Validation Dataset Research Articles

OPTIMIZING TIME SERIES FORECASTING: LEVERAGING MACHINE LEARNING MODELS FOR ENHANCED PREDICTIVE ACCURACY

Engaging in investment activities plays a crucial and strategic role in fostering the growth of businesses and ensuring their resilience in the market. This involvement entails expenditures on acquiring assets, embracing technological advancements, expanding production capacities, conducting research and development, among various other domains. Collectively, these aspects form the foundation for the sustained success of an organization over the long term. This thesis will delve into an exploration of leveraging machine learning techniques to forecast key parameters in business, including investments and their impact on the financial health of the company. In this research, explored a variety of time series models and identified that both the Random Forest Regressor and Decision Tree Regressor models deliver superior accuracy, showcasing identical RMSE values of 88.36 on the validation dataset. Furthermore, the Cat Boost and Light GBM models exhibited praiseworthy performance, registering RMSE values of 92.47 and 104.69, respectively. These findings highlight the robust performance of Random Forest Regressor and Decision Tree Regressor, emphasizing their capability to provide accurate predictions. It is noted that Random Forest Regressor and Decision Tree Regressor are distinguished by high accuracy in time series forecasting, and the choice between them should take into account the trade-offs between computational efficiency and interpretability of the model. These results allow us to propose practical strategies for managing investment resources to ensure the sustainable development and prosperity of the enterprise in the long term.

Capacity Prognostics of Marine Lithium-Ion Batteries Based on ICPO-Bi-LSTM Under Dynamic Operating Conditions

An accurate prognosis of the marine lithium-ion battery capacity is significant in guiding electric ships’ optimal operation and maintenance. Under real-world operating conditions, lithium-ion batteries are exposed to various external factors, making accurate capacity prognostication a complex challenge. The paper develops a marine lithium-ion battery capacity prognostic method based on ICPO-Bi-LSTM under dynamic operating conditions to address this. First, the battery is simulated according to the actual operating conditions of an all-electric ferry, and in each charge/discharge cycle, the sum, mean, and standard deviation of each parameter (current, voltage, energy, and power) during battery charging, as well as the voltage difference before and after the simulated operating conditions, are calculated to extract a series of features that capture the complex nonlinear degradation tendency of the battery, and then a correlation analysis is performed on the extracted features to select the optimal feature set. Next, to address the challenge of determining the neural network’s hyperparameters, an improved crested porcupine optimization algorithm is proposed to identify the optimal hyperparameters for the model. Finally, to prevent the interference of test data during model training, which could lead to evaluation errors, the training dataset is used for parameter fitting, the validation dataset for hyperparameter adjustment, and the test dataset for the model performance evaluation. The experimental results demonstrate that the proposed method achieves high accuracy and robustness in capacity prognostics of lithium-ion batteries across various operating conditions and types.

NMF typing and machine learning algorithm-based exploration of preeclampsia-related mechanisms on ferroptosis signature genes

