Tree Algorithm Research Articles

Study on optimization of trajectory of picking ARM robot in TOSHIBA DC650B machine system

The evolution of the modern manufacturing sector has ushered in a new era of intricate applications for industrial robots. The traditional manual programming of these robots demands a significant investment of time and energy. Hence, the realm of automatic path planning has emerged as a prominent avenue for the advancement of industrial robotics. This article introduces a novel approach aimed at crafting the most efficient trajectory for a picking arm robot tasked with lifting 1 kg molded products from a TOSHIBA DC650B horizontal pressure casting machine. Our proposed solution involves harnessing the capabilities of the Rapidly-exploring Random Tree (RRT) algorithm alongside its refined counterpart, RRT*, to chart the optimal path for the robot to traverse towards its target destination within the workspace while avoiding collisions. The findings of our study underscore a substantial 26 % reduction in path length when implementing the RRT* algorithm in comparison to the original RRT algorithm. Furthermore, the employment of the RRT* algorithm not only enhances the efficiency of the robot's trajectory but also contributes to an overall improvement in performance metrics, cost-effectiveness, and safety standards within industrial production environments. This innovative methodology exhibits a broad spectrum of applicability, offering a streamlined programming approach that not only saves valuable time but also bolsters automation efficiency across diverse industrial sectors. By leveraging the advanced capabilities of the RRT* algorithm, industrial processes can benefit from optimized path planning, leading to enhanced operational performance and a reduction in overhead costs. As a result, this methodological approach promises significant improvements in the realm of industrial automation while simultaneously paving the way for future advancements in robotic applications

Parallelization of the funnel tree algorithm for finding shortest paths on polyhedral surfaces

This paper presents an enhanced approach to computing shortest paths on triangulated polyhedral surfaces by parallelizing the Funnel Tree Algorithm (introduced by An et al. in The funnel tree algorithm for finding shortest paths on polyhedral surfaces, Optimization (2023)) and incorporating the Method of Orienting Curves. In particularly, we use the Method of Orienting Curves for finding the shortest path from the cusp of a funnel to its direct destination. As a result, the children of the funnel are also determined simultaneously. This combined approach leverages modern multi-core processors to achieve significant performance improvements. Experimental results demonstrate the effectiveness of this method on various polyhedral surfaces. The resulting implementation achieves significantly better speedups over corresponding sequential code given by An et al. when compared to their previous work.

Pelvic Kinematics during Gait Following Long-segment Spinal Fusion Due to Adult Spinal Deformity: An Analysis Using a Smartphone-based Inertial Measurement Unit

Gait changes could occur after thoracic to pelvic long-segment corrective fusion surgery, a common procedure for adult spinal deformity (ASD), potentially affecting the occurrence and progression of postoperative hip osteoarthritis. We aimed to clarify postoperative pelvic kinematics in patients with ASD by performing gait analysis using a system based on a smartphone-integrated inertial measurement unit (IMU). A total of 21 consecutive outpatients (73.6±4.6 years old, 2 men, 19 women) were enrolled. All had undergone long-segment fusion from the thoracic spine to the pelvis for ASD more than 1 year previously and could walk unassisted. A control group comprised 20 healthy volunteers. The IMU was fixed on the sacrum, and data were collected when subjects walked forward on a flat indoor floor. Acceleration in three axial directions and angular velocity around the three axes were recorded simultaneously during gait, and data were cut out for each gait cycle. Of 1043 features obtained, the top 20 features with the smallest p-value in a statistical comparison were selected. These features, plus gender and age, were classified using gradient boosting machine learning based on the decision tree algorithm. The classification accuracy and relative importance of the feature items were calculated. The accuracy rate for gait classification between groups was 96.7% and the F1-score was 0.968. The factor that contributed most to the classification of gait in both groups was "y-angular,_change_quantiles,_f_agg="var",_isabs=True,_qh=0.6,_ql=0.2," which means the variance of the change of the absolute value in the pelvic rotation angular velocity in the horizontal plane in the range of 20%-60% of the gait cycle. Its relative importance was 0.351, which was smaller in the group with fusion. Patients with ASD following long-segment fusion from the thoracic spine to the pelvis apparently have a gait style characterized by suppressed pelvic rotation in the horizontal plane.

