Moderate Accuracy Research Articles

Optic Nerve Sheath Diameter on Computed Tomography Scans Reflects Elevated Intracranial Pressure in Patients with Craniosynostosis.

Assessment for elevated intracranial pressure (ICP) helps guide interventional decision-making to treat craniosynostosis. However, non-invasive techniques for measuring ICP are limited. This study assesses whether optic nerve sheath diameter (ONSD) on low-dose computed tomography (CT) scans is associated with ICP in patients with craniosynostosis. Pediatric patients treated between 2014 and 2023 with craniosynostosis, intraoperative ICP measurements by direct subdural catheterization, and spectral domain-optical coherent tomography (SD-OCT) data were included. ONSD was retrospectively assessed on pre-operative CT scans by a masked neuroradiologist and compared to measures and proxies of ICP. Among 132 patients included, median age was 6.9 years (IQR 4.7-9.5) and 41 (31.1%) had a syndromic diagnosis. Maximum ONSD (ONSDmax) was increased in patients with ICP≥15mmHg (6.1mm vs. 5.5mm, p<0.01) and ICP≥20mmHg (6.3mm vs. 5.6mm, p<0.01). Maximum (r=0.32, p<0.001), minimum (r=0.26, p=0.003), and average (r=0.29, p<0.001) ONSD correlated with direct ICP measurements. ONSD and SD-OCT measurements were also correlated (RNFLmax: r=0.21, p=0.04; RTmax: r=0.24, p=0.02). An ONSDmax threshold of 5.75mm demonstrated 65% sensitivity and 64% specificity for detecting ICP≥15mmHg on optimized receiver operating characteristic curve analysis. Multivariable logistic regression generated an algorithm incorporating ONSDmax and age to detect ICP≥20mmHg with 64% sensitivity and 80% specificity. Optic nerve sheath diameter measured in low-dose CT scans detects elevated intracranial pressure with moderate accuracy, and precision increases when patient age is taken into consideration. Given the ease of accessing CT scan data, this may be a helpful ICP proxy for clinical decision-making.

The four-item PRECISE-DAPT score identifies coronary artery bypass grafting patients with increased risk for post-discharge major bleeding.

Early identification of patients with increased bleeding risk increases the possibility to individualize antithrombotic treatment. We validated the PRECISE-DAPT score, originally developed to estimate bleeding risk in patients on dual antiplatelet therapy (DAPT) after percutaneous coronary intervention (PCI), in coronary artery bypass grafting (CABG) patients. All patients who underwent first time, isolated CABG in Sweden 2009-2020 and survived until discharge were included. The four-item PRECISE-DAPT score, based on age, estimated glomerular filtration rate, preoperative haemoglobin concentration, and previous spontaneous bleeding, was calculated in patients discharged on DAPT (n=6838), or antiplatelet monotherapy (n=15406). High bleeding risk was defined as a score≥25 in accordance with previous studies and major bleeding as hospitalization due to bleeding. Associations were assessed by C-statistics and Cox regression models.Major bleeding occurred during the first postoperative year in 130 patients (1.9%) in the DAPT group, and in 197 patients (1.3%) in the monotherapy group. The score identified 32.9% of the patients in the DAPT group and 38.2% in monotherapy groups as having high bleeding risk. The area under the ROC-curve for the score was 0.67 (95%CI 0.62-0.72) for DAPT and 0.71 (0.67-0.74) for monotherapy. The hazard ratio for high bleeding risk vs. very low risk was 4.14 (2.07-8.26) for DAPT patients, and 4.95 (2.61-9.39) for monotherapy patients, both p<0.001. The PRECISE-DAPT identifies patients with increased risk for major bleeding after discharge following CABG with moderate accuracy. The accuracy is comparable to what previously has been reported for patients after PCI.

External validation of and improvement upon a model for the prediction of placenta accreta spectrum severity using prospectively collected multicenter ultrasound data.