BackgroundGlobally, pre-eclampsia (PE) poses a major threat to the health and survival of pregnant women and fetuses, contributing significantly to morbidity and mortality. Recent studies suggest a pathological link between PE and ferroptosis. We aim to utilize non-negative matrix factorization (NMF) clustering and machine learning algorithms to pinpoint disease-specific genes related to the process of ferroptosis in PE and investigate likely underlying biochemistry mechanisms.MethodsThe acquisition of four microarray datasets from the Gene Expression Omnibus (GEO) repository, the integration of these datasets, and the elimination of batch effects formed the core procedure. Genes related to ferroptosis in PE (DE-FRG) were identified. NMF clustering was performed on DE-FRG for unsupervised analysis, generating a heatmap for clustering validation via principal component analysis. Immunocyte infiltration differences between different subtypes were compared to elucidate the impact of ferroptosis on immune infiltration in the placental tissue of PE patients. The application of weighted gene co-expression network analysis (WGCNA) revealed important module genes linked to sample subtypes and disease status. The screening of PE feature genes involved employing SVM, RF, GLM, and XGB machine learning algorithms, and their predictive performance was validated using various analyses and an external dataset. The iRegulon tool was utilized to predict upstream transcription factors associated with ferroptosis feature genes, from which differentially expressed transcription factors were screened to construct a "Transcription Factor—FRG—ferroptosis" regulatory network. Finally, in vitro (cultured cells) and in vivo (rat) models were utilized to evaluate the regulatory mechanisms of ferroptosis in normal and PE placental tissues.ResultsDifferential analysis of the four merged GEO datasets identified 41 DE-FRGs. NMF clustering based on DE-FRGs revealed two PE subtypes. Immunocyte infiltration analysis indicated significant differences in immune levels between these subtypes. Further WGCNA analysis identified module genes associated with PE and these two subtypes. Subsequently, we developed an integrated machine learning model incorporating five FRGs and validated its predictive efficacy using various analyses and an external validation dataset. Finally, based on the transcription factor ARID3A and ferroptosis feature genes EPHB3 and PAPPA2, we constructed a "Transcription Factor—FRG—ferroptosis" regulatory network, with in vitro and in vivo experiments confirming that ARID3A promotes the progression of PE and ferroptosis by activating the expression of EPHB3 and PAPPA2.ConclusionThis analytical journey illuminated a critical regulatory nexus in PE, underscoring the central influence of ARID3A on PE through ferroptosis-mediated pathways.Graphical Ferroptosis in Preeclampsia: ARID3A regulates ferroptosis by upregulating EPHB3 and PAPPA2, exacerbating oxidative stress in placental cells.Machine Learning Insights: Integration of bioinformatics and machine learning identified ferroptosis-related genes and their subtypes in preeclampsia.Immune Microenvironment Alterations: Ferroptosis-driven immune infiltration heterogeneity contributes to the progression of preeclampsia.Therapeutic Potential: Targeting the ARID3A-EPHB3/PAPPA2 axis could offer novel interventions for managing preeclampsia.

Integration of bulk and single-cell RNA-seq reveals prognostic gene signatures in patients with bladder cancer treated with immune checkpoint inhibitors

Immune checkpoint inhibitors have significantly advanced research in oncology and are used to successfully treat patients with bladder cancer (BC). However, as the benefits of programmed death-1/ programmed death-ligand-1 blockade immunotherapy do not extend to all patients with BC, biomarkers are required to improve prognostic stratification. This study aims to identify reliable biomarkers to enhance the prediction of treatment outcomes. Bulk RNA expression data from a BC cohort (GSE176307) receiving ICI and single-cell sequencing data from patients with BC (GSE135337) were collected. We identified differentially expressed genes (DEGs) within cells that were associated with favorable survival outcomes and developed a predictive bladder cancer gene signature (BC-GS). Subsequently, we performed pathway enrichment analysis using the Reactome database. We used two independent datasets to validate the BC-GS. Patients with low BC-GS had a significantly shorter overall survival (OS) than those with high BC-GS (p < 0.05, p < 0.001, respectively). Additionally, patients with a concurrently low BC-GS score and low tumor mutation burden (TMB) in GSE176307 and the two validation datasets exhibited an increased risk of death. Genes in the BC-GS were predominantly involved in CD8+ T cell activation, antigen presentation, and immune checkpoint pathways. CIBERSORT analysis revealed differences in CD4+ T cells and macrophages between the high and low BC-GS groups. This study demonstrated the prognostic significance of the BC-GS in patients with BC treated with ICI. The combined assessment of the BC-GS and TMB may provide a sophisticated prognostic approach to enhance patient stratification for ICI treatment in BC.

Development and validation of non-invasive prediction models for assessing kidney histopathological activity index in lupus nephritis.

To develop and validate prediction models for estimating the activity index (AI) of kidney histopathology in lupus nephritis (LN) using clinical and laboratory data. This study used single-center data from patients with kidney biopsy-confirmed LN between January 2012 and March 2022. The development and validation datasets were temporally cut. We discriminated AI > 10 and ≤ 10 as high and mild/moderate activity status, respectively. We constructed four models for AI: Model 1 included demographic information; Model 2 additionally incorporated data of systemic conditions; Model 3 further included kidney-specific conditions; and Model 4 included all the aforementioned predictors. Logistic regression was employed in Models 1 to 3, while Model 4 utilized least absolute shrinkage and selection operator for predictor selection and model building. Internal validation was performed using 1000 bootstrap resampling, while external validation was performed in the temporal validation dataset. Both calibration and discrimination metrics were evaluated. There were 160 patients in the development dataset and 70 patients in the validation dataset. In the temporal validation, all the models achieved acceptable calibration and excellent discrimination. Model 2 which contained relatively fewer predictors achieved the highest area under the receiver operator characteristic curve of 0.86 (95% confidence interval 0.76 to 0.94). Our Model 2 incorporating demographic and systemic indicators exhibited good performance in estimating the AI of LN. We thus provide a simple yet effective algorithm to predict AI in patients with LN, potentially aiding clinicians in non-invasively assessing disease activity and guiding treatment decisions. Key Points • We developed a prediction model (Model 2) incorporating demographic and systemic indicators to predict AI in patients with LN. • The prediction model can aid clinicians in noninvasively assessing disease activity and guiding treatment decisions.

