Development of the lyrics-based deep learning algorithm for identifying alcohol-related words (LYDIA)
Ontology highlight
ABSTRACT:
SUBMITTER:
PROVIDER: S-EPMC10794165 | biostudies-literature | 2024 Jan
REPOSITORIES: biostudies-literature
Similar Datasets
Project description: Not available
| S-EPMC11297026 | biostudies-literature
Project description:Dyskalemia is a common electrolyte abnormality. Since dyskalemia can cause fatal arrhythmias and cardiac arrest in severe cases, it is crucial to monitor serum potassium (K+) levels on time. We developed deep learning models to detect hyperkalemia (K+ ≥ 5.5 mEq/L) and hypokalemia (K+ < 3.5 mEq/L) from electrocardiograms (ECGs), which are noninvasive and can be quickly measured. The retrospective cohort study was conducted at two hospitals from 2006 to 2020. The training set, validation set, internal testing cohort, and external validation cohort comprised 310,449, 15,828, 23,849, and 130,415 ECG-K+ samples, respectively. Deep learning models demonstrated high diagnostic performance in detecting hyperkalemia (AUROC 0.929, 0.912, 0.887 with sensitivity 0.926, 0.924, 0.907 and specificity 0.706, 0.676, 0.635 for 12-lead, limb-lead, lead I ECGs) and hypokalemia (AUROC 0.925, 0.896, 0.885 with sensitivity 0.912, 0.896, 0.904 and specificity 0.790, 0.734, 0.694) in the internal testing cohort. The group predicted to be positive by the hyperkalemia model showed a lower 30-day survival rate compared to the negative group (p < 0.001), supporting the clinical efficacy of the model. We also compared the importance of ECG segments (P, QRS, and T) on dyskalemia prediction of the model for interpretability. By applying these models in clinical practice, it will be possible to diagnose dyskalemia simply and quickly, thereby contributing to the improvement of patient outcomes.
| S-EPMC11445469 | biostudies-literature
Project description:ObjectiveAlthough distinctive structural abnormalities occur in patients with schizophrenia, detecting schizophrenia with magnetic resonance imaging (MRI) remains challenging. This study aimed to detect schizophrenia in structural MRI data sets using a trained deep learning algorithm.MethodFive public MRI data sets (BrainGluSchi, COBRE, MCICShare, NMorphCH, and NUSDAST) from schizophrenia patients and normal subjects, for a total of 873 structural MRI data sets, were used to train a deep convolutional neural network.ResultsThe deep learning algorithm trained with structural MR images detected schizophrenia in randomly selected images with reliable performance (area under the receiver operating characteristic curve [AUC] of 0.96). The algorithm could also identify MR images from schizophrenia patients in a previously unencountered data set with an AUC of 0.71 to 0.90. The deep learning algorithm's classification performance degraded to an AUC of 0.71 when a new data set with younger patients and a shorter duration of illness than the training data sets was presented. The brain region contributing the most to the performance of the algorithm was the right temporal area, followed by the right parietal area. Semitrained clinical specialists hardly discriminated schizophrenia patients from healthy controls (AUC: 0.61) in the set of 100 randomly selected brain images.ConclusionsThe deep learning algorithm showed good performance in detecting schizophrenia and identified relevant structural features from structural brain MRI data; it had an acceptable classification performance in a separate group of patients at an earlier stage of the disease. Deep learning can be used to delineate the structural characteristics of schizophrenia and to provide supplementary diagnostic information in clinical settings.
| S-EPMC7008229 | biostudies-literature
Project description:BackgroundNew drug treatments are regularly approved, and it is challenging to remain up-to-date in this rapidly changing environment. Fast and accurate visualization is important to allow a global understanding of the drug market. Automation of this information extraction provides a helpful starting point for the subject matter expert, helps to mitigate human errors, and saves time.ObjectiveWe aimed to semiautomate disease population extraction from the free text of oncology drug approval descriptions from the BioMedTracker database for 6 selected drug targets. More specifically, we intended to extract (1) line of therapy, (2) stage of cancer of the patient population described in the approval, and (3) the clinical trials that provide evidence for the approval. We aimed to use these results in downstream applications, aiding the searchability of relevant content against related drug project sources.MethodsWe fine-tuned a state-of-the-art deep learning model, Bidirectional Encoder Representations from Transformers, for each of the 3 desired outputs. We independently applied rule-based text mining approaches. We compared the performances of deep learning and rule-based approaches and selected the best method, which was then applied to new entries. The results were manually curated by a subject matter expert and then used to train new models.ResultsThe training data set is currently small (433 entries) and will enlarge over time when new approval descriptions become available or if a choice is made to take another drug target into account. The deep learning models achieved 61% and 56% 5-fold cross-validated accuracies for line of therapy and stage of cancer, respectively, which were treated as classification tasks. Trial identification is treated as a named entity recognition task, and the 5-fold cross-validated F1-score is currently 87%. Although the scores of the classification tasks could seem low, the models comprise 5 classes each, and such scores are a marked improvement when compared to random classification. Moreover, we expect improved performance as the input data set grows, since deep learning models need to be trained on a large enough amount of data to be able to learn the task they are taught. The rule-based approach achieved 60% and 74% 5-fold cross-validated accuracies for line of therapy and stage of cancer, respectively. No attempt was made to define a rule-based approach for trial identification.ConclusionsWe developed a natural language processing algorithm that is currently assisting subject matter experts in disease population extraction, which supports health authority approvals. This algorithm achieves semiautomation, enabling subject matter experts to leverage the results for deeper analysis and to accelerate information retrieval in a crowded clinical environment such as oncology.