Application of Decision Tree Algorithms for Predicting Trip Purposes in Makurdi, Nigeria

Decision tree models are versatile and interpretable machine learning algorithms widely used for both classification and regression tasks in transportation planning. This research focuses on analysing the suitability of decision tree algorithms in predicting trip purposes in Makurdi, Nigeria. The methodology involves formalizing household demographic and trip information datasets obtained through an extensive survey process. Modelling and prediction were conducted using Python programming language, and evaluation metrics such as R-squared and Mean Absolute Error (MAE) were used to assess the model’s performance. The results indicate that the model performed well, achieving accuracies of 84% and 68% and low MAE values of 0.188 and 0.314 on training and validation data, respectively. These findings suggest the model's reliability for future predictions. The study concludes that the decision tree-based model provides actionable insights for urban planners, transportation engineers, and policymakers to make informed decisions for improving transportation planning and management in Makurdi, Nigeria.

Multi-Indicator Heuristic Evaluation-Based Rapidly Exploring Random Tree Algorithm for Robot Path Planning in Complex Environments

Multi-Indicator Heuristic Evaluation-Based Rapidly Exploring Random Tree Algorithm for Robot Path Planning in Complex Environments

Multicenter Development and Prospective Validation of eCARTv5: A Gradient-Boosted Machine-Learning Early Warning Score.

Early detection of clinical deterioration using machine-learning early warning scores may improve outcomes. However, most implemented scores were developed using logistic regression, only underwent retrospective validation, and were not tested in important subgroups. The objective of our multicenter retrospective and prospective observational study was to develop and prospectively validate a gradient-boosted machine model (eCARTv5) for identifying clinical deterioration on the wards. All adult patients admitted to the inpatient medical-surgical wards at seven hospitals in three health systems for model development (2006-2022). All adult patients admitted to the inpatient medical-surgical wards and at 21 hospitals from three health systems for retrospective (2009-2023) and prospective (2023-2024) external validation. Predictor variables (demographics, vital signs, documentation, and laboratory values) were used in a gradient-boosted trees algorithm to predict ICU transfer or death in the next 24 hours. The developed model (eCARTv5) was compared with the Modified Early Warning Score (MEWS), the National Early Warning Score (NEWS), and eCARTv2 using the area under the receiver operating characteristic curve (AUROC). The development cohort included 901,491 admissions, the retrospective validation cohort included 1,769,461 admissions, and the prospective validation cohort included 205,946 admissions. In retrospective validation, eCARTv5 had the highest AUROC (0.834; 95% CI, 0.834-0.835), followed by eCARTv2 (0.775 [95% CI, 0.775-0.776]), NEWS (0.766 [95% CI, 0.766-0.767]), and MEWS (0.704 [95% CI, 0.703-0.704]). eCARTv5's performance remained high (AUROC ≥0.80) across a range of patient demographics, clinical conditions, and during prospective validation. We developed eCARTv5, which performed better than eCARTv2, NEWS, and MEWS retrospectively, prospectively, and across a range of subgroups. These results served as the foundation for Food and Drug Administration clearance for its use in identifying deterioration in hospitalized ward patients.

Massively parallel computation in a heterogeneous regime

Abstract Massively-parallel graph algorithms have received extensive attention over the past decade, with research focusing on three memory regimes: the superlinear regime, the near-linear regime, and the sublinear regime. The sublinear regime is the most desirable in practice, but conditional hardness results point towards its limitations. In this work we study a heterogeneous model, where the memory of the machines varies in size. We focus mostly on the heterogeneous setting created by adding a single near-linear machine to the sublinear MPC regime, and show that even a single large machine suffices to circumvent most of the conditional hardness results for the sublinear regime: for graphs with n vertices and m edges, we give (a) an MST algorithm that runs in $$O(\log \log (m/n))$$ O ( log log ( m / n ) ) rounds; (b) an algorithm that constructs an O(k)-spanner of size $$O(n^{1+1/k})$$ O ( n 1 + 1 / k ) in O(1) rounds; and (c) a maximal-matching algorithm that runs in $$O(\sqrt{\log (m/n)}\log \log (m/n))$$ O ( log ( m / n ) log log ( m / n ) ) rounds. We also observe that the best known near-linear MPC algorithms for several other graph problems which are conjectured to be hard in the sublinear regime (minimum cut, maximal independent set, and vertex coloring) can easily be transformed to work in the heterogeneous MPC model with a single near-linear machine, while retaining their original round complexity in the near-linear regime. If the large machine is allowed to have superlinear memory, all of the problems above can be solved in O(1) rounds.