Prediction of major adverse cardiovascular events following acute myocardial infarction using electrocardiogram DETERMINE score

BackgroundAcute myocardial infarction (AMI) remains a major cause of mortality and morbidity globally, with a high incidence of major adverse cardiovascular events (MACE) post-primary percutaneous coronary intervention (PPCI). The DETERMINE score, derived from electrocardiographic (ECG) markers, has shown promise as a predictor of adverse outcomes, but its clinical utility requires further validation.ObjectiveTo evaluate the predictive value of the DETERMINE score for MACE and provide early clinical warnings for high-risk patients.MethodsThis bidirectional cohort study included AMI patients from the Second Affiliated Hospital of Anhui Medical University between 2019 and 2023. The training cohort comprised 545 patients between January 2019 and January 2023, while the validation cohort consisted of 122 patients between February 2023 and July 2023. The primary endpoint was MACE within one-year post-PPCI. The relationship between the DETERMINE score and MACE was analyzed using Cox regression, trend tests, and restricted cubic splines to assess linear and nonlinear associations. Patients were stratified into risk groups based on tertiles or optimal cutoffs, and Kaplan-Meier survival curves compared MACE incidence across groups. Predictive accuracy was evaluated through time-dependent C-index, ROC curves, decision curve analysis, and calibration, and compared to other prognostic scores, including the Selvester, GRACE, and SYNTAX scores, as well as left ventricular ejection fraction (LVEF). Subgroup analyses by sex, age, and culprit artery involvement were also conducted.ResultsCox multivariate regression indicated that the DETERMINE score was an independent risk factor for MACE (HR = 1.56, 95% CI 1.38–1.75, P < 0.001). Trend test and RCS showed a positive correlation and non-linear relationship between the DETERMINE score and MACE (P-trend < 0.001, P-overall < 0.001, P-nonlinear: 0.003). Kaplan-Meier survival analysis revealed that, in both the training and validation datasets, groups with a higher DETERMINE score showed a higher cumulative risk of MACE. The DETERMINE score outperformed traditional prognostic scores (Selvester, GRACE, SYNTAX) in terms of predictive accuracy, with an AUROC of 0.840 at 12 months in the training cohort. The score also provided a substantial clinical net benefit, particularly over longer follow-up periods. Subgroup analyses confirmed its predictive power across different demographics and clinical presentations.ConclusionThe DETERMINE score has outstanding predictive power for MACE post-PPCI, which can guide the early identification of high-risk patients with poor prognosis of AMI in clinical practice.

An ultrasound-based nomogram for predicting central lymph node metastasis in papillary thyroid microcarcinoma.

Performing prophylactic central lymph node dissection for papillary thyroid microcarcinoma (PTMC) patients with clinically negative lymph node metastasis remains controversial - not all patients with PTMC are suitable for active surveillance. Therefore, we aimed to establish a nomogram based on ultrasound features for predicting CLNM in PTMC. This retrospective study included 636 patients with PTMC, in which the CLNM status was pathologically confirmed. Univariate and multivariate regression analyses were conducted to screen for risk factors associated with CLNM. Then, a CLNM prediction model was established, receiver operating characteristic, calibration, and decision curve analyses were used to assess the model's performance. Five variables, including age, sex, combined CLNM status, tumor size, and capsule invasion, were included in the nomogram. The values of the area under the receiver operating characteristic curve in the training and validation datasets were 0.720 (95% confidence interval [CI], 0.649-0.791) and 0.704 (95% CI, 0.622-0.786), respectively. An ultrasound-based nomogram was successfully established, of which the predictive model shows excellent predictive performance and can be used to evaluate the status of CLNM in PTMC. Thus, patients with high nomogram scores should be considered for prophylactic central neck dissection.

