelastic, 6-element and mead models while using the parameters given in the literature. In the sensitivity analysis we proposed to use dual Desensitized Linear Kalman Filter (DKF) and Extended Kalman Filter (EKF) method. In this method, state error covariance revealed the parameter sensitivities for each model.

RESULTS

Results showed that all models, except 2-comp. parallel and mead models, are both controllable and observable models. On the other hand all models, except mead model, are stable models. Regarding to the sensitivity analysis, dual DKF - EKF method estimated states of the models successfully with a low estimation error. Sensitivity analysis results showed that airway parameters have higher effects on the state estimation than the other parameters have.

CONCLUSION

We proved that state-space evaluation of the previously proposed parametric models of the respiratory system led us to quantitative and qualitative assessments of the respiratory models. Moreover parameter values found in the literature have different effects on the models.

Improving Lateral Flow Assay Performance Using Computational Modeling

The performance, field utility, and low cost of lateral flow assays (LFAs) have driven a tremendous shift in global health care practices by enabling diagnostic testing in previously unserved settings. This success has motivated the continued improvement of LFAs through increasingly sophisticated materials and reagents. However, our mechanistic understanding of the underlying processes that drive the informed design of these systems has not received commensurate attention. Here, we review the principles underpinning LFAs and the historical evolution of theory to predict their performance. As this theory is integrated into computational models and becomes testable, the criteria for quantifying performance and validating predictive power are critical. The integration of computational design with LFA development offers a promising and coherent framework to choose from an increasing number of novel materials, techniques, and reagents to deliver the low-cost, high-fidelity assays of the future.

Mathematical Modeling of Bioassays

The high affinity and specificity of biological receptors determine the demand for and the intensive development of analytical systems based on use of these receptors. Therefore, theoretical concepts of the mechanisms of these systems, quantitative parameters of their reactions, and relationships between their characteristics and ligand-receptor interactions have become extremely important. Many mathematical models describing different bioassay formats have been proposed. However, there is almost no information on the comparative characteristics of these models, their assumptions, and predictive insights. In this review we suggested a set of criteria to classify various bioassays and reviewed classical and contemporary publications on these bioassays with special emphasis on immunochemical analysis systems as the most common and in-demand techniques. The possibilities of analytical and numerical modeling are discussed, as well as estimations of the minimum concentrations that may be detected in bioassays and recommendations for the choice of assay conditions.

An improved detection limit and working range of lateral flow assays based on a mathematical model

The detection limit and working range of lateral flow assays are investigated experimentally and numerically.

ADAPTIVE UNSCENTED KALMAN FILTER FOR ONLINE SOFT TISSUES CHARACTERIZATION

Online soft tissue characterization is important for robotic-assisted minimally invasive surgery to achieve precise and stable robotic control with haptic feedback. This paper presents a new adaptive unscented Kalman filter based on the nonlinear Hunt–Crossley model for online soft tissue characterization without requiring the characteristics of system noise. This filter incorporates the concept of Sage windowing in the traditional unscented Kalman filter to adaptively estimate system noise covariance using predicted residuals within a time window. In order to account for the inherent relationship between the current and previous states of soft tissue deformation involved in robotic-assisted surgery and improve the estimation performance, a recursive estimation of system noise covariance is further constructed by introducing a fading scaling factor to control the contributions between noise covariance estimations at current and previous time points. The proposed adaptive unscented Kalman filter overcomes the limitation of the traditional unscented Kalman filter in requiring the characteristics of system noise. Simulations and comparisons show the efficacy of the suggested nonlinear adaptive unscented Kalman filter for online soft tissue characterization.

Mathematical Model of Serodiagnostic Immunochromatographic Assay

This article describes the mathematical model for an immunochromatographic assay for the detection of specific immunoglobulins against a target antigen (antibodies) in blood/serum (serodiagnosis). The model utilizes an analytical (non-numerical) approach and allows the calculation of the kinetics of immune complexes' formation in a continuous-flow system using commonly available software, such as Microsoft Excel. The developed model could identify the nature of the influence of immunochemical interaction constants and reagent concentrations on the kinetics of the formation of the detected target complex. On the basis of the model, recommendations are developed to decrease the detection limit for an immunochromatographic assay of specific immunoglobulins.