This study aimed to validate the Sargent risk stratification algorithm for the prediction of placenta accreta spectrum (PAS) severity using data collected from multiple centers and using the multicenter data to improve the model. We conducted a multicenter analysis using data collected for the IS-PAS database. The Sargent model's effectiveness in distinguishing between abnormally adherent placenta (FIGO grade 1) and abnormally invasive placenta (FIGO grades 2 and 3) was evaluated. A new model was developed using multicenter data from the IS-PAS database. The database included 315 cases of suspected PAS, of which 226 had fully documented standardized ultrasound signs. The final diagnosis was normal placentation in 5, abnormally adherent placenta/FIGO grade 1 in 43, and abnormally invasive placenta/FIGO grades 2 and 3 in 178. The external validation of the Sargent model revealed moderate predictive accuracy in a multicenter setting (C-index 0.68), compared to its higher accuracy in a single-center context (C-index 0.90). The newly developed model achieved a C-index of 0.74. The study underscores the difficulty in developing universally applicable PAS prediction models. While models like that of Sargent etal. show promise, their reproducibility varies across settings, likely due to the interpretation of the ultrasound signs. The findings support the need for updating the current ultrasound descriptors and for the development of any new predictive models to use data collected by different operators in multiple clinical settings.

Multifunctional Porous Pavement Prototype for Urban Pluvial Flood Protection: Preliminarily Findings on Contribution of Attitudes to Acceptance Willingness Toward Proposed Scientific and Engineering Solutions

Rapid urbanization and climate change correlate with an increase in frequency of flood events, globally. For many cities worldwide pluvial floods bring significant risk. Permeable/pervious paving (PePav), as an essential sustainable urban drainage technique, is one of the key environmental solutions for urban flooding. Waste and recycled materials in the construction industry for PePav production comply with the principles of circular economy and sustainable development. The influence of the human factor is well recognized in efforts to apply those solutions and make them self-sustainable. The preliminary results show that younger female students and students with lower family monthly income are prone to express more positive attitudes toward PePav generally. Students from different study groups (Departments of Psychology, Hydraulic and Environmental Engineering, and Construction Project Management) show significantly different attitudes toward PePav. Statistical models indicate that the PePav acceptance willingness can be predicted with moderate accuracy by knowing attitudes toward PePav and personal experience in construction. The contributions of the fact that someone's close person has been affected by a flood may vary depending on the type of PePav scientific and engineering solutions.

Breaking the mold of simulation-optimization: Direct forward machine learning methods for groundwater contaminant source identification

Groundwater Contaminant Source Identification (GCSI) is important for addressing environmental concerns. Currently, it is widely achieved using the Simulation/Optimization (S/O) method. However, the utilization of optimization techniques may cause high computation costs and parameter equifinality issues. We introduce two innovative GCSI methods, Direct Forward Machine Learning (DFML) and One-Hot Machine Learning (OHML), utilizing the classical Artificial Neuro Network (ANN) model in the machine learning field. Both new methods eliminate the need for an optimization algorithm in GCSI, thus reducing the construction effort and improving efficiency. The first method, DFML can directly estimate eight parameters, providing valuable insights into the contaminant location, release history, and aquifer properties. The second method, OHML can estimate the spatial probability distribution of the contaminant location through one-hot encoding, addressing uncertainties in the contaminant source location estimations realized by DFML.Evaluations demonstrate that both methods exhibit satisfying performances. DFML can estimate contaminant location and aquifer properties with high accuracy; and estimate the release history information with moderate accuracy. The OHML correctly assigns the higher contaminant probabilities to regions containing true contaminant locations. The combination of DFML and OHML offers a comprehensive framework. This study contributes valid methodologies to the GCSI field.

Evaluating accuracy and reproducibility of ChatGPT responses to patient-based questions in Ophthalmology: An observational study.