| S-EPMC10337300 | biostudies-literature
Project description:BackgroundProtein electrophoresis (PEP) is an important tool in supporting the analytical characterization of protein status in diseases related to monoclonal components, inflammation, and antibody deficiency. Here, we developed a deep learning-based PEP classification algorithm to supplement the labor-intensive PEP interpretation and enhance inter-observer reliability.MethodsA total of 2,578 gel images and densitogram PEP images from January 2018 to July 2019 were split into training (80%), validation (10%), and test (10.0%) sets. The PEP images were assessed based on six major findings (acute-phase protein, monoclonal gammopathy, polyclonal gammopathy, hypoproteinemia, nephrotic syndrome, and normal). The images underwent processing, including color-to-grayscale and histogram equalization, and were input into neural networks.ResultsUsing densitogram PEP images, the area under the receiver operating characteristic curve (AUROC) for each diagnosis ranged from 0.873 to 0.989, and the accuracy for classifying all the findings ranged from 85.2% to 96.9%. For gel images, the AUROC ranged from 0.763 to 0.965, and the accuracy ranged from 82.0% to 94.5%.ConclusionsThe deep learning algorithm demonstrated good performance in classifying PEP images. It is expected to be useful as an auxiliary tool for screening the results and helpful in environments where specialists are scarce.
| S-EPMC9401151 | biostudies-literature
Project description:IntroductionActive tuberculosis (ATB), instigated by Mycobacterium tuberculosis (M.tb), rises as a primary instigator of morbidity and mortality within the realm of infectious illnesses. A significant portion of M.tb infections maintain an asymptomatic nature, recognizably termed as latent tuberculosis infections (LTBI). The complexities inherent to its diagnosis significantly hamper the initiatives aimed at its control and eventual eradication.MethodologyUtilizing the Gene Expression Omnibus (GEO), we procured two dedicated microarray datasets, labeled GSE39940 and GSE37250. The technique of weighted correlation network analysis was employed to discern the co-expression modules from the differentially expressed genes derived from the first dataset, GSE39940. Consequently, a pyroptosis-related module was garnered, facilitating the identification of a pyroptosis-related signature (PRS) diagnostic model through the application of a neural network algorithm. With the aid of Single Sample Gene Set Enrichment Analysis (ssGSEA), we further examined the immune cells engaged in the pyroptosis process in the context of active ATB. Lastly, dataset GSE37250 played a crucial role as a validating cohort, aimed at evaluating the diagnostic prowess of our model.ResultsIn executing the Weighted Gene Co-expression Network Analysis (WGCNA), a total of nine discrete co-expression modules were lucidly elucidated. Module 1 demonstrated a potent correlation with pyroptosis. A predictive diagnostic paradigm comprising three pyroptosis-related signatures, specifically AIM2, CASP8, and NAIP, was devised accordingly. The established PRS model exhibited outstanding accuracy across both cohorts, with the area under the curve (AUC) being respectively articulated as 0.946 and 0.787.ConclusionThe present research succeeded in identifying the pyroptosis-related signature within the pathogenetic framework of ATB. Furthermore, we developed a diagnostic model which exuded a remarkable potential for efficient and accurate diagnosis.
| S-EPMC10646574 | biostudies-literature
Project description:Background and objectivesScreening and early diagnosis for heart failure (HF) are critical. However, conventional screening diagnostic methods have limitations, and electrocardiography (ECG)-based HF identification may be helpful. This study aimed to develop and validate a deep-learning algorithm for ECG-based HF identification (DEHF).MethodsThe study involved 2 hospitals and 55,163 ECGs of 22,765 patients who performed echocardiography within 4 weeks were study subjects. ECGs were divided into derivation and validation data. Demographic and ECG features were used as predictive variables. The primary endpoint was detection of HF with reduced ejection fraction (HFrEF; ejection fraction [EF]≤40%), and the secondary endpoint was HF with mid-range to reduced EF (≤50%). We developed the DEHF using derivation data and the algorithm representing the risk of HF between 0 and 1. We confirmed accuracy and compared logistic regression (LR) and random forest (RF) analyses using validation data.ResultsThe area under the receiver operating characteristic curves (AUROCs) of DEHF for identification of HFrEF were 0.843 (95% confidence interval, 0.840-0.845) and 0.889 (0.887-0.891) for internal and external validation, respectively, and these results significantly outperformed those of LR (0.800 [0.797-0.803], 0.847 [0.844-0.850]) and RF (0.807 [0.804-0.810], 0.853 [0.850-0.855]) analyses. The AUROCs of deep learning for identification of the secondary endpoint was 0.821 (0.819-0.823) and 0.850 (0.848-0.852) for internal and external validation, respectively, and these results significantly outperformed those of LR and RF.ConclusionsThe deep-learning algorithm accurately identified HF using ECG features and outperformed other machine-learning methods.