Electronic Detection of Garlic Density in Various Kinds of Yogurts Using Statistical Features

Accurate detection of food components plays a critical role in developing modern culinary technologies and food safety practices. This study uses electronic nose technology to determine garlic concentration in garlic yogurts. An electronic nose system consisting of 11 different MQ brand gas sensors was used in the study. Five different yogurt types were prepared with three different garlic concentrations: plain, low, and high. A total of 225 odor records were taken from 15 yogurt samples, and various features were extracted from these data, which were analyzed using four different classification algorithms. The Extra Trees algorithm was the most successful method, with 89.14% classification accuracy, 89.80% sensitivity, and 94.57% specificity rates. The results of the study show that electronic nose technology can be used in many application areas, especially in smart kitchen devices analyzing food ingredients to provide information about freshness and composition, in the food industry to ensure standardization of product quality in production processes and to ensure that intense aromatic ingredients such as garlic are used in the right amount, and in the development of food products suitable for consumers’ special diets or personal tastes.

Design and Implementation of Stock Market Prediction System for Used Cars in Nigeria

Stock market prediction of commodities have undergone changes from the traditional to modern methods of using machine learning. Hence, the objective of this study was to design and implement a stock market price prediction system for used cars in Nigeria using machine learning techniques, the extra tree algorithm and support vector machine (SVM). The dataset used included such attributes as fuel type, number of doors, number of cylinders, drive wheel and price amongst others. The model was designed by training and testing using pre-processed data. Python programming language was used in the implementation. The results obtained for the mean square error and the R-squared showed high accuracy and therefore made the model ideal for car price prediction in the automobile Nigerian market.

Advancing preeclampsia prediction: a tailored machine learning pipeline integrating resampling and ensemble models for handling imbalanced medical data

BackgroundConstructing a predictive model is challenging in imbalanced medical dataset (such as preeclampsia), particularly when employing ensemble machine learning algorithms.ObjectiveThis study aims to develop a robust pipeline that enhances the predictive performance of ensemble machine learning models for the early prediction of preeclampsia in an imbalanced dataset.MethodsOur research establishes a comprehensive pipeline optimized for early preeclampsia prediction in imbalanced medical datasets. We gathered electronic health records from pregnant women at the People’s Hospital of Guangxi from 2015 to 2020, with additional external validation using three public datasets. This extensive data collection facilitated the systematic assessment of various resampling techniques, varied minority-to-majority ratios, and ensemble machine learning algorithms through a structured evaluation process. We analyzed 4,608 combinations of model settings against performance metrics such as G-mean, MCC, AP, and AUC to determine the most effective configurations. Advanced statistical analyses including OLS regression, ANOVA, and Kruskal-Wallis tests were utilized to fine-tune these settings, enhancing model performance and robustness for clinical application.ResultsOur analysis confirmed the significant impact of systematic sequential optimization of variables on the predictive performance of our models. The most effective configuration utilized the Inverse Weighted Gaussian Mixture Model for resampling, combined with Gradient Boosting Decision Trees algorithm, and an optimized minority-to-majority ratio of 0.09, achieving a Geometric Mean of 0.6694 (95% confidence interval: 0.5855–0.7557). This configuration significantly outperformed the baseline across all evaluated metrics, demonstrating substantial improvements in model performance.ConclusionsThis study establishes a robust pipeline that significantly enhances the predictive performance of models for preeclampsia within imbalanced datasets. Our findings underscore the importance of a strategic approach to variable optimization in medical diagnostics, offering potential for broad application in various medical contexts where class imbalance is a concern.