Chat Generative Pre-Trained Transformer (ChatGPT) is an online large language model that appears to be a popular source of health information, as it can provide patients with answers in the form of human-like text, although the accuracy and safety of its responses are not evident. This study aims to evaluate the accuracy and reproducibility of ChatGPT responses to patients-based questions in ophthalmology. We collected 150 questions from the "Ask an ophthalmologist" page of the American Academy of Ophthalmology, which were reviewed and refined by two ophthalmologists for their eligibility. Each question was inputted into ChatGPT twice using the "new chat" option. The grading scale included the following: (1) comprehensive, (2) correct but inadequate, (3) some correct and some incorrect, and (4) completely incorrect. Totally, 117 questions were inputted into ChatGPT, which provided "comprehensive" responses to 70/117 (59.8%) of questions. Concerning reproducibility, it was defined as no difference in grading categories (1 and 2 vs 3 and 4) between the 2 responses for each question. ChatGPT provided reproducible responses to 91.5% of questions. This study shows moderate accuracy and reproducibility of ChatGPT responses to patients' questions in ophthalmology. ChatGPT may be-after more modifications-a supplementary health information source, which should be used as an adjunct, but not a substitute, to medical advice. The reliability of ChatGPT should undergo more investigations.

Discharge C-reactive protein predicts 90-day readmission after pancreatoduodenectomy: a conditional inference tree analysis

BackgroundThe aim of this study was to assess the predictive value of discharge C-reactive protein (CRP) and white blood cell (WBC) levels for 90-day readmission after pancreatoduodenectomy (PD). MethodsA two-centre, retrospective study was performed between 2008 and 2022. Receiver operating characteristic (ROC) curve analysis was used to determine the predictive value of CRP level and WBC count at discharge. A conditional inference tree (CTREE) was constructed to identify combined risks within subgroups using variables associated with readmission. ResultsOf 438 patients, 54 (12%) were readmitted. The median WBC count at discharge was comparable between the readmitted and not readmitted groups (9.1 vs. 8.5 G/l). The CRP levels at discharge were predictive of 90-day readmission, with an area under the ROC curve (AUC) of 0.63 (95% CI: 0.55−0.63). A CRP concentration below 105 mg/l ruled out 90-day readmission, with a negative predictive value (NPV) of 90% (95% CI: 81%–95%). CTREE confirmed the diagnostic value of CRP at discharge (AUC = 0.68, 95% CI 0.60–0.68). CTREE additionally identified previous wound infection as a second risk factor for readmission in patients with CRP levels less than 101 mg/l (P = 0.003). ConclusionCRP levels below 105 mg/l at discharge allow for a safe discharge with a low 90-day readmission rate. Wound infection, but not WBC count, was a positive predictor of 90-day readmission with moderate accuracy, suggesting the need for predischarge imaging for undetected complications in this patient cohort. Trial registrationOur retrospective analysis did not require registration with a publicly accessible registry.

Automated Analysis of Tooth Anatomy and Pathological Conditions from Orthopantomogram using Deep Neural Networks

This research project aims to automate the identification, labeling, and counting of teeth, as well as the classification of abnormalities and detection of caries in dental X-rays, specifically orthopantomograms (OPGs). It involves several deep neural networks and learning algorithms. The first module uses semantic segmentation with a U-net model to create masks for tooth detection, which are then refined with the YOLOv3 detector, achieving 80% accuracy. Canonical correlation analysis (CCA) helps find tooth midpoints and count the total number of teeth. The second module classifies abnormalities and pathologies using transfer learning with the Inceptionv3 model, yielding moderate accuracy. Caries detection is performed with thresholding and segmentation. The third module detects three treated pathologies—root canal treatments, crowns, and implants—using Faster RCNN and Inceptionv3, showing fair accuracy. Overall, the automated approach demonstrates promising results for enhancing X-ray image interpretation and diagnosing oral diseases.

Right anterior insula ASL hypoperfusion as a diagnostic biomarker of prodromal and mild dementia with Lewy bodies: preliminary evidence using a Bayesian approach.

