Independent Test Dataset Research Articles

Rhythm classifier: artificial intelligence for rhythm classification in insertable cardiac monitor

Abstract Background Though detections improved significantly, separating atrial tachycardia/atrial flutter (AT/AFL) from atrial fibrillations (AF) in insertable cardiac monitors (ICM) remains a challenge. Objective Using artificial intelligence (AI) algorithms to classify AF and AT/AFL in ICM detected episodes, which may reduce episode review burden. Methods Both raw EGM signal and the QRS diminished version of the raw signal were used to extract 14 AF and AT/AFL features from each episode. Extracted features were then represented as an ensemble of features and used to train and test the AI model. Modified ResNet18 network was used as the AI model. All episodes from a random selection of 80% of patients were used as training dataset, another 10% were used as validation, and 10% were used as test dataset, to evaluate model performance. AI model probability threshold was chosen based on validation set results. Episode detection sensitivity, specificity, and area under the curve (AUC) of receiver operating characteristics (ROC) curve in the independent test dataset are reported. Results AF and AT/AFL episodes detected in 918 LINQ ICM patients (5,253 AF and 4,705 AT/AFL – 745 pts and 8,950 false detection episodes) were included. Training dataset consisted of 15,183 episodes (735 pts), validation set consisted of 1,901 episodes (91 pts), and test set consisted of 1,824 episodes (92 pts). At chosen threshold in validation set, AF and AT/AFL detection had sensitivity of 90% and 73.3%, respectively, and specificity of 93.2% and 95%, respectively, with AUC of 0.966. In test data, AF and AT/AFL obtained relative sensitivity of 84.7% and 85.7%, respectively, with specificity of 88% and 86.5%, respectively (Fig. A). With ROC-AUC of 0.956 (Fig. B), the relative diagnostic yield of AF and AT/AFL on test data was 95.6%. Conclusion The trained AI model provides a relative sensitivity of about 85% for AF vs. AT/AFL classification and reduces inappropriate ICM detections by 79%. Adding more episodes may improve the performance.

Prediction of human O-linked glycosylation sites using stacked generalization and embeddings from pre-trained protein language model.

O-linked glycosylation, an essential post-translational modification process in Homo sapiens, involves attaching sugar moieties to the oxygen atoms of serine and/or threonine residues. It influences various biological and cellular functions. While threonine or serine residues within protein sequences are potential sites for O-linked glycosylation, not all serine and/or threonine residues undergo this modification, underscoring the importance of characterizing its occurrence. This study presents a novel approach for predicting intracellular and extracellular O-linked glycosylation events on proteins, which are crucial for comprehending cellular processes. Two base multi-layer perceptron models were trained by leveraging a stacked generalization framework. These base models respectively employ ProtT5 and Ankh O-linked glycosylation site-specific embeddings whose combined predictions are used to train the meta-multi-layer perceptron model. Trained on extensive O-linked glycosylation datasets, the stacked-generalization model demonstrated high predictive performance on independent test datasets. Furthermore, the study emphasizes the distinction between nucleocytoplasmic and extracellular O-linked glycosylation, offering insights into their functional implications that were overlooked in previous studies. By integrating the protein language model's embedding with stacked generalization techniques, this approach enhances predictive accuracy of O-linked glycosylation events and illuminates the intricate roles of O-linked glycosylation in proteomics, potentially accelerating the discovery of novel glycosylation sites. Stack-OglyPred-PLM produces Sensitivity, Specificity, Matthews Correlation Coefficient, and Accuracy of 90.50%, 89.60%, 0.464, and 89.70%, respectively on a benchmark NetOGlyc-4.0 independent test dataset. These results demonstrate that Stack-OglyPred-PLM is a robust computational tool to predict O-linked glycosylation sites in proteins. The developed tool, programs, training, and test dataset are available at https://github.com/PakhrinLab/Stack-OglyPred-PLM. Supplementary data are available at Bioinformatics online.