Prediction and mapping of leaf water content in Populus alba var. pyramidalis using hyperspectral imagery

Leaf water content (LWC) encapsulates critical aspects of tree physiology and is considered a proxy for assessing tree drought stress and the risk of forest decline; however, its measurement relies on destructive sampling and is thus less efficient. Advancements in hyperspectral imaging technology present new prospects for noninvasively evaluating LWC and mapping drought severity across forested regions. In this study, leaf samples were obtained from Populus alba var. pyramidalis, a species widely employed for constructing farmland shelterbelts in water-limited regions of northern China but notably susceptible to drought. These samples were dehydrated to varying degrees to generate concurrent LWC measurements and hyperspectral images, enabling the development of narrow-band and multivariate spectral prediction models for LWC estimation. Two visible-spectrum narrow-band indices identified, the single-band index (R627) and the band subtraction index (R437 − R444), demonstrated a strong correlation with LWC. Despite certain influences of variable preprocessing and selection on multivariate model performance, most models exhibited robust predictive accuracy for LWC. The FDRL-UVE-PLSR combination emerged as the optimal multivariate model, with R2 values reaching 0.9925 and 0.9853 and RMSE values below 0.0124 and 0.0264 for the calibration and validation datasets, respectively. Using this optimal model, along with localized spectral smoothing, moisture distribution across leaf surfaces was visualized, revealing lower water retention at the leaf margins compared to central regions. These methodologies provide critical insights into subtle water-associated physiological processes at the leaf scale and facilitate high-frequency, large-scale assessments and monitoring of drought stress levels and the risk of drought-induced tree mortality and forest degradation in drylands.

Plasma extracellular vesicles carry immune system-related peptides that predict human longevity.

Extracellular vesicles (EVs) play crucial roles in aging. In this National Institutes on Aging-funded study, we sought to identify circulating extracellular vesicle (EV) biomarkers indicative of longevity. The plasma EV proteome of 48 older adults (mean age 77.2 ± 1.7 years [range 72-80]; 50% female, 50% Black, 50% < 2-year survival, 50% ≥ 10-year survival) was analyzed by high-resolution mass spectrometry and flow cytometry. The ability of EV peptides to predict longevity was evaluated in discovery (n = 32) and validation (n = 16) datasets with areas under receiver operating characteristic curves (AUCs). Longevity-associated large EV (LEV) plasma subpopulations were mainly related to immune cells (HLA-ABC+, CD9+, and CD31+) and muscle cells (MCAD+ and RyR2+). Of 7960 identified plasma EV peptides (519 proteins), 46.4% were related to the immune system and 10.1% to muscle. Compared with short-lived older adults, 756 EV peptides (131 proteins) had a higher abundance, and 130 EV peptides (78 proteins) had a lower abundance in long-lived adults. Among longevity-associated peptides, 437 (58 proteins) were immune system related, and 12 (2 proteins) were muscle related. Using just three to five plasma EV peptides (mainly complement components C2-C6), we achieved high predictive accuracy for longevity (AUC range 0.91-1 in a hold-out validation dataset). Our findings suggest that immune cells produce longevity-associated plasma EVs and elucidate fundamental mechanisms regulating aging and longevity. EV longevity predictors suggest there may be merit in targeting complement pathways to extend lifespan, for instance, with any one of the multiple complement inhibitors currently available or in clinical development.

Development and internal validation of El-Shazly-Buchholz's nomogram to predict postoperative complications after PCNL: A multicenter study.

A model to predict the risk of surgical complications following percutaneous nephrolithotomy (PCNL) could be a useful tool to guide clinical decision-making. The aim of this study was to develop a simple and widely applicable stratification tool to be used for patient counseling, surgical planning, evaluation of outcomes, and academic reporting. Data of patients who underwent PCNL were retrieved from the database of the collaborating centers including demographics of patients, characteristics of their stones and urinary tracts, and perioperative data. The primary outcome was the development of postoperative complications. Data were randomly split into a training dataset (85%) and a validation dataset (15%). A univariate and multivariate logistic regression analysis of the training dataset was performed to identify independent predictors of postoperative complications. Model variables were used to construct a nomogram that was internally validated on the testing dataset by measuring calibration, discrimination, and plotting the decision curve. Six hundred thirty one patients (245 Males) with a median (IQR) age of 49 (37-56) years were included. Post-operative complications occurred in 147 (23.3%) patients. Significant predictors of complications included preoperative urine culture (p < 0.001), largest stone diameter (p = 0.02), and intraoperative blood loss (p = 0.002). A nomogram was developed from the predictors and applied to the validation dataset showing an area under the curve (95%CI) of 66.4% (52.2;80.6). This new scoring system emphasized patient characteristics and operative details rather than stone characters to predict the morbidity of PCNL. Furthermore, it should facilitate risk adjustment, enabling physicians to better define the nephrolithiasis disease continuum and identify patients who should be referred to tertiary care centers.