Application of decision tree model in diagnosis of mycoplasma pneumoniae pneumonia with plastic bronchitis

BackgroundTo establish a decision tree model of Mycoplasma pneumoniae pneumonia(MPP) complicated with plastic bronchitis(PB) in children, and to explore the application value of decision tree model in the auxiliary diagnosis of children.MethodsA retrospective study was conducted to collect the clinical data of 214 children who met the admission criteria in Fujian Children’s Hospital from June 2022 to June 2024, and they were divided into plastic bronchitis group (n = 66) and non-plastic bronchitis group (n = 148). Using R language, 70% of the data from each group of patients was randomly selected for training the model using decision tree algorithm analysis, thus generating a clinical diagnostic decision tree for Mycoplasma pneumoniae (MP) combined with PB. The generated decision tree model was validated on the validation sample dataset and the detection effect value of the model was calculated.ResultIn this study, a total of 22 indicators were employed to build the decision tree diagnostic model. Univariate statistical analysis was carried out prior to the model construction, and it was discovered that the differences of 13 indicators between the molded group and the non-molded group were statistically significant. A decision tree model with D-dimer ≥ 1.7ug/mL, C-reactive protein ≥ 15 mg/L, drug resistance or not, and serum ferritin<137 mg/L was constructed in the training sample dataset of the molded group and the non-molded group. The sensitivity of the decision tree model was 0.884, which was verified in the dataset of the remolded group and the non-molded group. The specificity was 0.727, and the area under the receiver operating characteristic curve was 0.831.ConclusionDecision tree model can provide reference for the application of auxiliary diagnosis in children with mycoplasma pneumoniae pneumonia complicated with plastic bronchitis. The model has good discriminative ability in general, and is worthy of clinical application and further study.

Distributed Denial of Service Attack Detection in Software-Defined Networks Using Decision Tree Algorithms

A software-defined network (SDN) is a new architecture approach for constructing and maintaining networks with the main goal of making the network open and programmable. This allows the achievement of specific network behavior by updating and installing software, instead of making physical changes to the network. Thus, SDNs allow far more flexibility and maintainability compared to conventional device-dependent architectures. Unfortunately, like their predecessors, SDNs are prone to distributed denial of service (DDoS) attacks. These attack paralyze networks by flooding the controller with bogus requests. The answer to this problem is to ignore machines in the network sending these requests. This can be achieved by incorporating classification algorithms that can distinguish between genuine and bogus requests. There is abundant literature on the application of such algorithms on conventional networks. However, because SDNs are relatively new, they lack such abundance both in terms of novel algorithms and effective datasets when it comes to DDoS attack detection. To address these issues, the present study analyzes several variants of the decision tree algorithm for detection of DDoS attacks while using two recently proposed datasets for SDNs. The study finds that a decision tree constructed with a hill climbing approach, termed the greedy decision tree, iteratively adds features on the basis of model performance and provides a simpler and more effective strategy for the detection of DDoS attacks in SDNs when compared with recently proposed schemes in the literature. Furthermore, stability analysis of the greedy decision tree provides useful insights about the performance of the algorithm. One edge that greedy decision tree has over several other methods is its enhanced interpretability in conjunction with higher accuracy.

Machine learning allows robust classification of lung neoplasm tissue using an electronic biopsy through minimally-invasive electrical impedance spectroscopy

New bronchoscopy techniques like radial probe endobronchial ultrasound have been developed for real-time sampling characterization, but their use is still limited. This study aims to use classification algorithms with minimally invasive electrical impedance spectroscopy to improve neoplastic lung tissue identification during biopsies. Decision Tree, Support Vector Machines (SVM), Ensemble Method, K-Nearest Neighbors, Naïve Bayes and Discriminant Analysis were applied using mean averaged bioimpedance modulus and phase angle spectra from lung tissue across 15 frequencies (15–307 kHz). Mann-Whitney U test assessed statistical significance between neoplasm and other tissues. Grid search analysis was conducted to determine the optimal hyperparameter configuration for each model, employing a 5-fold cross-validation approach. Model performance was evaluated using Receiver Operating Characteristic curves, with the Area Under Curve (AUC), precision, recall, and F1-score calculated. All the frequencies used to train and test the algorithms obtained high significant differences between neoplasm and the other types of tissues (P < 0.001). All the algorithms implemented obtained an accuracy, AUC and F1-score above the 95% except for Naïve Bayes. Decision Tree, Discriminant Analysis and SVM algorithms are suitable for the implementation of a new low-cost guidance method during bronchoscopy.