Precipitation Retrieval from FY-3G/MWRI-RM Based on SMOTE-LGBM

Using the FY-3G/MWRI-RM observations, this paper proposes a precipitation retrieval method that combines the Synthetic Minority Over-sampling Technique with Light Gradient Boosting Machine (SMOTE-LGBM) and analyzes the impact of MWRI-RM channel settings on precipitation retrieval. The SMOTE-LGBM-based model consists of two LGBM models for precipitation identification and estimation, respectively. The SMOTE method is used to address the imbalance between precipitation and non-precipitation samples. Using the Integrated Multi-Satellite Retrievals for the Global Precipitation Measurement (IMERG) product as a reference, we validate the retrieved precipitation by the SMOTE-LGBM-based model with an independent testing dataset. The critical success indexes are 0.483 and 0.526, and the Pearson correlation coefficients are 0.611 and 0.645 for the ocean and land regions, respectively. The spatial distributions of the retrieved and IMERG accumulated precipitation in the testing dataset are similar. In addition, we visualize and analyze the cases of Meiyu and two typhoons. The results indicate that the SMOTE-LGBM-based model effectively represents the spatial distribution characteristics of precipitation and achieves high agreement with IMERG precipitation products. Overall, the SMOTE-LGBM-based model successfully retrieves precipitation from MWRI-RM and provides accurate precipitation products for FY-3G/MWRI-RM for the first time.

MetaCGRP is a high-precision meta-model for large-scale identification of CGRP inhibitors using multi-view information

Migraine is considered one of the debilitating primary headache conditions with an estimated worldwide occurrence of approximately 14–15%, contributing highly to factors responsible for global disability. Calcitonin gene-related peptide (CGRP) is a neuropeptide that plays a crucial role in the pathophysiology of migraines and thus, its inhibition can help relieve migraine symptoms. However, conventional process of CGRP drug development has been laborious and time-consuming with incurred costs exceeding one billion dollars. On the other hand, machine learning (ML)-based approaches that are capable of accurately identifying CGRP inhibitors could greatly facilitate in expediting the discovery of novel CGRP drugs. Therefore, this study proposes a novel and high-accuracy meta-model, namely MetaCGRP, that can precisely identify CGRP inhibitors. To the best of our knowledge, MetaCGRP is the first SMILES-based approach that has been developed to identify CGRP inhibitors without the use of 3D structural information. In brief, we initially employed different molecular representation methods coupled with popular ML algorithms to construct a pool of baseline models. Then, all baseline models were optimized and used to generate multi-view features. Finally, we employed the feature selection method to optimize the multi-view features and determine the best feature subset to enable the construction of the meta-model. Both cross-validation and independent tests indicated that MetaCGRP clearly outperforms several conventional ML classifiers, with accuracies of 0.898 and 0.799 on the training and independent test datasets, respectively. In addition, MetaCGRP in conjunction with molecular docking was utilized to identify five potential natural product candidates from Thai herbal pharmacopoeia and analyze their binding affinity and interactions to CGRP. To facilitate community-wide efforts in expediting the discovery of novel CGRP inhibitors, a user-friendly web server for MetaCGRP is freely available at https://pmlabqsar.pythonanywhere.com/MetaCGRP.

MMG4: Recognition of G4-Forming Sequences Based on Markov Model.

G-quadruplexes (G4s) are special nucleic acid structures with various important biological functions. Existing tools and technologies for G4-forming sequences recognition are limited to time-consuming and costly methods such as circular dichroism and nuclear magnetic resonance. Developing a fast and accurate model for G4-forming sequences recognition has far-reaching significance. In this study, MMG4, a novel model to recognize G4-forming sequences based on Markov model (MM), was developed and the phenomenon of high recognition accuracy in the central region of the sequence and low accuracy in the two end regions was discovered. It was further found that the differences in base transfer probabilities, ratio distribution, and G4-motif structural content in different regions may be the causes of this phenomenon. The study also explored the impact of sequence length on recognition accuracy and found the optimal recognition interval to be [910-1049], with the highest recognition accuracy reaching 85.95%. By extracting sequence features, the study constructed three types of machine learning models: random forest (RF), support vector machine, and back-propagation neural network. It was found that recognition performance of MM was significantly better than that of the other three machine learning models, proving that the recognition method based on MM can effectively capture the correlation information between adjacent nucleotides of G4. By combining MM with the three machine learning models, the predictive performance of MMG4 improved. Among them, the RF model combined with MM has the best performance, achieving an area under the receiver operating characteristic curve value of 0.93 and an area under the precision-recall curve value of 0.9. Finally, the study validated the model robustness and generalization ability through independent testing dataset.