A light-weight factorized convolutions based dual-input fuzzy-CNN for efficient motor bearing fault diagnosis

ABSTRACT Efficient and timely identification of bearing faults is imperative to ensure operational normalcy, reduced down-times and health hazards in motor fault tolerant control systems. This paper proposes a fault diagnosis method that combines the vibration information with time-varying rotational speed for effective fault diagnosis under non-stationary conditions. The complex wavelet transform is used to encode both the vibrational and rotational signals in 2d representations for spatial feature extraction. An efficient algorithm is proposed to select the mother wavelet with the least average entropy. Moreover, a spatial decomposition-based approach using factorised convolutions is used to create a light-weight fuzzy convolutional neural network named Split-Operation Fuzzy Convolutional Neural Network (SOF-CNN) for semi rule-based feature extraction and classification. The performance was evaluated on the University of Ottawa (UOO) dataset for multiple speed conditions and cross-condition validation with the highest accuracy yield being 99.92% from the fourth condition and 3rd trial acquired via averaged 10-Fold Cross Validation. The average accuracy yield across all the scenario was 99.89% with 64.73% being the highest accuracy for cross-condition validation. The performance was evaluated across a diverse range of evaluation criterion including both quantitative and statistical tests.

Uncertainty-aware automatic TNM staging classification for [18F] Fluorodeoxyglucose PET-CT reports for lung cancer utilising transformer-based language models and multi-task learning

Background[18F] Fluorodeoxyglucose (FDG) PET-CT is a clinical imaging modality widely used in diagnosing and staging lung cancer. The clinical findings of PET-CT studies are contained within free text reports, which can currently only be categorised by experts manually reading them. Pre-trained transformer-based language models (PLMs) have shown success in extracting complex linguistic features from text. Accordingly, we developed a multi-task ‘TNMu’ classifier to classify the presence/absence of tumour, node, metastasis (‘TNM’) findings (as defined by The Eight Edition of TNM Staging for Lung Cancer). This is combined with an uncertainty classification task (‘u’) to account for studies with ambiguous TNM status.Methods2498 reports were annotated by a nuclear medicine physician and split into train, validation, and test datasets. For additional evaluation an external dataset (n = 461 reports) was created, and annotated by two nuclear medicine physicians with agreement reached on all examples. We trained and evaluated eleven publicly available PLMs to determine which is most effective for PET-CT reports, and compared multi-task, single task and traditional machine learning approaches.ResultsWe find that a multi-task approach with GatorTron as PLM achieves the best performance, with an overall accuracy (all four tasks correct) of 84% and a Hamming loss of 0.05 on the internal test dataset, and 79% and 0.07 on the external test dataset. Performance on the individual TNM tasks approached expert performance with macro average F1 scores of 0.91, 0.95 and 0.90 respectively on external data. For uncertainty an F1 of 0.77 is achieved.ConclusionsOur ‘TNMu’ classifier successfully extracts TNM staging information from internal and external PET-CT reports. We concluded that multi-task approaches result in the best performance, and better computational efficiency over single task PLM approaches. We believe these models can improve PET-CT services by assisting in auditing, creating research cohorts, and developing decision support systems. Our approach to handling uncertainty represents a novel first step but has room for further refinement.

Learning to Train and to Explain a Deep Survival Model with Large-Scale Ovarian Cancer Transcriptomic Data.

