Comparison of the Diagnostic Performance of Deep Learning Algorithms for Reducing the Time Required for COVID-19 RT-PCR Testing.

(1) Background: Rapid and accurate negative discrimination enables efficient management of scarce isolated bed resources and adequate patient accommodation in the majority of areas experiencing an explosion of confirmed cases due to Omicron mutations. Until now, methods for artificial intelligence or deep learning to replace time-consuming RT-PCR have relied on CXR, chest CT, blood test results, or clinical information. (2) Methods: We proposed and compared five different types of deep learning algorithms (RNN, LSTM, Bi-LSTM, GRU, and transformer) for reducing the time required for RT-PCR diagnosis by learning the change in fluorescence value derived over time during the RT-PCR process. (3) Results: Among the five deep learning algorithms capable of training time series data, Bi-LSTM and GRU were shown to be able to decrease the time required for RT-PCR diagnosis by half or by 25% without significantly impairing the diagnostic performance of the COVID-19 RT-PCR test. (4) Conclusions: The diagnostic performance of the model developed in this study when 40 cycles of RT-PCR are used for diagnosis shows the possibility of nearly halving the time required for RT-PCR diagnosis.

Electrochemical Gene Amplification Signal Detection of Disposable Biochips Using Electrodes.

Real-time Polymerase Chain Reaction (RT-PCR), a molecular diagnostic technology, is spotlighted as one of the quickest and fastest diagnostic methods for the actual coronavirus (SARS-CoV-2). However, the fluorescent label-based technology of the RT-PCR technique requires expensive equipment and a sample pretreatment process for analysis. Therefore, this paper proposes a biochip based on Electrochemical Impedance Spectroscopy (EIS). In this paper, it was possible to see the change according to the concentration by measuring the impedance with a chip made of two electrodes with different shapes of sample DNA.

Compact Camera Fluorescence Detector for Parallel-Light Lens-Based Real-Time PCR System.

The polymerase chain reaction is an important technique in biological research. However, it is time consuming and has a number of disadvantages. Therefore, real-time PCR technology that can be used in real-time monitoring has emerged, and many studies are being conducted regarding its use. Real-time PCR requires many optical components and imaging devices such as expensive, high-performance cameras. Therefore, its cost and assembly process are limitations to its use. Currently, due to the development of smart camera devices, small, inexpensive cameras and various lenses are being developed. In this paper, we present a Compact Camera Fluorescence Detector for use in parallel-light lens-based real-time PCR devices. The proposed system has a simple optical structure, the system cost can be reduced, and the size can be miniaturized. This system only incorporates Fresnel lenses without additional optics in order for the same field of view to be achieved for 25 tubes. In the center of the Fresnel lens, one LED and a complementary metal-oxide semiconductor camera were placed in directions that were as similar as possible. In addition, to achieve the accurate analysis of the results, image processing was used to correct them. As a result of an experiment using a reference fluorescent substance and double-distilled water, it was confirmed that stable fluorescence detection was possible.

The application of a deep learning system developed to reduce the time for RT-PCR in COVID-19 detection.

Reducing the time to diagnose COVID-19 helps to manage insufficient isolation-bed resources and adequately accommodate critically ill patients. There is currently no alternative method to real-time reverse transcriptase polymerase chain reaction (RT-PCR), which requires 40 cycles to diagnose COVID-19. We propose a deep learning (DL) model to improve the speed of COVID-19 RT-PCR diagnosis. We developed and tested a DL model using the long short-term memory method with a dataset of fluorescence values measured in each cycle of 5810 RT-PCR tests. Among the DL models developed here, the diagnostic performance of the 21st model showed an area under the receiver operating characteristic (AUROC), sensitivity, and specificity of 84.55%, 93.33%, and 75.72%, respectively. The diagnostic performance of the 24th model showed an AUROC, sensitivity, and specificity of 91.27%, 90.00%, and 92.54%, respectively.

Cost-Effective Multiplex Fluorescence Detection System for PCR Chip.