Tree-based classification model for Long-COVID infection prediction with age stratification using data from the national COVID cohort collaborative (N3C)

Abstract Objectives We propose and validate a domain knowledge-driven classification model for diagnosing Post-acute sequelae of SARS-CoV-2 infection (PASC), also known as Long COVID, using Electronic Health Records (EHR) data. Materials and Methods We developed a robust model that incorporates features strongly indicative of PASC or associated with the severity of COVID-19 symptoms as identified in our literature review. The XGBoost tree-based architecture was chosen for its ability to handle class-imbalanced data and its potential for high interpretability. Using the training data provided by the Long COVID Computation Challenge (L3C), which was a sample of the National COVID Collaborative (N3C), our models were fine-tuned and calibrated to optimize Area Under the Receiver Operating Characteristic curve (AUROC) and the F1 score, following best practices for the class-imbalanced N3C data. Results Our age-stratified classification model demonstrated strong performance with an average 5-fold cross-validated AUROC of 0.844 and F1 score of 0.539 across the young adult, mid-aged, and older-aged populations in the training data. In an independent testing dataset, which was made available after the challenge was over, we achieved an overall AUROC score of 0.814 and F1 score of 0.545. Discussion The results demonstrated the utility of knowledge-driven feature engineering in a sparse EHR data and demographic stratification in model development to diagnose a complex and heterogeneously presenting condition like PASC. The model's architecture, mirroring natural clinician decision-making processes, contributed to its robustness and interpretability, which are crucial for clinical translatability. Further, the model’s generalizability was evaluated over a new cross-sectional data as provided in the later stages of the L3C challenge. Conclusion The study proposed and validated the effectiveness of age-stratified, tree-based classification models to diagnose PASC. Our approach highlights the potential of machine learning in addressing the diagnostic challenges posed by the heterogeneity of Long COVID symptoms.

CoNglyPred: Accurate Prediction of N-Linked Glycosylation Sites Using ESM-2 and Structural Features With Graph Network and Co-Attention.

N-Linked glycosylation is crucial for various biological processes such as protein folding, immune response, and cellular transport. Traditional experimental methods for determining N-linked glycosylation sites entail substantial time and labor investment, which has led to the development of computational approaches as a more efficient alternative. However, due to the limited availability of 3D structural data, existing prediction methods often struggle to fully utilize structural information and fall short in integrating sequence and structural information effectively. Motivated by the progress of protein pretrained language models (pLMs) and the breakthrough in protein structure prediction, we introduced a high-accuracy model called CoNglyPred. Having compared various pLMs, we opt for the large-scale pLM ESM-2 to extract sequence embeddings, thus mitigating certain limitations associated with manual feature extraction. Meanwhile, our approach employs a graph transformer network to process the 3D protein structures predicted by AlphaFold2. The final graph output and ESM-2 embedding are intricately integrated through a co-attention mechanism. Among a series of comprehensive experiments on the independent test dataset, CoNglyPred outperforms state-of-the-art models and demonstrates exceptional performance in case study. In addition, we are the first to report the uncertainty of N-linked glycosylation predictors using expected calibration error and expected uncertainty calibration error.

IACP-DFSRA: Identification of Anticancer Peptides Based on a Dual-channel Fusion Strategy of ResCNN and Attention

Anticancer peptides (ACPs) have been widely applied in the treatment of cancer owing to good safety, rational side effects, and high selectivity. However, the number of ACPs that have been experimentally validated is limited as identification of ACPs is extremely expensive. Hence, accurate and cost-effective identification methods for ACPs are urgently needed. In this work, we proposed a deep learning-based model, named iACP-DFSRA, for ACPs identification. Specifically, we adopted two kinds of sequence embedding technologies, ProtBert_BFD pre-training language model and handcrafted features to encode protein sequences. Then, the LightGBM was used for feature selection, and the selected features were input into ResCNN and Attention mechanism, respectively, to extract local and global features. Finally, the concatenate features were deeply fused by using the Attention mechanism to allow key features to be paid more attention to by the model and make predictions by fully connected layer. The results of 10-fold cross-validation demonstrated that the iACP-DFSRA model delivered improved results in most metrics with Sp of 94.15%, Sn of 95.32%, Acc of 94.74% and MCC of 89.48% compared to the latest AACFlow model. Indeed, the iACP-DFSRA model is the only model with Acc > 90% and MCC > 80% on this independent test dataset. Furthermore, we have further demonstrated the superiority of our model on additional datasets. In addition, t-SNE and SHAP interpretation analysis demonstrated that it is crucial to use two channels for feature extraction and use the Attention mechanism for deep fusion, which helps the iACP-DFSRA to predict ACPs more effectively.