Background/Objectives: Ovarian cancer is a complex disease with poor outcomes that affects women worldwide. The lack of successful therapeutic options for this malignancy has led to the need to identify novel biomarkers for patient stratification. Here, we aim to develop the outcome predictors based on the gene expression data as they may serve to identify categories of patients who are more likely to respond to certain therapies. Methods: We used The Cancer Genome Atlas (TCGA) ovarian cancer transcriptomic data from 372 patients and approximately 16,600 genes to train and evaluate the deep learning survival models. In addition, we collected an in-house validation dataset of 12 patients to assess the performance of the trained survival models for their direct use in clinical practice. Despite deceptive generalization capabilities, we demonstrated how our model can be interpreted to uncover biological processes associated with survival. We calculated the contributions of the input genes to the output of the best trained model and derived the corresponding molecular pathways. Results: These pathways allowed us to stratify the TCGA patients into high-risk and low-risk groups (p-value 0.025). We validated the stratification ability of the identified pathways on the in-house dataset consisting of 12 patients (p-value 0.229) and on the external clinical and molecular dataset consisting of 274 patients (p-value 0.006). Conclusions: The deep learning-based models for survival prediction with RNA-seq data could be used to detect and interpret the gene-sets associated with survival in ovarian cancer patients and open a new avenue for future research.

Landslide susceptibility mapping using combined geospatial, FR and AHP models: a case study from Ethiopia’s highlands

This study performed landslide susceptibility mapping in Awabel Woreda, situated in the east Gojjam zone of the Amhara region in Ethiopia. The occurrence of landslides and slope instability is widespread in Awabel Woreda, leading to the devastation of agricultural fields, crops, and residences, the demise of animal life, and the forced relocation of local inhabitants from their dwellings. The present investigation utilized remote sensing and GIS techniques in conjunction with the Frequency Ratio (FR) and Analytical Hierarchy Process (AHP) models to map landslide risk zones. For the landslide susceptibility mapping of this study, nine causative factors, such as “elevation, slope, aspect, drainage density, lineament density, land use and land cover (LULC), soil texture, rainfall, and lithology”, were considered. A total of 130 past landslide events were identified via field survey and Google Earth image, of which 70% (91) of them were used as training datasets and 30% (39) were used as validation datasets of the proposed models (i.e., FR and AHP). The nine contributing variables and their classes were evaluated, and factor weights were computed using the IDRISI Selva 17.0 expert programme. The landslide susceptibility indexes (LSI) of the FR and AHP models were calculated and categorized into five relative zones using ArcGIS 10.7. The landslide susceptibility map (LSM) produced by the FR model shows that 93.2 km2 (14.52%) of the study area is classified as very low susceptibility, 167.51 km2 (26.09%) as low susceptibility, 174.10 km2 (27.12%) as moderate susceptibility, 137.03 km2 (21.34%) as high susceptibility, and 70.30 km2 (10.93%) as very high susceptibility to landslides. Based on the AHP model's LSM, different landslide susceptibility zones were identified in the study area. Specifically, 140.46 km2 (21.88%) of the region falls into the very low susceptibility zone, 116.78 km2 (18.59%) falls into the low susceptibility zone, 147.94 km2 (23.04%) falls into the moderate susceptibility zone, 154.04 km2 (23.99%) falls into the high susceptibility zone, and 82.78 km2 (12.89%) falls into the very high susceptibility zone. The validation investigation demonstrated that the FR and AHP models had accuracy rates of 89.73 and 87.18%, respectively. The FR model exhibited marginally more accurate results than AHP, primarily because of the direct correlation between previous and current occurrences of landslides. Nevertheless, the AHP model’s effectiveness relies on the individual’s expertise and the characteristics of the components that cause the outcome. The landslide susceptibility maps generated through these models provide valuable insights for land management and disaster mitigation efforts, with delineated zones indicating very low to very high susceptibility areas.

Metabolomic studies reveal and validate potential biomarkers of diabetic retinopathy in two Chinese datasets with type 2 diabetes: a cross-sectional study