Identifying and validating a biomarker with high specificity in early-stage dementia with Lewy bodies (DLB) using a feasible method is crucial to enhance the current suboptimal diagnostic procedure. Previous research revealed abnormalities, including hypoperfusion in the right anterior insular cortex at group level, in prodromal DLB. Exploring hypoperfusion of the right anterior insula, at an individual-level and assessing its relevance as a potential imaging biomarker in early DLB, has, to our knowledge, not been investigated. Our preliminary study aims to assess the feasibility of the technique and to provide a methodological framework for further investigation. We assessed the feasibility and accuracy of the hypoperfusion of the right anterior insula per arterial spin labelling magnetic resonance imaging (ASL-MRI) as a diagnostic biomarker in early DLB and provided rough estimates of its sensitivity and specificity. Defining the region of interest based on previous research, we established the biomarker as the hypoperfusion of the right anterior insula. Discriminative and analytical performances were assessed in comparison to a control group of treatment-resistant depression patients. Bayesian diagnostic reasoning was employed to assess the biomarker diagnostic usability in early DLB in two scenarios: healthy elderly controls and mild cognitive impairment. Additionally, we updated probabilities by integrating data from the Mayo-clinic cognitive fluctuations scale and real-time quaking-induced conversion (RT-QuIC) α-synuclein data. Lastly, a whole-brain perfusion analysis of DLB patients was conducted to identify further brain regions with discriminative abilities. We successfully replicated the right anterior insular hypoperfusion (RAI-Hypo) in all DLB patients at the individual level. The overall sensitivity of the biomarker was 96%, and the specificity was 92%. Bayesian testing revealed the biomarker's highest performance in early-stage DLB with cognitive fluctuations, showcasing a diagnostic potential associated with a high precision and moderate accuracy. In a cognitively non-impaired population, the RAI-Hypo showed a limited usability and lacked in selectivity to qualify as a screening tool. The exploratory whole-brain analysis revealed perfect discriminative capacities in the bilateral anterior insulae and the left inferior parietal lobule. Further studies are needed to confirm our preliminary results. If performance is maintained in subsequent studies and is compared to a more suitable control population, the proposed biomarker may be eventually sufficient to discriminate early-stage DLB from non-DLB.

General process detection through side channel characterization

ABSTRACT In this study, we demonstrate the feasibility of using physical sensors in System-on-a-Chip (SoC) for real-time security monitoring by detecting and characterizing different process types, such as file I/O, CPU/ALU-intensive tasks, network I/O and virtualization. We present models that use sensor data for binary classification to determine whether a specific process category is active or inactive. Analyzing the detection results, we determine the importance of each sensor in identifying process types and providing insights into process behaviors and sensor impacts. We developed adaptive ensemble classifiers to accommodate varying load conditions, enhancing detection accuracy across diverse operational scenarios, including regular background activities that simulate real-world conditions. Our results show effective detection of file I/O and CPU/ALU-intensive processes under various loads. Virtualization processes are accurately detected under light loads but show a moderate accuracy decline under heavier conditions. Network I/O detection faces challenges due to fewer relevant sensors. Our ability to predict process categories consistently and with high performance allows us to discover behaviors of various malicious activities. This research underlines the efficacy of sensor-based analysis for reliable and adaptable real-time process monitoring, demonstrating the feasibility of using sensor data for security purposes in various environmental conditions.

Development of Uv-visible and Hplc Analytical Methods for Quantitative Analysis of Antiulcer Drugs in Dosage Forms

Precise measurement of active pharmaceutical ingredients (APIs) in dosage forms is crucial in the pharmaceutical business to guarantee the quality, effectiveness, and safety of the product. This research is focused on creating and confirming analytical techniques for accurately measuring the amount of antiulcer medicines in different types of medication. The measurement of antiulcer medicines in their respective formulations was carried out using two analytical techniques: UV-Visible spectroscopy and High-Performance Liquid Chromatography (HPLC). UV-Visible spectroscopy is a straightforward and cost-effective method that allows for quick examination. It is particularly useful for regular quality control applications. HPLC, in contrast, offers exceptional sensitivity, specificity, and precision, facilitating the isolation and measurement of many constituents in complex matrices. The technique development procedure included optimizing experimental parameters, including solvent systems, detection wavelengths, column types, mobile phase composition, and flow rates. The reliability and robustness of the suggested procedures were ensured by validation studies done in accordance with ICH recommendations. The adequacy of the procedures for their intended purpose was assessed by evaluating parameters such as linearity, precision, accuracy, specificity, and robustness. The UV-Visible and HPLC techniques that were developed showed exceptional linearity over the examined concentration range, with correlation values (R²) above 0.99. The precision experiments showed low relative standard deviations (RSDs), suggesting excellent repeatability and moderate accuracy. The accuracy checks demonstrated that the recovery values were within acceptable ranges, therefore verifying the dependability of the methodologies used for quantitative analysis.