Muscle Fat and Volume Differences in People With Hip-Related Pain Compared With Controls: A Machine Learning Approach.

Hip-related pain (HRP) affects young to middle-aged active adults and impacts physical activity, finances and quality of life. HRP includes conditions like femoroacetabular impingement syndrome and labral tears. Lateral hip muscle dysfunction and atrophy in HRP are more pronounced in advanced hip pathology, with limited evidence in younger populations. While MRI use for assessing hip muscle morphology is increasing, with automated deep-learning techniques showing promise, studies assessing their accuracy are limited. Therefore, we aimed to compare hip intramuscular fat infiltrate (MFI) and muscle volume, in individuals with and without HRP as well as assess the reliability and accuracy of automated machine-learning segmentations compared with human-generated segmentation. This cross-sectional study included sub-elite/amateur football players (Australian football and soccer) with a greater than 6-month history of HRP [n = 180, average age 28.32, (standard deviation 5.88) years, 19% female] and a control group of sub-elite/amateur football players without pain [n = 48, 28.89 (6.22) years, 29% female]. Muscle volume and MFI of gluteus maximus, medius, minimis and tensor fascia latae were assessed using MRI. Associations between muscle volume and group were explored using linear regression models, controlling for body mass index, age, sport and sex. A convolutional neural network (CNN) machine-learning approach was compared with human-performed muscle segmentations in a subset of participants (n = 52) using intraclass correlation coefficients and Sorensen-Dice index. When considering adjusted estimates of muscle volume, there were significant differences observed between groups for gluteus medius (adjusted mean difference 23 858 mm3 [95% confidence interval 7563, 40 137]; p = 0.004) and tensor fascia latae (6660 mm3 [2440, 13 075]; p = 0.042). No differences were observed between groups for gluteus maximus (18 265 mm3 [-21 209, 50 782]; p = 0.419) or minimus (3893 mm3 [-2209, 9996]; p = 0.21). The CNN was trained for 30 000 iterations and assessed its accuracy and reliability on an independent testing dataset, achieving high segmentation accuracy (mean Sorenson-Dice index >0.900) and excellent muscle volume and MFI reliability (ICC2,1 > 0.900). The CNN outperformed manual raters, who had slightly lower interrater accuracy (Sorensen-Dice index >0.800) and reliability (ICC2,1 > 0.800). The increased muscle volumes in the symptomatic group compared with controls could be associated with increased myofibrillar size, sarcoplasmic hypertrophy or both. These changes may facilitate greater muscular efficiency for a given load, enabling the athlete to maintain their normal level of function. In addition, the CNNs for muscle segmentation was more efficient and demonstrated excellent reliability in comparison to manual segmentations.

Using AI explainable models and handwriting/drawing tasks for psychological well-being

This study addresses the increasing threat to Psychological Well-Being (PWB) posed by Depression, Anxiety, and Stress conditions. Machine learning methods have shown promising results for several psychological conditions. However, the lack of transparency in existing models impedes practical application. The study aims to develop explainable machine learning models for depression, anxiety and stress prediction, focusing on features extracted from tasks involving handwriting and drawing.Two hundred patients completed the Depression, Anxiety, and Stress Scale (DASS-21) and performed seven tasks related to handwriting and drawing. Extracted features, encompassing pressure, stroke pattern, time, space, and pen inclination, were used to train the explainable-by-design Entropy-based Logic Explained Network (e-LEN) model, employing first-order logic rules for explanation. Performance comparison was performed with XGBoost, enhanced by the SHAP explanation method.The trained models achieved notable accuracy in predicting depression (0.749 ±0.089), anxiety (0.721 ±0.088), and stress (0.761 ±0.086) through 10-fold cross-validation (repeated 20 times). The e-LEN model’s logic rules facilitated clinical validation, uncovering correlations with existing clinical literature. While performance remained consistent for depression and anxiety on an independent test dataset, a slight degradation was observed for stress prediction in the test task.