The lack of portability and high cost of multiplex real-time PCR systems limits the device to be used in POC. To overcome this issue, this paper proposes a compact and cost-effective fluorescence detection system that can be integrated to a multiplex real-time PCR equipment. An open platform camera with embedded lens was used instead of photodiodes or an industrial camera. A compact filter wheel using a sliding tape is integrated, and the excitation LEDs are fixed at a 45° angle near the PCR chip, eliminating the need of additional filter wheels. The results show precise positioning of the filter wheel with an error less than 20 µm. Fluorescence detection results using a reference dye and standard DNA amplification showed comparable performance to that of the photodiode system.

Cloud-Based Software Architecture for Fully Automated Point-of-Care Molecular Diagnostic Device.

This paper proposes a cloud-based software architecture for fully automated point-of-care molecular diagnostic devices. The target system operates a cartridge consisting of an extraction body for DNA extraction and a PCR chip for amplification and fluorescence detection. To facilitate control and monitoring via the cloud, a socket server was employed for fundamental molecular diagnostic functions such as DNA extraction, amplification, and fluorescence detection. The user interface for experimental control and monitoring was constructed with the RESTful application programming interface, allowing access from the terminal device, edge, and cloud. Furthermore, it can also be accessed through any web-based user interface on smart computing devices such as smart phones or tablets. An emulator with the proposed software architecture was fabricated to validate successful operation.

Multiple Compact Camera Fluorescence Detector for Real-Time PCR Devices.

The polymerase chain reaction is an important technique in biological research because it tests for diseases with a small amount of DNA. However, this process is time consuming and can lead to sample contamination. Recently, real-time PCR techniques have emerged which make it possible to monitor the amplification process for each cycle in real time. Existing camera-based systems that measure fluorescence after DNA amplification simultaneously process fluorescence excitation and emission for dozens of tubes. Therefore, there is a limit to the size, cost, and assembly of the optical element. In recent years, imaging devices for high-performance, open platforms have benefitted from significant innovations. In this paper, we propose a fluorescence detector for real-time PCR devices using an open platform camera. This system can reduce the cost, and can be miniaturized. To simplify the optical system, four low-cost, compact cameras were used. In addition, the field of view of the entire tube was minimized by dividing it into quadrants. An effective image processing method was used to compensate for the reduction in the signal-to-noise ratio. Using a reference fluorescence material, it was confirmed that the proposed system enables stable fluorescence detection according to the amount of DNA.

Open Platform Cameras Based Bio-Imaging Evaluation System.

With the active development of mobile devices, a variety of ultra-small, high-definition, and open platform-based cameras are being mass-produced. In this paper, we established an emulation system to verify the bio-imaging performance of the bulky and expensive high-performance cameras and various smartphone cameras that have been used in bio-imaging devices. In the proposed system, the linearity of the brightness gradient change of four types of cameras was compared and analyzed. Based on these results, three cameras were selected in order of excellent linearity, and gel image analysis results were compared.

Prediction of Hemorrhagic Transformation after Ischemic Stroke Using Machine Learning.

Hemorrhagic transformation (HT) is one of the leading causes of a poor prognostic marker after acute ischemic stroke (AIS). We compared the performances of the several machine learning (ML) algorithms to predict HT after AIS using only structured data. A total of 2028 patients with AIS, who were admitted within seven days of symptoms onset, were included in this analysis. HT was defined based on the criteria of the European Co-operative Acute Stroke Study-II trial. The whole dataset was randomly divided into a training and a test dataset with a 7:3 ratio. Binary logistic regression, support vector machine, extreme gradient boosting, and artificial neural network (ANN) algorithms were used to assess the performance of predicting the HT occurrence after AIS. Five-fold cross validation and a grid search technique were used to optimize the hyperparameters of each ML model, which had its performance measured by the area under the receiver operating characteristic (AUROC) curve. Among the included AIS patients, the mean age and number of male subjects were 69.6 years and 1183 (58.3%), respectively. HT was observed in 318 subjects (15.7%). There were no significant differences in corresponding variables between the training and test dataset. Among all the ML algorithms, the ANN algorithm showed the best performance in terms of predicting the occurrence of HT in our dataset (0.844). Feature scaling including standardization and normalization, and the resampling strategy showed no additional improvement of the ANN's performance. The ANN-based prediction of HT after AIS showed better performance than the conventional ML algorithms. Deep learning may be used to predict important outcomes for structured data-based prediction.