Augmenting machine learning of Grad–Shafranov equilibrium reconstruction with Green's functions

This work presents a method for predicting plasma equilibria in tokamak fusion experiments and reactors. The approach involves representing the plasma current as a linear combination of basis functions using principal component analysis of plasma toroidal current densities (Jt) from the EFIT-AI equilibrium database. Then utilizing EFIT's Green's function tables, basis functions are created for the poloidal flux (ψ) and diagnostics generated from the toroidal current (Jt). Similar to the idea of a physics-informed neural network (NN), this physically enforces consistency between ψ, Jt, and the synthetic diagnostics. First, the predictive capability of a least squares technique to minimize the error on the synthetic diagnostics is employed. The results show that the method achieves high accuracy in predicting ψ and moderate accuracy in predicting Jt with median R2 = 0.9993 and R2 = 0.978, respectively. A comprehensive NN using a network architecture search is also employed to predict the coefficients of the basis functions. The NN demonstrates significantly better performance compared to the least squares method with median R2 = 0.9997 and 0.9916 for Jt and ψ, respectively. The robustness of the method is evaluated by handling missing or incorrect data through the least squares filling of missing data, which shows that the NN prediction remains strong even with a reduced number of diagnostics. Additionally, the method is tested on plasmas outside of the training range showing reasonable results.

Diagnostic Accuracy of Artificial Intelligence Compared to Biopsy in Detecting Early Oral Squamous Cell Carcinoma: A Systematic Review and Meta Analysis.

To summarize and compare the existing evidence on diagnostic accuracy of artificial intelligence (AI) models in detecting early oral squamous cell carcinoma (OSCC). Review was performed in accordance to Preferred Reporting Items for Systematic Reviews and Meta-Analysis - Diagnostic Test Accuracy (PRISMA- DTA) checklist and the review protocol is registered under PROSPERO(CRD42023456355). PubMed, Google Scholar, EBSCOhost were searched from January 2000 to November 2023 to identify the diagnostic potential of AI based tools and models. True-positive, false-positive, true-negative, false-negative, sensitivity, specificity values were extracted or calculated if not present for each study. Quality of selected studies was evaluated based on QUADAS (Quality assessment of diagnostic accuracy studies)- 2 tool. Meta-analysis was performed in Meta-Disc 1.4 software and Review Manager 5.3 RevMan using a bivariate model parameter for the sensitivity and specificity and summary points, summary receiver operating curve (SROC), diagnostic odds ratio (DOR) confidence region, and area under curve (AUC) were calculated. Fourteen studies were included for qualitative synthesis and for meta-analysis. Included studies had presence of low to moderate risk of bias. Pooled sensitivity and specificity of 0.43 (CI 0.18- 0.71) and 0.50 (CI 0.20- 0.80) was observed with a pooled positive likelihood ratio of (PLR) 0.86 (0.43 - 1.71) and negative likelihood ratio (NLR) of 1.04 (0.42 - 1.68) was observed with DOR of 0.78 (0.12 - 5.18) and overall accuracy (AUC) being 0.45 respectively. AI based tools has poor to moderate overall diagnostic accuracy. However, to validate our study findings further more standardized diagnostic accuracy studies should be conducted with proper reporting through QUADAS-2 tool. Thus, we can conclude AI based based tool for secondary level of prevention for early OSCC under early diagnosis and prompt treatment.