Enhanced prognostic signature for lung adenocarcinoma through integration of adjacent normal and tumor gene expressions

BackgroundCancer prognosis-related signatures have traditionally been constructed based on gene expression profiles derived from tumor or normal tissues. However, the potential benefits of incorporating gene expression profiles from both tumor and normal tissues to improve signature performance have not been explored. MethodsIn this study, we developed three prognostic models for lung adenocarcinoma (LUAD) using gene expression profiles from tumor tissues, normal tissues, and a combination (COM) of both, sourced from The Cancer Genome Atlas (TCGA). To ensure comparability, the same workflow was followed for all three models. ResultsWhen applied to the TCGA LUAD dataset, the tumor-derived model exhibited the best overall performance, except in calibration analysis, where the normal-derived model performed better. The COM-derived model demonstrated intermediate performance. Validation on three independent test datasets revealed that the COM-derived model showed the best performance, while the normal-derived model showed the worst. In overall survival (OS) analysis, the low-risk group defined by the COM-derived model consistently exhibited longer mean survival times. The tumor-derived model did not consistently show this trend, and the normal-derived model produced opposite results. In discrimination analysis, no significant differences were observed. The COM-derived model demonstrated good discrimination ability for short periods, while the tumor-derived model performed better for longer periods. In calibration analysis, both the COM and tumor-derived models had similar absolute prediction errors, which were better than those of the normal-derived model. However, the tumor-derived model tended to underestimate survival rates. The clinical feature analysis and validation in GSE229705 indicate that the risk score (RS) from the COM model is the most clinically significant. These results demonstrate that the COM model's RS aligns more closely with clinical data, maintaining stable performance and the strongest generalizability. ConclusionsOverall, the COM-derived model demonstrated the best generalization ability. The superior performance of the tumor-derived model in the TCGA LUAD dataset might be due to overfitting. Our results suggest that appropriate combinations of gene expression data from tumor and normal tissues can enhance the predictive power of prognostic signatures.

Comprehensive Assessment of BERT-Based Methods for Predicting Antimicrobial Peptides.

In recent years, the prediction of antimicrobial peptides (AMPs) has gained prominence due to their high antibacterial activity and reduced susceptibility to drug resistance, making them potential antibiotic substitutes. To advance the field of AMP recognition, an increasing number of natural language processing methods are being applied. These methods exhibit diversity in terms of pretraining models, pretraining data sets, word vector embeddings, feature encoding methods, and downstream classification models. Here, we provide a comprehensive survey of current BERT-based methods for AMP prediction. An independent benchmark test data set is constructed to evaluate the predictive capabilities of the surveyed tools. Furthermore, we compared the predictive performance of these computational methods based on six different AMP public databases. LM_pred (BFD) outperformed all other surveyed tools due to abundant pretraining data set and the unique vector embedding approach. To avoid the impact of varying training data sets used by different methods on prediction performance, we performed the 5-fold cross-validation experiments using the same data set, involving retraining. Additionally, to explore the applicability and generalization ability of the models, we constructed a short peptide data set and an external data set to test the retrained models. Although these prediction methods based on BERT can achieve good prediction performance, there is still room for improvement in recognition accuracy. With the continuous enhancement of protein language model, we proposed an AMP prediction method based on the ESM-2 pretrained model called iAMP-bert. Experimental results demonstrate that iAMP-bert outperforms other approaches. iAMP-bert is freely accessible to the public at http://iamp.aielab.cc/.

MGCN-PolyA: An Integrated Computational Framework for Predicting Poly(A) Signals with Multiscale-gated Convolutional Networks

Background: The accurate recognition of the polyadenylation signal (PAS) from DNA sequences is essential for understanding gene transcriptional regulation. A variety of machine learning-based computational methods have been developed to predict PAS in recent years; however, their performance and their generalization ability are unsatisfactory. It is highly desirable to design more preferable computational approaches for PAS prediction. Methods: In this work, we developed an integrated framework MGCN-PolyA for PAS prediction across four species, including Homo sapiens, Bos taurus, Mus musculus, and Drosophila melanogaster. MGCN-Poly(A) benefits from the diversity of feature engineering and the effectiveness of the model architecture. We combined features from different perspectives, such as word embedding, One-hot encoding, K-mer frequency, and Enhanced Nucleic Acid Composition (ENAC), which complement each other and provide rich and comprehensive information for model learning. In model architecture, MGCN-Poly(A) leverages a two-channel multi-scale gated convolutional network to effectively learn high-level feature representations at different scales, and then combines the statistical features to predict PAS using random forest algorithm. These designs not only speed up network training, but also improves the generalization ability Results: The benchmarking experiments on the independent test datasets demonstrate that MGCNPolyA outperforms other state-of-the-art algorithms in identifying PAS. MGCN-PolyA has the highest accuracy on all test datasets, and its excellent performance on cross-species validation also demonstrates the robustness of our model. Conclusion: Extracting features from different perspectives is important for PAS recognition, and the integration of DNNs and shallow machine learning algorithms can improve the model performance.