Word Embedding Reveals Cyfra 21-1 as a Biomarker for Chronic Obstructive Pulmonary Disease.

BACKGROUND: Although patients with chronic obstructive pulmonary disease (COPD) experience high morbidity and mortality worldwide, few biomarkers are available for COPD. Here, we analyzed potential biomarkers for the diagnosis of COPD by using word embedding. METHODS: To determine which biomarkers are likely to be associated with COPD, we selected respiratory disease-related biomarkers. Degrees of similarity between the 26 selected biomarkers and COPD were measured by word embedding. And we infer the similarity with COPD through the word embedding model trained in the large-capacity medical corpus, and search for biomarkers with high similarity among them. We used Word2Vec, Canonical Correlation Analysis, and Global Vector for word embedding. We evaluated the associations of selected biomarkers with COPD parameters in a cohort of patients with COPD. RESULTS: Cytokeratin 19 fragment (Cyfra 21-1) was selected because of its high similarity and its significant correlation with the COPD phenotype. Serum Cyfra 21-1 levels were determined in patients with COPD and controls (4.3 ± 5.9 vs. 3.9 ± 3.6 ng/mL, P = 0.611). The emphysema index was significantly correlated with the serum Cyfra 21-1 level (correlation coefficient = 0.219, P = 0.015). CONCLUSION: Word embedding may be used for the discovery of biomarkers for COPD and Cyfra 21-1 may be used as a biomarker for emphysema. Additional studies are needed to validate Cyfra 21-1 as a biomarker for COPD.

Quantitative Analysis of Fluorescence Detection Using a Smartphone Camera for a PCR Chip.

Most existing commercial real-time polymerase chain reaction (RT-PCR) instruments are bulky because they contain expensive fluorescent detection sensors or complex optical structures. In this paper, we propose an RT-PCR system using a camera module for smartphones that is an ultra small, high-performance and low-cost sensor for fluorescence detection. The proposed system provides stable DNA amplification. A quantitative analysis of fluorescence intensity changes shows the camera's performance compared with that of commercial instruments. Changes in the performance between the experiments and the sets were also observed based on the threshold cycle values in a commercial RT-PCR system. The overall difference in the measured threshold cycles between the commercial system and the proposed camera was only 0.76 cycles, verifying the performance of the proposed system. The set calibration even reduced the difference to 0.41 cycles, which was less than the experimental variation in the commercial system, and there was no difference in performance.

Deep learning based prediction of prognosis in nonmetastatic clear cell renal cell carcinoma.

Survival analyses for malignancies, including renal cell carcinoma (RCC), have primarily been conducted using the Cox proportional hazards (CPH) model. We compared the random survival forest (RSF) and DeepSurv models with the CPH model to predict recurrence-free survival (RFS) and cancer-specific survival (CSS) in non-metastatic clear cell RCC (nm-cRCC) patients. Our cohort included 2139 nm-cRCC patients who underwent curative-intent surgery at six Korean institutions between 2000 and 2014. The data of two largest hospitals' patients were assigned into the training and validation dataset, and the data of the remaining hospitals were assigned into the external validation dataset. The performance of the RSF and DeepSurv models was compared with that of CPH using Harrel's C-index. During the follow-up, recurrence and cancer-specific deaths were recorded in 190 (12.7%) and 108 (7.0%) patients, respectively, in the training-dataset. Harrel's C-indices for RFS in the test-dataset were 0.794, 0.789, and 0.802 for CPH, RSF, and DeepSurv, respectively. Harrel's C-indices for CSS in the test-dataset were 0.831, 0.790, and 0.834 for CPH, RSF, and DeepSurv, respectively. In predicting RFS and CSS in nm-cRCC patients, the performance of DeepSurv was superior to that of CPH and RSF. In no distant time, deep learning-based survival predictions may be useful in RCC patients.

Prediction of Stroke Outcome Using Natural Language Processing-Based Machine Learning of Radiology Report of Brain MRI.