Association between lipid accumulation product index and chronic kidney disease: A systematic review and meta-analysis.

Diabetes mellitus and lipid metabolism disorders are increasingly recognized as key contributors to the development of chronic kidney disease (CKD). The lipid accumulation product (LAP) index, a novel marker of lipid accumulation, has potential implications for CKD risk assessment. The present meta-analysis aimed to assess the association between LAP index and CKD, with an emphasis on varying impacts in diabetic and non-diabetic populations. A comprehensive search for relevant publications was performed using PubMed/MEDLINE, Scopus, Cochrane Library, ScienceDirect and Google Scholar databases, and a meta-analysis of 17 studies was performed to investigate the relationship between LAP index and CKD. The random-effects inverse-variance model employing the DerSimonian-Laird estimator for τ² was utilized to calculate pooled odds ratios (ORs). Diagnostic accuracy was assessed using summary receiver operating characteristic (ROC) curves, with calculations of the area under the ROC curve (AUROC), sensitivity, specificity, likelihood ratios and diagnostic OR. The pooled OR for the association between higher quintiles or tertiles of LAP index and CKD was 1.098 (95% CI: 1.043-1.152), with substantial heterogeneity (I²=91.2%) and evidence of publication bias. Subgroup analysis revealed a stronger association in non-diabetic (OR=2.422, 95% CI: 1.802-3.042) compared with diabetic patients (OR=1.018, 95% CI: 0.993-1.043). The diagnostic accuracy of LAP index for CKD was moderate (AUROC=0.64), with sensitivity and specificity estimates of 0.58 and 0.63, respectively. In conclusion, in the present study, LAP index demonstrated a modest but significant association with CKD, particularly in non-diabetic patients. Despite its moderate diagnostic accuracy, the LAP index could serve as a valuable tool in CKD risk stratification, particularly when integrated with other clinical markers.

Impacts of heat stress on the accuracy of an ear-tag accelerometer for monitoring rumination and eating behavior in dairy-beef cross cattle using an automated gold standard

Accelerometer-based technologies can be utilized for precision monitoring of feeding behaviors, but limited information is available regarding the impact of varying environmental conditions on sensor performance. The objective of this study was to determine if a commercially available ear-tag sensor (CM; CowManager SensOor, Agis Automatisering BV) could accurately quantify eating and rumination time under heat stress conditions. Data obtained from CM sensors was compared with data collected using an automated gold standard (RW; Rumiwatch System; Itin+Hoch). Automated measurements were obtained from 2 experiments in which cattle were exposed to heat stress conditions. In the principal study (Experiment 1), 3428 h of data were collected from 9 Holstein × Angus steers (470.9 ± 23.9 kg) subjected to either thermoneutral (TN; 21.0°C; 64.0% humidity; temperature-humidity index [THI] = 67; 12- and 12-h light and dark cycle; n = 1714 h), or heat stress conditions (HS; cyclical daily temperatures to mimic diurnal patterns; 0800 - 2000 h: 33.6°C, 40.0% RH, THI: 83.5; 2000 - 0800 h: 23.2°C, 70.0% RH; THI: 70.3; n = 1714 h). Data (n = 719 h) from 6 Holstein x Angus steers (487.9 ± 9.1 kg) were obtained from a subsequent experiment (Experiment 2) to confirm consistency of ear-tag accelerometer performance under elevated THI (HS conditions as described above). In Experiment 1, CM was capable of quantifying rumination time with high accuracy under TN conditions (concordance correlation coefficient [CCC]: 0.75 - 0.81). Overall, agreement between CM and the automated gold standard declined 6 - 7% during HS, which was most apparent later in the day when cattle had been subjected to HS for multiple hours (moderate agreement; CCC: 0.68). Accuracy for rumination time was also only moderate for data collected during Experiment 2 (CCC: 0.55 - 0.61). In contrast, CM reported total eating (eating with the head down + head up while masticating) time with moderate accuracy for TN (CCC: 0.53 - 0.54), only achieved negligible to low accuracy during HS (CCC: 0.39 - 0.44 [Experiment 1] and 0.17 - 0.34 [Experiment 2]). Sensor performance did improve when CM eating time was compared specifically to the time spent with the head down reported by RW; HS still negatively influenced sensor performance, however, with high agreement during TN (CCC: 0.72 - 0.73) but low to moderate agreement during HS (CCC: 0.65 - 0.69 [Experiment 1] and 0.40 - 0.58 [Experiment 2]). Results of this study suggest accuracy of ear-tag accelerometers may be impaired when cattle are subjected to heat stress.