Automatic motion artifact detection in electrodermal activity signals using 1D U-net architecture

We developed a method for automated detection of motion and noise artifacts (MNA) in electrodermal activity (EDA) signals, based on a one-dimensional U-Net architecture. EDA has been widely employed in diverse applications to assess sympathetic functions. However, EDA signals can be easily corrupted by MNA, which frequently occur in wearable systems, particularly those used for ambulatory recording. MNA can lead to false decisions, resulting in inaccurate assessment and diagnosis. Several approaches have been proposed for MNA detection; however, questions remain regarding the generalizability and the feasibility of implementation of the algorithms in real-time especially those involving deep learning approaches. In this work, we propose a deep learning approach based on a one-dimensional U-Net architecture using spectrograms of EDA for MNA detection. We developed our method using four distinct datasets, including two independent testing datasets, with a total of 9602 128-s EDA segments from 104 subjects. Our proposed scheme, including data augmentation, spectrogram computation, and 1D U-Net, yielded balanced accuracies of 80.0 ± 13.7 % and 75.0 ± 14.0 % for the two independent test datasets; these results are better than or comparable to those of other five state-of-the-art methods. Additionally, the computation time of our feature computation and machine learning classification was significantly lower than that of other methods (p < .001). The model requires only 0.28 MB of memory, which is far smaller than the two deep learning approaches (4.93 and 54.59 MB) which were used as comparisons to our study. Our model can be implemented in real-time in embedded systems, even with limited memory and an inefficient microprocessor, without compromising the accuracy of MNA detection.

AScirRNA: A novel computational approach to discover abiotic stress-responsive circular RNAs in plant genome

In the realm of plant biology, understanding the intricate regulatory mechanisms governing stress responses stands as a pivotal pursuit. Circular RNAs (circRNAs), emerging as critical players in gene regulation, have garnered attention in recent days for their potential roles in abiotic stress adaptation. A comprehensive grasp of circRNAs' functions in stress response offers avenues for breeders to manipulating plants to develop abiotic stress resistant crop cultivars to thrive in challenging climates. This study pioneers a machine learning-based model for predicting abiotic stress-responsive circRNAs. The K-tuple nucleotide composition (KNC) and Pseudo KNC (PKNC) features were utilized to numerically represent circRNAs. Three different feature selection strategies were employed to select relevant and non-redundant features. Eight shallow and four deep learning algorithms were evaluated to build the final predictive model. Following five-fold cross-validation process, XGBoost learning algorithm demonstrated superior performance with LightGBM-chosen 260 KNC features (Accuracy: 74.55 %, auROC: 81.23 %, auPRC: 76.52 %) and 160 PKNC features (Accuracy: 74.32 %, auROC: 81.04 %, auPRC: 76.43 %), over other combinations of learning algorithms and feature selection techniques. Further, the robustness of the developed models were evaluated using an independent test dataset, where the overall accuracy, auROC and auPRC were found to be 73.13 %, 72.34 % and 72.68 % for KNC feature set and 73.52 %, 79.53 % and 73.09 % for PKNC feature set, respectively. This computational approach was also integrated into an online prediction tool, AScirRNA (https://iasri-sg.icar.gov.in/ascirna/) for easy prediction by the users. Both the proposed model and the developed tool are poised to augment ongoing efforts in identifying stress-responsive circRNAs in plants.

Automated plan generation for prostate radiotherapy patients using deep learning and scripted optimization