Image-based quantitative analysis of gold immunochromatographic strip via cellular neural network approach

Gold immunochromatographic strip assay provides a rapid, simple, single-copy and on-site way to detect the presence or absence of the target analyte. This paper aims to develop a method for accurately segmenting the test line and control line of the gold immunochromatographic strip (GICS) image for quantitatively determining the trace concentrations in the specimen, which can lead to more functional information than the traditional qualitative or semi-quantitative strip assay. The canny operator as well as the mathematical morphology method is used to detect and extract the GICS reading-window. Then, the test line and control line of the GICS reading-window are segmented by the cellular neural network (CNN) algorithm, where the template parameters of the CNN are designed by the switching particle swarm optimization (SPSO) algorithm for improving the performance of the CNN. It is shown that the SPSO-based CNN offers a robust method for accurately segmenting the test and control lines, and therefore serves as a novel image methodology for the interpretation of GICS. Furthermore, quantitative comparison is carried out among four algorithms in terms of the peak signal-to-noise ratio. It is concluded that the proposed CNN algorithm gives higher accuracy and the CNN is capable of parallelism and analog very-large-scale integration implementation within a remarkably efficient time.

Modeling Throughput of Emergency Departments via Time Series

In this article, the expectation maximization (EM) algorithm is applied for modeling the throughput of emergency departments via available time-series data. The dynamics of emergency department throughput is developed and evaluated, for the first time, as a stochastic dynamic model that consists of the noisy measurement and first-order autoregressive (AR) stochastic dynamic process. By using the EM algorithm, the model parameters, the actual throughput, as well as the noise intensity, can be identified simultaneously. Four real-world time series collected from an emergency department in West London are employed to demonstrate the effectiveness of the introduced algorithm. Several quantitative indices are proposed to evaluate the inferred models. The simulation shows that the identified model fits the data very well.

Time Series Modeling of Nano-Gold Immunochromatographic Assay via Expectation Maximization Algorithm

In this paper, the expectation maximization (EM) algorithm is applied to the modeling of the nano-gold immunochromatographic assay (nano-GICA) via available time series of the measured signal intensities of the test and control lines. The model for the nano-GICA is developed as the stochastic dynamic model that consists of a first-order autoregressive stochastic dynamic process and a noisy measurement. By using the EM algorithm, the model parameters, the actual signal intensities of the test and control lines, as well as the noise intensity can be identified simultaneously. Three different time series data sets concerning the target concentrations are employed to demonstrate the effectiveness of the introduced algorithm. Several indices are also proposed to evaluate the inferred models. It is shown that the model fits the data very well.

An improved constraint filtering technique for inferring hidden states and parameters of a biological model

MOTIVATION

In systems biology, kinetic models represent the biological system using a set of ordinary differential equations (ODEs). The correct values of the parameters within these ODEs are critical for a reliable study of the dynamic behaviour of such systems. Typically, it is only possible to experimentally measure a fraction of these parameter values. The rest must be indirectly determined from measurements of other quantities. In this article, we propose a novel statistical inference technique to computationally estimate these unknown parameter values. By characterizing the ODEs with non-linear state-space equations, this inference technique models the unknown parameters as hidden states, which can then be estimated from noisy measurement data.

RESULTS

Here we extended the square-root unscented Kalman filter SR-UKF proposed by Merwe and Wan to include constraints with the state estimation process. We developed the constrained square-root unscented Kalman filter (CSUKF) to estimate parameters of non-linear state-space models. This probabilistic inference technique was successfully used to estimate parameters of a glycolysis model in yeast and a gene regulatory network. We showed that our method is numerically stable and can reliably estimate parameters within a biologically meaningful parameter space from noisy observations. When compared with the two common non-linear extensions of Kalman filter in addition to four widely used global optimization algorithms, CSUKF is shown to be both accurate and computationally efficient. With CSUKF, statistical analysis is straightforward, as it directly provides the uncertainty on the estimation result.

AVAILABILITY AND IMPLEMENTATION

Matlab code available upon request from the author.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A hybrid EKF and switching PSO algorithm for joint state and parameter estimation of lateral flow immunoassay models

In this paper, a hybrid extended Kalman filter (EKF) and switching particle swarm optimization (SPSO) algorithm is proposed for jointly estimating both the parameters and states of the lateral flow immunoassay model through available short time-series measurement. Our proposed method generalizes the well-known EKF algorithm by imposing physical constraints on the system states. Note that the state constraints are encountered very often in practice that give rise to considerable difficulties in system analysis and design. The main purpose of this paper is to handle the dynamic modeling problem with state constraints by combining the extended Kalman filtering and constrained optimization algorithms via the maximization probability method. More specifically, a recently developed SPSO algorithm is used to cope with the constrained optimization problem by converting it into an unconstrained optimization one through adding a penalty term to the objective function. The proposed algorithm is then employed to simultaneously identify the parameters and states of a lateral flow immunoassay model. It is shown that the proposed algorithm gives much improved performance over the traditional EKF method.