Machine Learning Models for Predicting Risky Pregnancies in Early Clinical Interventions

Background: Risky pregnancies present significant challenges in maternal healthcare, often requiring accurate prediction to prevent adverse outcomes. Machine learning (ML) models offer a promising approach for predicting such risks, enabling timely interventions. This study evaluates five machine learning models—Logistic Regression, Decision Tree, K-Nearest Neighbors (KNN), Naive Bayes, and Support Vector Machine (SVM)—for their effectiveness in predicting risky pregnancies using clinical datasets. Methods: The study developed and evaluated five ML models, each implemented using Python’s scikit-learn library. The dataset was split into 75% for training and 25% for testing. Standard classification metrics, including accuracy, precision, recall, and F1-score, were used to assess model performance. Hyperparameter tuning was conducted using grid search and cross-validation to optimize model parameters. The models’ performance was compared to identify the most suitable for clinical applications. Results: The Decision Tree model achieved the highest accuracy (100% on training data, 95.6% on testing data), along with excellent precision, recall, and F1-scores for both classes, making it the most accurate and interpretable model for predicting risky pregnancies. Logistic Regression also performed well, particularly in identifying high-risk cases, with testing accuracy of 82%. KNN and SVM provided moderate accuracy, with KNN achieving 78% testing accuracy and SVM 80%. Naive Bayes, however, performed poorly, achieving only 43.2% accuracy due to its assumption of feature independence, which was not suitable for the dataset. Conclusion: The Decision Tree and Logistic Regression models emerged as the most effective for predicting risky pregnancies, offering high accuracy and interpretability, crucial for clinical decision-making.

Peripheral biomarkers to differentiate bipolar depression from major depressive disorder: a real-world retrospective study

BackgroundBipolar depression (BPD) is often misdiagnosed as a major depressive disorder (MDD) in clinical practice, which may be attributed to a lack of robust biomarkers indicative of differentiated diagnosis. This study analysed the differences in various hormones and inflammatory markers to explore peripheral biomarkers that differentiate BPD from MDD patients.MethodsA total of 2,048 BPD and MDD patients were included. A panel of blood tests was performed to determine the levels of sex hormones, stress hormones, and immune-related indicators. Propensity score matching (PSM) was used to control for the effect of potential confounders between two groups and further a receiver operating characteristic (ROC) curve was used to analyse the potential biomarkers for differentiating BPD from MDD.ResultsCompared to patients with MDD, patients with BPD expressed a longer duration of illness, more hospitalisations within five years, and an earlier age of onset, along with fewer comorbid psychotic symptoms. In terms of biochemical parameters, MDD patients presented higher IgA and IgM levels, while BPD patients featured more elevated neutrophil and monocyte counts. ROC analysis suggested that combined biological indicators and clinical features could moderately distinguish between BPD and MDD. In addition, different biological features exist in BPD and MDD patients of different ages and sexes.ConclusionsDifferential peripheral biological parameters were observed between BPD and MDD, which may be age-sex specific, and a combined diagnostic model that integrates clinical characteristics and biochemical indicators has a moderate accuracy in distinguishing BPD from MDD.