Background and PurposeTreatment planning is a time-intensive task that could be automated. We aimed to develop a “single-click” workflow, fully deployed within a commercial treatment planning system (TPS), for autoplanning prostate radiotherapy treatment plans using predictions from a deep learning model (DLM). Materials and MethodsAutomatically generated treatment plans were created with a single script, executed from within a commercial TPS scripting environment, that performed two stages sequentially. Initially, a 3D dose distribution was predicted with a ResUNet DLM. The DLM was trained and validated using previously treated datasets (n = 120) which used 3D contours as inputs. Following this, dose predictions were converted into treatment plans by extracting dose-volume metrics from the predictions to use as objectives for the inverse optimizer within the TPS. An independent test dataset (n = 20) was used to evaluate the similarity between automated and clinical plans. ResultsFor planning target volumes, the median percentage difference and interquartile range between the automatically generated plans and clinical plans were 0.4% [0.2-1.1%] for the V100%, −0.5% [(−1.0)-(−0.2)%] for D99% and −0.5% [(−1.0)-(−0.2)%] for D95%. Bladder and rectum volume-at-dose objectives agreed within −6.1% [(−12.5)-0.9%]. The conversion of the DLM prediction into a treatment plan took 15 min [13-16 min]. ConclusionsAn automatic plan generation workflow that uses a DL model with scripted optimization was fully deployed in a commercial TPS. Autoplans were compared to previously treated clinical plans and were found to be non-inferior.

7T-high resolution MRI-derived radiomic analysis for the identification of symptomatic intracranial atherosclerotic plaques.

High-resolution magnetic resonance imaging (HR-MRI) allows for detailed visualization of intracranial atherosclerotic plaques. Radiomics can be used as a tool for objective quantification of the plaque's characteristics. We analyzed the radiomics features (RFs) obtained from 7 T HR-MRI of patients with intracranial atherosclerotic disease (ICAD) to determine distinct characteristics of culprit and non-culprit plaques. Patients with stroke due to ICAD underwent HR-MRI. Culprit plaques in the vascular territory of the stroke were identified. Degree of stenosis, area degree of stenosis and plaque burden were calculated. A three-dimensional segmentation of the plaque was performed, and RFs were obtained. A machine learning model for prediction and identification of culprit plaques using significantly different RFs was evaluated. The study included 33 patients with ICAD as stroke etiology. Univariate analysis revealed 24 RFs in pre-contrast MRI, 21 in post-contrast MRI, 13 RFs that were different between pre and post contrast MRIs. Additionally, six shape-based RFs significantly differed from culprit and non-culprit plaques. The random forest model achieved an accuracy rate of 81% (88% sensitivity and 75% specificity) in identifying culprit plaques in the independent testing dataset. This model successfully identified the culprit plaques in all patients during the testing phase. Symptomatic plaques had a distinct signature RFs compared to other plaques within the same subject. A machine learning model built with RFs successfully identified the symptomatic atherosclerotic plaques in most cases. Radiomics is a promising tool for stratification of plaques in patients with ICAD.

Applicability of calibrated diffuse reflectance spectroscopy models across spatial and temporal boundaries

Diffuse reflectance spectroscopy (DRS) is an emerging soil testing approach. Although several studies have validated the DRS approach, limited efforts are made to assess the applicability of calibrated DRS models on new samples collected at different locations and/or time. To test such spatio-temporal applicability of calibrated DRS models, we collected surface soil samples from 1,112 smallholder farms during 2018 (T2018) and 607 farms during 2021 (T2021) covering seven districts of the Bundelkhand region of central India. The T2018 samples covered 7 development blocks; the T2021 samples were also collected from these blocks but from different sampling locations. Additionally, a new sampling site (Jhansi-Bamour block) was added during 2021 to create an independent test dataset. Collected samples were analysed for 17 soil parameters (basic soil properties, macronutrients, and micronutrients) and spectral reflectance over the visible to near-infrared region. Corresponding soil test crop response (STCR) ratings were also estimated. The Cubist model was calibrated in the T2018 dataset and tested against the T2021 dataset using the coefficient of determination (R2), root-mean-squared error (RMSE), and percentage relative error deviation (PRED) at 30% error threshold as performance statistics. Model applicability was assessed at each block level (site-specific), by dividing the study site into their two geology-specific regions, and by treating the entire dataset as a regional-scale spectral library. Results showed that DRS models calibrated on a finer scale (site-specific) are less efficient in estimating soil parameters in broader scale (geology-specific and regional-scale) test T2021 samples although their STCR ratings may safely be estimated at local scales. When site-specific data were aggregated to broader scales and T2018 dataset was spiked with 20% samples from the T2021 dataset, model performance improved for critical soil parameters such as soil organic carbon (SOC) contents and several plant nutrients and their ratings; application of such large-scale models also improved the estimation accuracy when applied to site-specific datasets. Exchangeable Ca and Mg, clay and SOC contents were frequently well-estimated with R2 values ranging from 0.54 to 0.93. Fine sand was the next best estimated soil property with R2 values in the range of 0.40–0.75. The STCR ratings estimated in the DRS approach matched the wet chemistry-based STCR ratings to the tune of 43 to 100%. Overall, as many as 60% of all new samples could be estimated with more than 70% accuracy for 8 out of 17 parameters. With the DRS approach tested on both spatially- and temporally-independent test datasets and, specifically, with high estimation accuracy of STCR ratings, our results suggest that the DRS approach may safely be used as a viable alternative to conventional soil testing in smallholder farms.