BackgroundDiabetic retinopathy (DR) is a major microvascular complication of diabetes mellitus and causes vision impairment and blindness. The presence of major risk factors for DR, such as high levels of HbA1c, does not predict all DR pathogenesis in the clinic, which suggests that uncovering the underlying mechanisms and identifying novel markers are needed. Previous evidence has shown that the serum metabolic signature of DR is unique and detectable compared with that of diabetes mellitus (DM). Here, we aimed to identify serum metabolites as reliable biomarkers for the presence of DR in type 2 DM (T2DM) patients.MethodsWe performed untargeted and targeted metabolomic studies using liquid chromatography‒mass spectrometry (LC‒MS) and multiple reaction monitoring (MRM) methods on the serum samples of T2DM patients. For the discovery dataset, 39 DR patients and 39 non-DR (NDR) patients were included. For the validation dataset, 95 DR patients and 95 non-DR (NDR) patients were included. Receiver operating characteristic curve analysis was performed to evaluate the discriminating power of the metabolites. Binary logistic regression models were fit to evaluate the associations of metabolite peak areas or neurotransmitter concentrations with the presence of DR and adjusted for known risk factors.ResultsA total of 7123 metabolites were tested. The 39 DR patients had a mean age of 56 years with an average diabetes duration of 12 years, and the 39 NDR patients had a mean age of 57 years with an average diabetes duration of 11 years. Nine serum candidate markers were further identified. Six out of nine markers were associated with DR after we adjusted for covariates, including blood pressure, HbA1c, diabetes duration, fasting blood glucose, triglyceride, eGFR etc. Among them, eicosapentaenoic acid (EPA) and L-tyrosine were validated in an independent, risk factor-matched sample set. The serum L-tyrosine concentration was decreased in DR group by 47% (-0.22 ± 0.87 vs. 0.48 ± 1.05, P < 0.001), of which the cutoff value was 0.10 mg/ml, with 86% sensitivity and 40% specificity (AUC = 0.62, 95% CI = 0.54–0.70, P = 0.005).ConclusionsLow levels of circulating L-Tyrosine indicate retinopathy occurrence in T2DM population.

A Novel Effective Models for Identifying BRCA Patients and Optimizing Clinical Treatments.

This study aimed to develop an effective model that identifies high-risk breast cancer (BRCA) patients and optimizes clinical treatments. This study includes five public datasets, TCGA-BRCA as the training dataset and other cohorts as the validation datasets. Machine learning algorithms for finding key tumor-associated immune gene pairs (TAIGPs). These TAIGPs were used to construct tumor-associated immune gene pair index (TAIGPI) by multivariate analysis and further validated on the validation datasets. In addition, the differences in clinical prognosis, biological characteristics, and treatment benefits between high and low TAIGPI groups were further analyzed. The TAIGPI was established by 36 TAIGPs. Better clinical outcomes in the low TAIGPI patients, with consistent results, were also obtained in the validation datasets. The study showed that patients in the low TAIGPI group had a high infiltration of immune cells and low proliferative activity of tumor cells. In contrast, patients in the high TAIGPI group exhibited low infiltration of immune cells and high proliferative activity of tumor cells. In addition, patients in the low TAIGPI group are more likely to benefit from chemotherapy, adjuvant chemotherapy, or immunotherapy. The TAIGPI can be an effective predictive strategy for the clinical prognosis of breast cancer patients, providing new insights into personalized treatment options for breast cancer patients.

Preliminary clinical performance of a Cas13a-based lateral flow assay for detecting Neisseria gonorrhoeae in urine specimens.

Nucleic acid amplification testing (NAAT) for N. gonorrhoeae is unavailable in resource-limited settings. We previously developed a CRISPR-based lateral flow assay for detecting N. gonorrhoeae. We aimed to pair that assay with point-of-care DNA extraction, assess performance in clinical urine specimens, and optimize assay kinetics. We collected urine specimens among men presenting with urethritis enrolling in a clinical trial at the Massachusetts General Hospital Sexual Health Clinic. We assessed the quantified DNA yield of detergent-based extractions with and without heat. We selected one detergent for extracting all specimens, paired with isothermal recombinase polymerase amplification for 90 minutes and lateral flow Cas13a detection, interpreted via pixel intensity analysis. We also trained a smartphone-based machine-learning model on 1,008 images to classify lateral flow results. We used the model to interpret lateral flow results from the clinical specimens. We also tested a modified amplification chemistry with a second forward primer lacking the T7-promoter to accelerate reaction kinetics. Extraction with 0.02% Triton X resulted in an average DNA yield of 2.6 × 106 copies/µL (SD ± 6.7 × 105). We treated 40 urine specimens (n = 12 positive) with 0.02% Triton X, and using quantified pixel intensity analysis, the Cas13a-based assay correctly classified all specimens (100% agreement; 95% CI 91.2%-100%). The machine-learning model correctly classified 45/45 strips in the validation data set and all 40 lateral flow strips from clinical specimens. Including the second forward primer reduced incubation time to 60 minutes. Using point-of-care DNA extraction, our Cas13a-based lateral flow N. gonorrhoeae assay demonstrated promising performance among clinical urine specimens.IMPORTANCEUsing a CRISPR-based assay we previously developed for Neisseria gonorrhoeae detection, we developed new techniques to facilitate point-of-care use. We then demonstrated the promising performance of that assay in clinical specimens. Furthermore, we developed a smartphone-based machine learning application for assisting interpretation of lateral flow strip results. Such an assay has the potential to transform the care of sexually transmitted infections in low-resource settings where diagnostic tests are unavailable. A point-of-care pathogen-specific assay, paired with the connectivity offered by a smartphone application, can also support public health surveillance efforts in such areas.