A survey on machine learning system for intraductal papillary mucinous neoplasms detection

IPMN cysts, a pre-malignant risk to the pancreas, have the potential to develop into pancreatic cancer. Accurately identifying and evaluating the risk level is crucial for planning an efficient treatment strategy. However, this task is immensely challenging due to the varied and irregular shapes, textures, and sizes of IPMN cysts, as well as those of the pancreas itself. In this study, we introduce a new computer-aided diagnostic approach for classifying IPMN risk levels based on multi-contrast MRI scans. The proposed analysis framework comprises an efficient volumetric self-adapting segmentation strategy for delineating the pancreas, followed by a newly developed deep learning-based classification scheme incorporating a radiomics-based predictive approach. To evaluate the proposed decision-fusion model, we use multi-centre datasets and multi-contrast MRI scans, aiming to achieve superior performance compared to the current state of the art in this field. The ablation studies illustrate the importance of both radiomics and deep learning modules in achieving a new state-of-the-art (SOTA) performance compared to international guidelines and published studies (81.9% vs 61.3% in accuracy). These key findings carry significant implications for clinical decision-making, potentially revolutionizing the way IPMN risk levels are classified. Through a series of rigorous experiments on multi-centre datasets (involving more MRI scans from five centers), we attained unprecedented performance levels with moderate accuracy. The code will be made available upon publication.

Gender variability in machine learning based subcortical neuroimaging for Parkinson’s disease diagnosis

PurposeThis study evaluates machine learning (ML) classifiers for diagnosing Parkinson’s disease (PD) using subcortical brain region data from 3D T1 magnetic resonance imaging (MRI) Parkinson’s Progression Markers Initiative (PPMI database). We aim to identify top-performing algorithms and assess gender-related differences in accuracy.Design/methodology/approachMultiple ML algorithms will be compared for their ability to classify PD vs healthy controls using MRI scans of the brain structures like the putamen, thalamus, brainstem, accumbens, amygdala, caudate, hippocampus and pallidum. Analysis will include gender-specific performance comparisons.FindingsThe study reveals that ML classifier performance in diagnosing PD varies across subcortical brain regions and shows gender differences. The Extra Trees classifier performed best in men (86.36% accuracy in the putamen), while Naive Bayes performed best in women (69.23%, amygdala). Regions like the accumbens, hippocampus and caudate showed moderate accuracy (65–70%) in men and poor performance in women. The results point out a significant gender-based performance gap, highlighting the need for gender-specific models to improve diagnostic precision across complex brain structures.Originality/valueThis study highlights the significant impact of gender on machine learning diagnosis of PD using data from subcortical brain regions. Our novel focus on these regions uncovers their diagnostic potential, improves model accuracy and emphasizes the need for gender-specific approaches in medical AI. This work could ultimately lead to earlier PD detection and more personalized treatment.

Can ChatGPT Fool the Match? Artificial Intelligence Personal Statements for Plastic Surgery Residency Applications: A Comparative Study.

Introduction: Personal statements can be decisive in Canadian residency applications. With the rise in AI technology, ethical concerns regarding authenticity and originality become more pressing. This study explores the capability of ChatGPT in producing personal statements for plastic surgery residency that match the quality of statements written by successful applicants. Methods: ChatGPT was utilized to generate a cohort of personal statements for CaRMS (Canadian Residency Matching Service) to compare with previously successful Plastic Surgery applications. Each AI-generated and human-written statement was randomized and anonymized prior to assessment. Two retired members of the plastic surgery residency selection committee from the University of British Columbia, evaluated these on a 0 to 10 scale and provided a binary response judging whether each statement was AI or human written. Statistical analysis included Welch 2-sample t tests and Cohen's Kappa for agreement. Results: Twenty-two personal statements (11 AI-generated by ChatGPT and 11 human-written) were evaluated. The overall mean scores were 7.48 (SD 0.932) and 7.68 (SD 0.716), respectively, with no significant difference between AI and human groups (P = .4129). The average accuracy in distinguishing between human and AI letters was 65.9%. The Cohen's Kappa value was 0.374. Conclusions: ChatGPT can generate personal statements for plastic surgery residency applications with quality indistinguishable from human-written counterparts, as evidenced by the lack of significant scoring difference and moderate accuracy in discrimination by experienced surgeons. These findings highlight the evolving role of AI and the need for updated evaluative criteria or guidelines in the residency application process.