| S-EPMC6597456 | biostudies-literature
Project description:Purpose: The number of patients with alcohol-related problems is steadily increasing. A large-scale survey of alcohol-related problems has been conducted. However, studies that predict hazardous drinkers and identify which factors contribute to the prediction are limited. Thus, the purpose of this study was to predict hazardous drinkers and the severity of alcohol-related problems of patients using a deep learning algorithm based on a large-scale survey data. Materials and Methods: Datasets of National Health and Nutrition Examination Survey of South Korea (K-NHANES), a nationally representative survey for the entire South Korean population, were used to train deep learning and conventional machine learning algorithms. Datasets from 69,187 and 45,672 participants were used to predict hazardous drinkers and the severity of alcohol-related problems, respectively. Based on the degree of contribution of each variable to deep learning, it was possible to determine which variable contributed significantly to the prediction of hazardous drinkers. Results: Deep learning showed the higher performance than conventional machine learning algorithms. It predicted hazardous drinkers with an AUC (Area under the receiver operating characteristic curve) of 0.870 (Logistic regression: 0.858, Linear SVM: 0.849, Random forest classifier: 0.810, K-nearest neighbors: 0.740). Among 325 variables for predicting hazardous drinkers, energy intake was a factor showing the greatest contribution to the prediction, followed by carbohydrate intake. Participants were classified into Zone I, Zone II, Zone III, and Zone IV based on the degree of alcohol-related problems, showing AUCs of 0.881, 0.774, 0.853, and 0.879, respectively. Conclusion: Hazardous drinking groups could be effectively predicted and individuals could be classified according to the degree of alcohol-related problems using a deep learning algorithm. This algorithm could be used to screen people who need treatment for alcohol-related problems among the general population or hospital visitors.
| S-EPMC8299053 | biostudies-literature
Project description:This paper proposes Deep Semantic Mining based Recommendation (DSMR), which can extract user features and item attribute features more accurately by deeply mining the semantic information of review text and item description documents recommend. First, the proposed model uses the BERT pre-training model to process review texts and item description documents, and deeply mine user characteristics and item attributes, which effectively alleviates the problems of data sparseness and item cold start; Then, the forward LSTM is used to pay attention to the changes of user preferences over time, and a more accurate recommendation is obtained; finally, in the model training stage, the experimental data are randomly divided into 1 to 5 points, 1:1:1:1:1. Extraction ensures that the amount of data for each score is equal, so that the results are more accurate and the model is more robust. Experiments are carried out on four commonly used Amazon public data sets, and the results show that with the root mean square error as the evaluation index, the error of DSMR recommendation results is at least 11.95% lower on average than the two classic recommendation models based only on rating data. At the same time, it is better than the three latest recommendation models based on review text, and it is 5.1% lower than the best model on average.
| S-EPMC9512199 | biostudies-literature
Project description:BackgruoundSince image-based fracture prediction models using deep learning are lacking, we aimed to develop an X-ray-based fracture prediction model using deep learning with longitudinal data.MethodsThis study included 1,595 participants aged 50 to 75 years with at least two lumbosacral radiographs without baseline fractures from 2010 to 2015 at Seoul National University Hospital. Positive and negative cases were defined according to whether vertebral fractures developed during follow-up. The cases were divided into training (n=1,416) and test (n=179) sets. A convolutional neural network (CNN)-based prediction algorithm, DeepSurv, was trained with images and baseline clinical information (age, sex, body mass index, glucocorticoid use, and secondary osteoporosis). The concordance index (C-index) was used to compare performance between DeepSurv and the Fracture Risk Assessment Tool (FRAX) and Cox proportional hazard (CoxPH) models.ResultsOf the total participants, 1,188 (74.4%) were women, and the mean age was 60.5 years. During a mean follow-up period of 40.7 months, vertebral fractures occurred in 7.5% (120/1,595) of participants. In the test set, when DeepSurv learned with images and clinical features, it showed higher performance than FRAX and CoxPH in terms of C-index values (DeepSurv, 0.612; 95% confidence interval [CI], 0.571 to 0.653; FRAX, 0.547; CoxPH, 0.594; 95% CI, 0.552 to 0.555). Notably, the DeepSurv method without clinical features had a higher C-index (0.614; 95% CI, 0.572 to 0.656) than that of FRAX in women.ConclusionDeepSurv, a CNN-based prediction algorithm using baseline image and clinical information, outperformed the FRAX and CoxPH models in predicting osteoporotic fracture from spine radiographs in a longitudinal cohort.
| S-EPMC9449110 | biostudies-literature