Classification of auditory ERPs for ADHD detection in children

Attention deficit hyperactivity disorder (ADHD) is one of the children’s most common neurodevelopmental conditions. ADHD diagnosis is based on evaluating inattention, hyperactivity, and impulsivity symptoms that interfere with or reduce daily functioning. Although electroencephalography (EEG) tests are used for ADHD diagnosis, they are generally considered a complement to clinical evaluation. This paper proposes an approach to classify EEG records of children with ADHD and control cases. We identified and extracted relevant features from EEG signals of 47 children (22 diagnosed with ADHD and 25 controls) and evaluated machine learning techniques for classification. We used the 2-tone oddball paradigm to elicit the subjects’ auditory event-related potentials (ERP), and we recorded EEG signals with a portable headset for approximately five minutes. In the feature extraction stage, we included measures from cognitive evoked potentials, frequency bands power, chaos quantification, and bispectral analysis, in addition to the age of the children and the number of high-pitched tones the children counted during the test. The SVM and Trees algorithms obtained the best performance for 86.36% accuracy and 95.45% sensitivity. These findings demonstrate the potential of portable EEG-based systems to complement standard clinical assessments, offering an objective, time-efficient, and accessible approach to support early ADHD diagnosis. Achieving high accuracy and sensitivity in classification is critical to reducing the risk of misdiagnosis and ensuring timely intervention, ultimately improving patient outcomes.

Leveraging mixed-effects regression trees for the analysis of high-dimensional longitudinal data to identify the low and high-risk subgroups: simulation study with application to genetic study

BackgroundThe linear mixed-effects model (LME) is a conventional parametric method mainly used for analyzing longitudinal and clustered data in genetic studies. Previous studies have shown that this model can be sensitive to parametric assumptions and provides less predictive performance than non-parametric methods such as random effects-expectation maximization (RE-EM) and unbiased RE-EM regression tree algorithms. These longitudinal regression trees utilize classification and regression trees (CART) and conditional inference trees (Ctree) to estimate the fixed-effects components of the mixed-effects model. While CART is a well-known tree algorithm, it suffers from greediness. To mitigate this issue, we used the Evtree algorithm to estimate the fixed-effects part of the LME for handling longitudinal and clustered data in genome association studies.MethodsIn this study, we propose a new non-parametric longitudinal-based algorithm called “Ev-RE-EM” for modeling a continuous response variable using the Evtree algorithm to estimate the fixed-effects part of the LME. We compared its predictive performance with other tree algorithms, such as RE-EM and unbiased RE-EM, with and without considering the structure for autocorrelation between errors within subjects to analyze the longitudinal data in the genetic study. The autocorrelation structures include a first-order autoregressive process, a compound symmetric structure with a constant correlation, and a general correlation matrix. The real data was obtained from the longitudinal Tehran cardiometabolic genetic study (TCGS). The data modeling used body mass index (BMI) as the phenotype and included predictor variables such as age, sex, and 25,640 single nucleotide polymorphisms (SNPs).ResultsThe results demonstrated that the predictive performance of Ev-RE-EM and unbiased RE-EM was nearly similar. Additionally, the Ev-RE-EM algorithm generated smaller trees than the unbiased RE-EM algorithm, enhancing tree interpretability.ConclusionThe results showed that the unbiased RE-EM and Ev-RE-EM algorithms outperformed the RE-EM algorithm. Since algorithm performance varies across datasets, researchers should test different algorithms on the dataset of interest and select the best-performing one. Accurately predicting and diagnosing an individual’s genetic profile is crucial in medical studies. The model with the highest accuracy should be used to enhance understanding of the genetics of complex traits, improve disease prevention and diagnosis, and aid in treating complex human diseases.

Development and validation comparison of multiple models for perioperative neurocognitive disorders during hip arthroplasty

This study aims to develop optimal predictive models for perioperative neurocognitive disorders (PND) in hip arthroplasty patients, thereby advancing clinical practice. Data from all hip arthroplasty patients in the MIMIC-IV database were utilized to predict PND. With 62 variables, we applied multiple logistic regression, artificial neural network (ANN), Naive Bayes, support vector machine, and decision tree (XgBoost) algorithms to forecast PND. Feature analysis, receiver operating characteristic curve (ROC) and calibration curve plotting, and sensitivity, specificity, and F-measure β = 1 (F1-score) assessments were conducted on both training and validation sets for classifying models’ effectiveness. Brier score and Index of prediction accuracy (IPA) were employed to compare prediction capabilities in both sets. Among 3,292 hip arthroplasty patients in the MIMIC database, 331 developed PND. Five models using different algorithms were constructed. After thorough comparison and validation, the ANN model emerged as the most effective model. Performance metrics on the training set for the ANN model were: ROC: 0.954, Accuracy: 0.938, Precision: 0.758, F1-score: 0.657, Brier Score: 0.048, IPA: 90.8%. On the validation set, the ANN model performed as follows: ROC: 0.857, Accuracy: 0.903, Precision: 0.539, F1-score: 0.432, Brier Score: 0.071, IPA: 71.4%. An online visualization tool was developed (https://xyyy.pythonanywhere.com/).

