Informatics in Medicine Unlocked最新文献

Introduction

In intensity-modulated radiation therapy (IMRT) techniques, although the dose conformity increases, the out-of-field doses would not decrease. This study aimed to assess the dose error calculated by the treatment planning system (TPS) in the out-of-field regions using the dynamic IMRT (D-IMRT) method in nasopharyngeal cancer (NPC) patients.

Methods

The out-of-field doses were measured for the chiasm and parotid organs using the D-IMRT technique (6 MV energy) with Monaco TPS. Computed tomography (CT) images of 10 NPC patients (54–77 years, mean: 61.6 ± 12.2 years) were considered and countered using 7-field and 11-field methods. The OCTAVIUS 4D phantom was utilized for dose assessment.

Results

According to the OCTAVIUS measurements, the Monaco TPS dose errors ranged from −58.8 to 105.5%. The average dose error for optic chiasm and parotid organs was −25% and 8.5%, respectively, with several cases falling within tolerance (±5%).

Conclusion

There were considerable dose calculation errors by Monaco TPS for organs located in out-of-field regions (optic chiasm and parotid) during IMRT for NPC patients. Therefore, accurate dose estimation in the out-of-field regions should be considered in clinical practices.

Helicobacter pylori (H. pylori) is mainly considered for causing chronic gastritis, which can lead to several secondary complications like peptic ulcer and pre-malignant lesions for example atrophic gastritis, intestinal dysplasia and metaplasia, with the etiological factor of developing gastric cancer. Recent research demonstrates that H.pylori colonizes the stomach mucosa of more than fifty populations around the globe. This research focuses on unveiling hub genes, and diagnostic and drug targets against said organism by utilizing various types of networking biology and biophysical approaches. In data retrieval, the GSE19826 dataset was obtained from the gene expression omnibus database and microarray data set from array express. Geo2r analysis predicted a total number of 7 DEGs and 10 hub genes, next functional protein association network analysis (STRING) unveiled that among 10 Hub genes only 3 genes were found more interactive with other genes and involved in pathogenesis, The shortlisted three genes were further analyzed for survival analysis using Gene Expression Profiling Interactive Analysis (GEPIA) and predicted the survival rate of targeted genes. Moreover, functional enchainment analysis was done using the ToppFun server, the server predicted that COL11A1 and COL10A1 were more involved in the pathogenesis of the H. pylori infection. Furthermore, the COL10A1 gene was subjected to protein structure prediction. In molecular docking analysis, the asinex antibacterial library was screened for potential inhibitors, and one compound was predicted as a strong inhibitor with the best binding at −10.23 kcal/mol. The docking results were further validated through molecular dynamic simulation analysis and the MD simulation analysis evaluated the dynamic movement of the docked complex in various nanoseconds, the MD simulation results predicted that the docked complexes are stable throughout the simulation and can be used as a potential inhibitor against the said pathogen, however experimental study is required to further validate the predicted results and design drug against targeted pathogen.

Pancreatic cancer (PC) is a devastating malignancy with intricate genetic underpinnings and a complex etiology. Emerging evidence suggests the presence of lynch syndrome (LS) and type 2 diabetes (T2D) associated susceptibility to PC. This study presents integrated computational and systems biology approaches to identify the genetic risk factors underlying the association between PC, LS, and T2D. Patient data for these three diseases have been collected from NCBI and differentially expressed genes (DEGs) identified by the GREIN web platform. Furthermore, protein-protein interaction (PPI), gene ontology (GO), and signaling pathway networks were analyzed through STRING and DAVID databases, respectively. Autodock Vina has been used for prospective analysis of ligand-protein interaction. About 60 unique common DEGs were identified by statistical analysis. In addition to the utilization of five distinct algorithms within the Cytoscape framework, we have reported three potential target candidates: TNF, CXCL1, and TNFSF10. In particular, the immune and inflammatory response, the chemokine-mediated signaling pathway, rheumatoid arthritis, and IL-17 signaling pathways emerged as prominently enriched pathways. Furthermore, the interaction of 162 phytochemicals from Nigella sativa was assessed with the identified hub proteins. Among these, thujopsene emerged as a notable ligand candidate, demonstrating the most favorable binding energy against the TNF (−9.6 kca/mol TNFSF10 (−8.5 kcal/mol), and CXCL1 (−9.1 kcal/mol) proteins. Besides, pharmacokinetics, toxicity, and drug-likeness properties of the thujopsene ligand showed an acceptable range for selection of a drug candidate. Collectively, these findings shed light on the intricate interplay of genes, pathways, and potential therapeutic compounds, providing a basis for further exploration and validation in the context of relevant diseases.

Nowadays, Polycystic Ovary Syndrome (PCOS) affects many women, making it a prevalent concern. It is a hormonal disorder that causes irregular, delayed, or absent menstrual cycles in the female body. This condition can lead to the development of type 2 diabetes, gestational diabetes, weight gain, unwanted body hair, and various other complications. In severe cases, PCOS can result in infertility, posing a challenge for patients trying to conceive. Statistics show that the incidence rate of PCOS has significantly increased in recent years, which is alarming. If PCOS is identified early, people may follow their doctor's recommendations and live a better life. The dataset used for this research contains records for 541 patients. The aim of this study is to employ machine learning models to identify patterns in this disorder. The information learned is then inputted into various algorithms to assess accuracy, specificity, sensitivity, and precision using different ML models, such as Logistic Regression (LR), Decision Tree (DT), AdaBoost (AB), Random Forest (RF), and Support Vector Machine (SVM) among others. The research utilized the Mutual Information model for feature selection and compared the models to determine the most accurate one. Employing the Mutual Information model for feature engineering, AB and RF achieved the highest accuracy of 94 %.