Renal tumor segmentation, visualization, and segmentation confidence using ensembles of neural networks in patients undergoing surgical resection.

To develop an automatic segmentation model for solid renal tumors on contrast-enhanced CTs and to visualize segmentation with associated confidence to promote clinical applicability. The training dataset included solid renal tumor patients from two tertiary centers undergoing surgical resection and receiving CT in the corticomedullary or nephrogenic contrast media (CM) phase. Manual tumor segmentation was performed on all axial CT slices serving as reference standard for automatic segmentations. Independent testing was performed on the publicly available KiTS 2019 dataset. Ensembles of neural networks (ENN, DeepLabV3) were used for automatic renal tumor segmentation, and their performance was quantified with DICE score. ENN average foreground entropy measured segmentation confidence (binary: successful segmentation with DICE score > 0.8 versus inadequate segmentation ≤ 0.8). N = 639/n = 210 patients were included in the training and independent test dataset. Datasets were comparable regarding age and sex (p > 0.05), while renal tumors in the training dataset were larger and more frequently benign (p < 0.01). In the internal test dataset, the ENN model yielded a median DICE score = 0.84 (IQR: 0.62-0.97, corticomedullary) and 0.86 (IQR: 0.77-0.96, nephrogenic CM phase), and the segmentation confidence an AUC = 0.89 (sensitivity = 0.86; specificity = 0.77). In the independent test dataset, the ENN model achieved a median DICE score = 0.84 (IQR: 0.71-0.97, corticomedullary CM phase); and segmentation confidence an accuracy = 0.84 (sensitivity = 0.86 and specificity = 0.81). ENN segmentations were visualized with color-coded voxelwise tumor probabilities and thresholds superimposed on clinical CT images. ENN-based renal tumor segmentation robustly performs in external test data and might aid in renal tumor classification and treatment planning. Ensembles of neural networks (ENN) models could automatically segment renal tumors on routine CTs, enabling and standardizing downstream image analyses and treatment planning. Providing confidence measures and segmentation overlays on images can lower the threshold for clinical ENN implementation. Ensembles of neural networks (ENN) segmentation is visualized by color-coded voxelwise tumor probabilities and thresholds. ENN provided a high segmentation accuracy in internal testing and in an independent external test dataset. ENN models provide measures of segmentation confidence which can robustly discriminate between successful and inadequate segmentations.

PLM-T3SE: Accurate Prediction of Type III Secretion Effectors Using Protein Language Model Embeddings.

The Type III secretion effectors (T3SEs) are bacterial proteins synthesized by Gram-negative pathogens and delivered into host cells via the Type III secretion system (T3SS). These effectors usually play a pivotal role in the interactions between bacteria and hosts. Hence, the precise identification of T3SEs aids researchers in exploring the pathogenic mechanisms of bacterial infections. Since the diversity and complexity of T3SE sequences often make traditional experimental methods time-consuming, it is imperative to explore more efficient and convenient computational approaches for T3SE prediction. Inspired by the promising potential exhibited by pre-trained language models in protein recognition tasks, we proposed a method called PLM-T3SE that utilizes protein language models (PLMs) for effective recognition of T3SEs. First, we utilized PLM embeddings and evolutionary features from the position-specific scoring matrix (PSSM) profiles to transform protein sequences into fixed-length vectors for model training. Second, we employed the extreme gradient boosting (XGBoost) algorithm to rank these features based on their importance. Finally, a MLP neural network model was used to predict T3SEs based on the selected optimal feature set. Experimental results from the cross-validation and independent test demonstrated that our model exhibited superior performance compared to the existing models. Specifically, our model achieved an accuracy of 98.1%, which is 1.8%-42.4% higher than the state-of-the-art predictors based on the same independent data set test. These findings highlight the superiority of the PLM-T3SE and the remarkable characterization ability of PLM embeddings for T3SE prediction.