Path Planning of Intelligent Mobile Robots with an Improved RRT Algorithm

The Rapidly Exploring Random Tree algorithm, renowned for its randomness, asymptotic properties, and local planning capabilities, is extensively employed in autonomous driving for path planning. Addressing issues such as pronounced randomness, low search efficiency, inefficient utilization of effective points, suboptimal path smoothness, and potential deviations from the optimal path in the RRT algorithm based on random sampling, we proposed an optimization algorithm that integrates Kalman filtering to eliminate redundant points along the path. Initially, this algorithm addresses the issue of inverse growth in the RRT algorithm’s search tree by implementing a variable steering angle strategy, thereby minimizing oscillations and unnecessary pose adjustments. Secondly, by merging collision detection with Kalman filtering, and by comparing the step sizes between newly generated child nodes and random tree nodes towards the root node, we filtered redundant points from the path, thereby reducing the count of effective points and optimizing the path. Lastly, we utilized a second-order Bezier curve to smoothen the path, eliminating sharp corners and discontinuities, ultimately yielding the optimal path. Across diverse map environments and two distinct dimensional scenarios, we conducted multiple sets of simulation experiments to validate the algorithm’s feasibility. The experimental outcomes demonstrate notable improvements in parameters like average path length, average planning time, average count of effective points, and average sampling points, highlighting the enhanced accuracy and efficiency of the improved algorithm in path planning.

Prediction of prostate biopsy outcomes at different cut-offs of prostate-specific antigen using machine learning: a multicenter study

BackgroundMachine learning (ML) is a significant area of artificial intelligence, which can improve the accuracy of predictive or diagnostic models for differentiating between prostate biopsy outcomes. This study aims to develop a novel decision-support ML model for classifying patients with biopsy-negative (cancer-free), clinically significant, and non-clinically significant prostate cancer across two prostate-specific antigen (PSA) cut-offs ≤ 10 ng/ml and > 10 ng/ml.MethodsThe data for the current study were retrieved from the records of two main hospitals in Riyadh, Saudi Arabia from July 2018 through July 2024. Six machine learning algorithms were employed, and the dataset was randomly divided into a training set and a validation set at a ratio of 8:2. The following metrics were used as performance indicators across the six algorithms: Accuracy, Precision, Recall, F1-score, and area under the curve. Recent data from the two hospitals was utilized for external validation.ResultsThe metrics for Random Forest, Extra Tree, and Decision Tree algorithms showed excellent capability in classifying the outcomes of prostate biopsy for the two PSA cut-offs. However, the metrics for the PSA cut-off > 10 ng/ml are higher than those for PSA ≤ 10 ng/ml. For the three-class classification, the accuracy and area under the curve for the cut-off > 10 ng/ml were 0.96 and 0.99, respectively. While for the cut-off ≤ 10 ng/ml they were 0.92 and 0.94 for Random Forest and 0.94 and 0.95 for the Extra Tree algorithm. The metrics of non-clinically significant and biopsy-negative cases outperformed those of clinically significant cases.ConclusionML models are proving to be effective tools in differentiating between prostate biopsy outcomes, enhancing diagnostic accuracy, and potentially transforming clinical practices in prostate cancer management.