Brain magnetic resonance imaging (MRI) is useful for predicting the outcome of patients with acute ischemic stroke (AIS). Although deep learning (DL) using brain MRI with certain image biomarkers has shown satisfactory results in predicting poor outcomes, no study has assessed the usefulness of natural language processing (NLP)-based machine learning (ML) algorithms using brain MRI free-text reports of AIS patients. Therefore, we aimed to assess whether NLP-based ML algorithms using brain MRI text reports could predict poor outcomes in AIS patients. This study included only English text reports of brain MRIs examined during admission of AIS patients. Poor outcome was defined as a modified Rankin Scale score of 3-6, and the data were captured by trained nurses and physicians. We only included MRI text report of the first MRI scan during the admission. The text dataset was randomly divided into a training and test dataset with a 7:3 ratio. Text was vectorized to word, sentence, and document levels. In the word level approach, which did not consider the sequence of words, and the "bag-of-words" model was used to reflect the number of repetitions of text token. The "sent2vec" method was used in the sensation-level approach considering the sequence of words, and the word embedding was used in the document level approach. In addition to conventional ML algorithms, DL algorithms such as the convolutional neural network (CNN), long short-term memory, and multilayer perceptron were used to predict poor outcomes using 5-fold cross-validation and grid search techniques. The performance of each ML classifier was compared with the area under the receiver operating characteristic (AUROC) curve. Among 1840 subjects with AIS, 645 patients (35.1%) had a poor outcome 3 months after the stroke onset. Random forest was the best classifier (0.782 of AUROC) using a word-level approach. Overall, the document-level approach exhibited better performance than did the word- or sentence-level approaches. Among all the ML classifiers, the multi-CNN algorithm demonstrated the best classification performance (0.805), followed by the CNN (0.799) algorithm. When predicting future clinical outcomes using NLP-based ML of radiology free-text reports of brain MRI, DL algorithms showed superior performance over the other ML algorithms. In particular, the prediction of poor outcomes in document-level NLP DL was improved more by multi-CNN and CNN than by recurrent neural network-based algorithms. NLP-based DL algorithms can be used as an important digital marker for unstructured electronic health record data DL prediction.

Comparison of named entity recognition methodologies in biomedical documents.

BACKGROUND: Biomedical named entity recognition (Bio-NER) is a fundamental task in handling biomedical text terms, such as RNA, protein, cell type, cell line, and DNA. Bio-NER is one of the most elementary and core tasks in biomedical knowledge discovery from texts. The system described here is developed by using the BioNLP/NLPBA 2004 shared task. Experiments are conducted on a training and evaluation set provided by the task organizers. RESULTS: Our results show that, compared with a baseline having a 70.09% F1 score, the RNN Jordan- and Elman-type algorithms have F1 scores of approximately 60.53% and 58.80%, respectively. When we use CRF as a machine learning algorithm, CCA, GloVe, and Word2Vec have F1 scores of 72.73%, 72.74%, and 72.82%, respectively. CONCLUSIONS: By using the word embedding constructed through the unsupervised learning, the time and cost required to construct the learning data can be saved.

Best serum biomarker combination for ovarian cancer classification.

BACKGROUND: Screening test using CA-125 is the most common test for detecting ovarian cancer. However, the level of CA-125 is diverse by variable condition other than ovarian cancer. It has led to misdiagnosis of ovarian cancer. METHODS: In this paper, we explore the 16 serum biomarker for finding alternative biomarker combination to reduce misdiagnosis. For experiment, we use the serum samples that contain 101 cancer and 92 healthy samples. We perform two major tasks: Marker selection and Classification. For optimal marker selection, we use genetic algorithm, random forest, T-test and logistic regression. For classification, we compare linear discriminative analysis, K-nearest neighbor and logistic regression. RESULTS: The final results show that the logistic regression gives high performance for both tasks, and HE4-ELISA, PDGF-AA, Prolactin, TTR is the best biomarker combination for detecting ovarian cancer. CONCLUSIONS: We find the combination which contains TTR and Prolactin gives high performance for cancer detection. Early detection of ovarian cancer can reduce high mortality rates. Finding a combination of multiple biomarkers for diagnostic tests with high sensitivity and specificity is very important.

A method of inferring the relationship between Biomedical entities through correlation analysis on text.