One of the medicinal herbs utilized in treating diabetes traditionally is Annona muricata. This work investigates the effect of phytochemicals from A. muricata on the therapeutically important protein targets associated with type II diabetes mellitus (T2DM) using a computational approach. Compounds (Phytochemicals) previously identified in A. muricata were docked against proteins of interest to find therapeutic hit compounds. The stability of the ligand-protein complexes was examined after selecting proteins that bind well with the discovered hits, and ADMET properties of the ligands were also predicted to determine their toxicity and drug-likeness. In addition to studying the compounds' softness, hardness, electron affinity, and electrostatic potential, the Schrödinger material science Jaguar fast engine was used to study their frontier molecular orbital (FMO). The targets aldose reductase (ALR), 11beta-hydroxysteroid dehydrogenase type 1 (11-HSD1), and diacylglycerol O-acyltransferase 1 (DGAT1) exhibited the highest binding affinities from the early screening of compounds against fifteen (15) proteins linked with T2DM. While eight (8) phenolic compounds of the plants had comparatively high docking scores with 11β-HSD1 and ALR, seven (7) acetogenins had good binding affinities with DGAT1. These top-scoring compounds exhibited considerable ADMET profiles. Additionally, the phenolic compounds that are considered as hits adhered to the Lipinski rule of 5 and can be thought of as potential drug candidates. Genistein and kaempferol are the most reactive ligands in terms of quantum mechanics. The information from this study could be used to create an alternative anti-diabetic drug with better efficacy.

Prototypical networks (PN) have emerged as one of multiple effective approaches for few-shot learning (FSL), even in medical image classification. This study focuses on implementing a PN for skin lesion classification to assess its performance, generalizability, and robustness when applied across 11 dermoscopic image domains. Unlike conventional FSL scenarios, where the performance is evaluated for unseen classes in the test set, our analysis extends this to evaluate PNs on a complete hold-out dataset with the same classes from a different domain. Differences in a patient’s age, lesion localization, or image acquisition systems variations mimic real-world cross-domain conditions in a clinic. Given the scarcity of medical datasets, this assessment is crucial for potentially translating such systems into real-world clinical settings to support physicians with the diagnosis. Our primary focus is two-fold: investigating whether a PN performs on par with a baseline classifier, even using only a limited number of reference samples from the hold-out test set (in-domain) and whether a PN can generalize to the same classes of unseen domains (cross-domain). Our analysis uncovers that a PN can perform on par with the baseline classifier in an in-domain setting, even with only a few support samples. However, in cross-domain scenarios, a PN exhibits improved performance only on specific domains, while others demonstrate similar or even decreased performance when confronted with a smaller number of images. Our findings contribute to comprehending potential opportunities and limitations of FSL in dermatological practice.

In the dynamic world of healthcare technology, efficiently and accurately extracting medical data from physician–patient conversations is vital. This paper presents a new approach in healthcare technology, employing Natural Language Processing (NLP) to identify and extract critical information from doctor–patient conversations on mobile devices. Unlike traditional methods that rely on Electronic Health Records, our novel application enables the extraction of symptoms, diagnoses, and treatments directly on a mobile device during medical consultations, significantly enhancing patient privacy. We managed to integrate both Bidirectional Encoder Representations from Transformers (BERT) models and optimized Large Language Models (LLMs) on a mobile device without compromising performance significantly. Our findings reveal that the BERT model attained an F1-score of 85.1%, while FLERT and its compressed variant DistilFLERT showed superior performance. The FLAN-T5 model outperformed all models we tested with scores up to 92.7%. These results highlight the efficacy of leveraging advanced NLP and LLM technologies in healthcare environments on a mobile device, offering a promising direction for accessible and efficient patient care.

Introduction

Objective

Methods

Results

Conclusion

Background

Methods

Results

Conclusion

Glioma is the most prevalent primary malignant brain tumor. Segmentation of glioma regions using magnetic resonance imaging (MRI) is essential for treatment planning. However, segmentation of glioma regions is usually based on four MRI modalities, which are T1, T2, T1Gd, and FLAIR. Acquiring these four modalities will increase patients' time inside the scanner and drive up the segmentation process's processing time. Nevertheless, not all these modalities are acquired in some cases due to the limited available time on the MRI scanner or uncooperative patients. Therefore, U-Net-based fully convolutional neural networks were employed for automated segmentation to answer the urgent question: does a smaller number of MRI modalities limit the segmentation accuracy? The proposed approach was trained, validated, and tested on 100 high-grade glioma (HGG) cases twice, once with all MRI sequences and second with only FLAIR and T1Gd. The results on the test set showed that the baseline U-Net model gave a mean Dice score of 0.9166 and 0.9190 on all MRI sequences using FLAIR and T1Gd, respectively. To check for possible performance improvement of the U-Net on FLAIR and T1Gd modalities, an ensemble of pre-trained VGG16, VGG19, and ResNet50 as modified U-Net encoders were employed for automated glioma segmentation based on T1Gd and FLAIR modalities only and compared with the baseline U-Net. The proposed models were trained, validated, and tested on 259 high-grade gliomas (HGG) cases. The results showed that the proposed baseline U-Net model and the ensemble of pre-trained VGG16, VGG19, or ResNet50 as modified U-Net encoders have a mean Dice score of 0.9395, 0.9360, 0.9359, and 0.9356, respectively. The results were also compared to other studies based on four MRI modalities. The work indicates that FLAIR and T1Gd are the most prominent contributors to the segmentation process. The proposed baseline U-Net is robust enough for segmenting HGG sub-tumoral structures and competitive with other state-of-the-art works.