Interpolation of Imaging Mass Spectrometry Data by a Window-Based Adversarial Autoencoder Method.

Imaging mass spectrometry (IMS) is a technique for simultaneously acquiring the expression and distribution of molecules on the surface of a sample, and it plays a crucial role in spatial omics research. In IMS, the time cost and instrument load required for large data sets must be considered, as IMS typically involves tens of thousands of pixels or more. In this study, we developed a high-resolution method for IMS data reconstruction using a window-based Adversarial Autoencoder (AAE) method. We acquired IMS data from partial cerebellum regions of mice with a pitch size of 75 μm and then down-sampled the data to a pitch size of 150 μm, selecting 22 m/z peak intensity values per pixel. We established an AAE model to generate three pixels from the surrounding nine pixels within a window to reconstruct the image data at a pitch size of 75 μm. Compared with two alternative interpolation methods, Bilinear and Bicubic interpolation, our window-based AAE model demonstrated superior performance on image evaluation metrics for the validation data sets. A similar model was constructed for larger mouse kidney tissues, where the AAE model achieved high image evaluation metrics. Our method is expected to be valuable for IMS measurements of large animal organs across extensive areas.

Use Self-Training Random Forest for Predicting Winter Wheat Yield

The effectiveness of supervised ML heavily depends on having a large, accurate, and diverse annotated dataset, which poses a challenge in applying ML for yield prediction. To address this issue, we developed a self-training random forest algorithm capable of automatically expanding the annotated dataset. Specifically, we trained a random forest regressor model using a small amount of annotated data. This model was then utilized to generate new annotations, thereby automatically extending the training dataset through self-training. Our experiments involved collecting data from over 30 winter wheat varieties during the 2019–2020 and 2021–2022 growing seasons. The testing results indicated that our model achieved an R2 of 0.84, RMSE of 627.94 kg/ha, and MAE of 516.94 kg/ha in the test dataset, while the validation dataset yielded an R2 of 0.81, RMSE of 692.96 kg/ha, and MAE of 550.62 kg/ha. In comparison, the standard random forest resulted in an R2 of 0.81, RMSE of 681.02 kg/ha, and MAE of 568.97 kg/ha in the test dataset, with validation results of an R2 of 0.79, RMSE of 736.24 kg/ha, and MAE of 585.85 kg/ha. Overall, these results demonstrate that our self-training random forest algorithm is a practical and effective solution for expanding annotated datasets, thereby enhancing the prediction accuracy of ML models in winter wheat yield forecasting.

Factors Influencing the Firm Performance of Digital Merchandising Businesses in Thailand

Thailand's national strategy focuses on making the country more competitive by helping modern entrepreneurs succeed both within the country and internationally. This research aims for development a new framework that shows how critical elements like market capabilities, IT capabilities, digitalization, and innovation contribute to the success of digital businesses in Thailand. The study uses a mixed method research, combining an integration of qualitative and quantitative techniques are utilized. Within qualitative research, we gathering insights from 20 experts, by using e-Rough Set Delphi method. In quantitative research, The research was based on a collection of 611 valid datasets, which were gathered from digital merchandising business owners in Thailand via online questionnaires. The results indicate that market capabilities, IT capabilities, digitalization, and innovation all have an influential and meaningful effect on business performance. Research also emphasizes importance of investing in IT infrastructure, adopting advanced IT strategies, and improving market capabilities to stay competitive in the fast-changing digital market. The insights from this study are valuable for strengthening the digital economy, as adopting digital technologies and innovations can help businesses grow sustainably and positively impact the overall economy and society.