BACKGROUND: One of the most important processes in a machine learning-based natural language processing is to represent words. The one-hot representation that has been commonly used has a large size of vector and assumes that the features that make up the vector are independent of each other. On the other hand, it is known that word embedding has a great effect in estimating the similarity between words because it expresses the meaning of the word well. In this study, we try to clarify the correlation between various terms in the biomedical texts based on the excellent ability of estimating similarity between words shown by word embedding. Therefore, we used word embedding to find new biomarkers and microorganisms related to a specific diseases. METHODS: In this study, we try to analyze the correlation between diseases-markers and diseases-microorganisms. First, we need to construct a corpus that seems to be related to them. To do this, we extract the titles and abstracts from the biomedical texts on the PubMed site. Second, we express diseases, markers, and microorganisms' terms in word embedding using Canonical Correlation Analysis (CCA). CCA is a statistical based methodology that has a very good performance on vector dimension reduction. Finally, we tried to estimate the relationship between diseases-markers pairs and diseases-microorganisms pairs by measuring their similarity. RESULTS: In the experiment, we tried to confirm the correlation derived through word embedding using Google Scholar search results. Of the top 20 highly correlated disease-marker pairs, about 85% of the pairs have actually undergone a lot of research as a result of Google Scholars search. Conversely, for 85% of the 20 pairs with the lowest correlation, we could not actually find any other study to determine the relationship between the disease and the marker. This trend was similar for disease-microbe pairs. CONCLUSIONS: The correlation between diseases and markers and diseases and microorganisms calculated through word embedding reflects actual research trends. If the word-embedding correlation is high, but there are not many published actual studies, additional research can be proposed for the pair.

Permanent magnet actuation for magnetic bead-based DNA extraction.

BACKGROUND: Recently, automatic molecular diagnostic devices to extract DNA have been extensively developed using magnetic beads. While various methods can be applied to the control of the beads, the efficiency of the control when incorporated in automatic devices has not been studied. This paper proposes a compact magnet actuation method for the control of magnetic beads for DNA extraction, and compares the efficiency to the already available magnetic bead-based DNA extraction device. A permanent magnet was preferred for its compactness, while an electro-magnet provides easy operation. After investigating various methods to actuate the magnet with perspective to the size, circuit complexity, and power requirement, we determined the solenoid actuation method to be most efficient. To further reduce the dimension of the overall actuation device, direct actuation of the permanent magnet to control the hold/release of the beads was employed in this paper. The proposed method was compared with the conventional solenoid actuator with a metal plunger. An experimental fluidics device was set up with a fluidic channel and a syringe pump. The bead holding performance against the fluid speed was tested while a fixed amount of beads was loaded into the center of the channel. The group velocity of the beads was analyzed via image processing to determine whether the magnet was sufficient to hold the beads. The required power and space was analyzed and compared qualitatively and quantitatively. RESULT: The proposed direct actuation method was capable of holding the beads at faster fluidic speed than the conventional solenoid actuator. The required power was comparable contemplating the high initial power of the solenoid actuator, and required much smaller space since no plunger was needed. CONCLUSIONS: The direct actuation of the permanent magnet using a solenoid coil showed enhanced performance in holding the beads via permanent magnet, with less complexity of the actuation circuit and space. The proposed method therefore can efficiently improve the overall performance of the bead-based DNA extraction.

Performance evaluation of optimal real-time polymerase chain reaction achieved with reduced voltage.