A Stepwise decision tree model for differential diagnosis of Kimura’s disease in the head and neck

ObjectivesThis study aims to differentiate Kimura’s disease (KD) from Sjogren’s syndrome with mucosa-associated lymphoid tissue lymphoma (SS&MALT), neurofibromatosis (NF), and lymphoma in the head and neck by using a stepwise decision tree approach.Materials and methodsA retrospective analysis of 202 patients with pathologically confirmed KD, SS&MALT, NF, or lymphoma was conducted. Demographic and magnetic resonance imaging (MRI) data were collected, with qualitative features (e.g., skin thickening, lesion morphology, lymphadenopathy, MRI signal intensity) and quantitative variables (e.g., age, lesion size, apparent diffusion coefficients (ADCs), wash-in rate, time to peak (TTP), time-signal intensity curve (TIC) patterns) examined. A stepwise decision-tree model using the classification and regression trees (CART) algorithm was developed to aid in the differential diagnosis of KD in the head and neck. The model’s diagnostic accuracy and misclassification risk were assessed to evaluate its reliability and effectiveness.ResultsKey characteristics for KD included male predominance (91.7%), frequent lymphadenopathy (86.1%), and skin thickening (72.2%). Primary lesions of NF typically exhibited higher ADCs compared to those of KD, SS&MALT, and lymphoma. In lymphadenopathy, however, unique ADC patterns were observed: in KD, the ADCs of lymphadenopathy were lower than those of primary lesions, whereas in lymphoma, the ADCs of lymphadenopathy were comparable to those of primary lesions. Predictors for distinguishing KD included lesion’s location, ADCs, lymphadenopathy, and sizes (all p < 0.001). The decision-tree model achieved an impressive 99.0% accuracy in the differential diagnosis across the overall cohort, with a 10-fold cross-validated misclassification risk of 0.079 ± 0.024.ConclusionThe stepwise decision tree model, based on MRI features, showed high accuracy in differentiating KD from other head and neck diseases, offering a reliable diagnostic tool in clinical practice.Clinical relevanceKD is characterized by male predominance, skin thickening, and high incidence of lymphadenopathy. ADCs and TIC patterns are distinguishable in differentiating KD from SS&MALT, NF, and lymphoma in the head and neck. The decision tree model enhances the understanding of KD imaging features and facilitates accurate KD diagnosis, offering an easily accessible and convenient diagnostic tool for radiologists and physicians in daily practice and guiding tailored clinical management plans for affected patients.Clinical trial numberNot applicable.

State-of-the-art for automated machine learning predicts outcomes in poor-grade aneurysmal subarachnoid hemorrhage using routinely measured laboratory & radiological parameters: coagulation parameters and liver function as key prognosticators

The objective of this study was to develop and evaluate automated machine learning (aML) models for predicting short-term (1-month) and medium-term (3-month) functional outcomes [Modified Rankin Scale (mRS)] in patients suffering from poor-grade aneurysmal subarachnoid hemorrhage (aSAH), using readily available and routinely measured laboratory and radiological parameters at admission. Data from a pilot non-randomized trial of 60 poor-grade aSAH patients (Hunt-Hess grades IV or V) were analyzed. Patients were evenly divided between targeted temperature management (TTM) and standard treatment groups. The current state-of-the-art for aML was adopted to employ nine ML algorithms with hyperparameter tuning to develop algorithmic models predicting 1 month and 3-months mRS scores. Model performance was evaluated using macro-weighted average Area Under the Receiver Operating Curve (mWA-AUROC) analysis and additional metrics. Logistic regression algorithmic models achieved perfect prediction (mWA-AUROC = 1, accuracy = 100%, sensitivity and specificity = 100% [95% CI: 83.16 − 100%]) for both 1-month and 3-month mRS outcomes. For 1-month outcomes, neutrophil count, platelet count, and gamma-glutamyl transferase levels were identified as key predictors. For 3-month outcomes, patient gender, activated partial thromboplastin time, and serum aspartate aminotransferase levels were most impactful. Decision tree algorithms (mWA-AUROC = 0.9-0.925) identified specific cut-points for various parameters, providing actionable information for clinical decision-making. Positive prognostic factors included alkaline phosphatase levels higher than mid-value of their normal range, absence of hydrocephalus, use of targeted temperature management (TTM), and specific cut-offs for coagulation and liver function parameters. The use of TTM was reinforced as a key prognosticator of mRS outcomes at both time points. We have made our developed models and the associated architecture available at GitHub. This study demonstrated the potential of aML in predicting functional outcomes for poor-grade aSAH patients. The identification of novel predictors, including liver function and coagulation parameters, opens new avenues for research and intervention. While the perfect predictive performance warrants cautious interpretation and further validation, these models represent a step towards personalized medicine in aSAH management, potentially improving prognostication and treatment strategies.