BACKGROUND: Polymerase chain reaction (PCR) is used in nucleic acid tests of infectious diseases in point-of-care testing. Previous studies have demonstrated real-time PCR that uses a micro-PCR chip made of packing tape, double-sided tape, and a plastic cover with polycarbonate or polypropylene on a black matte printed circuit board substrate. Despite the success of DNA amplification and fluorescence detection using an early version of the micro-PCR chip, reaching the target temperature was fairly slow and, as a result, the total running time was getting longer. To reduce this runtime, the micro-PCR chip was modified by reducing the heater pattern size of the PCB substrate to one-quarter of the original size or less, while maintaining the ability of the heating pattern to cover the reservoir area of the microfluidic channel. In subsequent experiments, DNA amplification failed several times. During the analysis of the cause of this failure, it was found that the reagent was boiling with the heating range from 25 to 95 °C. METHODS: As a method of DNA amplification verification, images were captured by digital single-lens reflex camera to detect FAM fluorescence using diagonal illumination from a blue LED light source. The images were automatically captured at 72 °C (the extension step in nucleic acid amplification) and the brightness of the captured images was analyzed to con-firm the success of DNA amplification. RESULTS: Compared to the previous chip with a larger heating pattern size, the current chip appears to generate excess energy as the size of the heating pattern was reduced. To reduce this excess energy, the initial voltage was lowered to 2 V and 2.5 V, which is equivalent to a one-fifth and one-quarter voltage-power reduction in pulse width modulation control, respectively. In both voltage reduction cases, the DNA amplification was successful. CONCLUSIONS: DNA amplification tests may fail due to the excess energy generated by reducing the heater pattern size of the PCB substrate. However, the tests succeeded when the voltage was reduced to 2 V or 2.5 V. The 2.5 V power test was more efficient for reducing the overall running time.

Evaluation-independent system for DNA section amplification.

BACKGROUND: In general, the image analysis of nucleic acid for detecting DNA is dependent on the gel documentation system. These experiments may deal with harmful staining agents and are time consuming. To address these issues, real-time polymerase chain reaction (PCR) devices have been developed. The advantages of real-time PCR are its capabilities for real-time diagnosis, improved sensitivity, and digitization of measurement results. However, real-time PCR equipment is still too bulky and expensive for use in small hospitals and laboratories. METHODS: This paper describes an evaluation-independent real-time PCR system that differs from conventional systems in that it uses a side-illumination optical detection system and a temperature adjustment coefficient for DNA detection. The overall configuration of the evaluation-independent system includes the PCR chip and system hardware and software. The use of the side-illumination method for detection enables the system size to be reduced compared to systems using a typical illumination method. Furthermore, the results of a PCR test are strongly affected by the reaction temperature. Thus, extremely precise control of the temperature of the reaction is needed to obtain accurate results and good reliability. We derived a temperature compensation coefficient that allows us to compensate for the differences between the measured temperature of the negative temperature coefficient (NTC) thermistor sensor and the real temperature of the thermocouple. RESULTS: Applying the temperature compensation coefficient parameter using the NTC thermistor and using the side-illumination method resulted in an increase in the initial sensor value. The occurrence of the DNA section amplification decreased to 22 cycles from 24 cycles. CONCLUSIONS: The proposed system showed comparable performance to that of an existing real-time PCR, even with the use of simpler and smaller optical devices.

A study of polymerase chain reaction device control via cloud using Firebase Cloud Messaging protocol.

BACKGROUND: In this paper, we propose a system for data monitoring and control of polymerase chain reaction (PCR) externally. PCR is a technique for amplifying a desired DNA molecule by repeatedly synthesizing a specific part of DNA sequence. Currently, commercially available systems are standalone systems or operate PCR devices through a computer in the vicinity of devices for control purposes. These systems are limited in the number of devices that the host system can monitor at the same time, and there are limitations in controlling devices or accessing experimental data externally. Therefore, we propose a system to control the PCR device via the cloud for the convenience of the user and to overcome the limitation of the place. METHODS: The cloud system used in this study is Google's Firebase. At this time, we use Firebase Cloud Messaging (FCM) protocol to send and receive data. In this paper, we have experimented on the possibility of data transmission and reception using FCM between device, cloud and user. Since the PCR chips used in the research are generally operated at about 10°/s, and the temperature can be controlled within 0.5°, the processing period of the control process should be made much smaller than 1/20 s (50 ms). RESULTS: As a result of experiments, the time of the data round-trip using FCM was measured at 150 ms on the average. Therefore, the data exchange time using FCM is three times slower than the reference time of 50 ms. CONCLUSIONS: Since the data round-trip time using FCM is measured to be three times slower than the reference time of 50 ms, it is impossible for the user to control the device such as the PCR device used in this study through the cloud. However, it is possible for the user to monitor the status of the PCR device from the outside in real time.