Classification And Regression Tree Model Research Articles

Magnetic resonance imaging-based prediction models for differentiating intraspinal schwannomas from meningiomas: classification and regression tree and random forest analysis.

Due to the variations in surgical approaches and prognosis between intraspinal schwannomas and meningiomas, it is crucial to accurately differentiate between the two prior to surgery. Currently, there is limited research exploring the implementation of machine learning (ML) methods for distinguishing between these two types of tumors. This study aimed to establish a classification and regression tree (CART) model and a random forest (RF) model for distinguishing schwannomas from meningiomas. We retrospectively collected 88 schwannomas (52 males and 36 females) and 51 meningiomas (10 males and 41 females) who underwent magnetic resonance imaging (MRI) examinations prior to the surgery. Simple clinical data and MRI imaging features, including age, sex, tumor location and size, T1-weighted images (T1WI) and T2-weighted images (T2WI) signal characteristics, degree and pattern of enhancement, dural tail sign, ginkgo leaf sign, and intervertebral foramen widening (IFW), were reviewed. Finally, a CART model and RF model were established based on the aforementioned features to evaluate their effectiveness in differentiating between the two types of tumors. Meanwhile, we also compared the performance of the ML models to the radiologists. The receiver operating characteristic (ROC) curve, accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were used to evaluate the models and clinicians' discrimination performance. Our investigation reveals significant variations in ten out of 11 variables in the training group and five out of 11 variables in the test group when comparing schwannomas and meningiomas (P<0.05). Ultimately, the CART model incorporated five variables: enhancement pattern, the presence of IFW, tumor location, maximum diameter, and T2WI signal intensity (SI). The RF model combined all 11 variables. The CART model, RF model, radiologist 1, and radiologist 2 achieved an area under the curve (AUC) of 0.890, 0.956, 0.681, and 0.723 in the training group, and 0.838, 0.922, 0.580, and 0.659 in the test group, respectively. The RF prediction model exhibits more exceptional performance than an experienced radiologist in discriminating intraspinal schwannomas from meningiomas. The RF model seems to be better in discriminating the two tumors than the CART model.

Development of clinical screening tool for exocrine pancreatic insufficiency in patients with definite chronic pancreatitis

Background/ObjectivesNo simple, accurate diagnostic tests exist for exocrine pancreatic insufficiency (EPI), and EPI remains underdiagnosed in chronic pancreatitis (CP). We sought to develop a digital screening tool to assist clinicians to predict EPI in patients with definite CP. MethodsThis was a retrospective case-control study of patients with definite CP with/without EPI. Overall, 49 candidate predictor variables were utilized to train a Classification and Regression Tree (CART) model to rank all predictors and select a parsimonious set of predictors for EPI status. Five-fold cross-validation was used to assess generalizability, and the full CART model was compared with 4 additional predictive models. EPI misclassification rate (mRate) served as primary endpoint metric. Results274 patients with definite CP from 6 pancreatitis centers across the United States were included, of which 58 % had EPI based on predetermined criteria. The optimal CART decision tree included 10 variables. The mRate without/with 5-fold cross-validation of the CART was 0.153 (training error) and 0.314 (prediction error), and the area under the receiver operating characteristic curve was 0.889 and 0.682, respectively. Sensitivity and specificity without/with 5-fold cross-validation was 0.888/0.789 and 0.794/0.535, respectively. A trained second CART without pancreas imaging variables (n = 6), yielded 8 variables. Training error/prediction error was 0.190/0.351; sensitivity was 0.869/0.650, and specificity was 0.728/0.649, each without/with 5-fold cross-validation. ConclusionWe developed two CART models that were integrated into one digital screening tool to assess for EPI in patients with definite CP and with two to six input variables needed for predicting EPI status.

Quantifying Seagrass Density Using Sentinel-2 Data and Machine Learning

Seagrasses, rooted aquatic plants growing completely underwater, are extremely important for the coastal ecosystem. They are an important component of the total carbon burial in the ocean, they provide food, shelter, and nursery to many aquatic organisms in coastal ecosystems, and they improve water quality. Due to human activity, seagrass coverage has been rapidly declining, and there is an urgent need to monitor seagrasses consistently. Seagrass coverage has been closely monitored in the Chesapeake Bay since 1970 using air photos and ground samples. These efforts are costly and time-consuming. Many studies have used remote sensing data to identify seagrass bed outlines, but few have mapped seagrass bed density. This study used Sentinel-2 satellite data and machine learning in Google Earth Engine and the Chesapeake Bay Program field data to map seagrass density. We used seagrass density data from the Chincoteague and Sinepuxent Bay to train machine learning algorithms and evaluate their accuracies. Out of the four machine learning models tested (Naive Bayes (NB), Classification and Regression Trees (CART), Support Vector Machine (SVM), and Random Forest (RF)), the RF model outperformed the other three models with overall accuracies of 0.874 and Kappa coefficients of 0.777. The SVM and CART models performed similarly and NB performed the poorest. We tested two different approaches to assess the models’ accuracy. When we used all the available ground samples to train the models, whereby our analysis showed that model performance was associated with seagrass density class, and that higher seagrass density classes had better consumer accuracy, producer accuracy, and F1 scores. However, the association of model performance with seagrass density class disappeared when using the same training data size for each class. Very sparse and dense seagrass classes had replacedhigherbetter accuracies than the sparse and moderate seagrass density classes. This finding suggests that training data impacts machine learning model performance. The uneven training data size for different classes can result in biased assessment results. Selecting proper training data and machine learning models are equally important when using machine learning and remote sensing data to map seagrass density. In summary, this study demonstrates the potential to map seagrass density using satellite data.

Reserve-building as a buffer for depression among individuals living with disability: a longitudinal study of current activities related to brain health.

This study examined whether reserve-building activities are associated with attenuated reported depression among people who were disabled from work due to a medical condition as compared to employed, retired, and unemployed participants. This secondary analysis included 771 individuals who provided data at three time points: baseline (late Spring 2020), follow-up 1 (Spring 2021), and follow-up 2 (Fall 2021). The DeltaQuest Reserve-Building Measure assessed current activities related to brain health. An analysis of variance and Pearson correlation coefficients assessed group differences in reserve-building activity scores. Classification and regression tree (CART) modeling investigated factors associated with higher and lower reported depression by employment group. The random effects (RE) models tested two buffering hypotheses: (1) comparing all groups to the employed group and (2) examining within-group effects. Engaging in outdoor activities, exercise, and religious/spiritual activities was associated with reduced depression over time in the overall sample. While disabled participants endorsed lower levels of being Active in the World, Outdoor activities, and Exercise and higher levels of Inner Life and Passive Media Consumption than the other employment groups, more reserve-building activities distinguished depression levels in the disabled group's CART models compared to the others. Among the disabled, unemployed, and retired participants, engaging in any reserve-building activities was also associated with lower depression scores, which was distinct from the employed participants. In the RE models that used the employed group as the reference category, only the disabled group's level of depression was buffered by engaging in creative activities. In the within-group RE models, the disabled group's engagement in Religious/Spiritual, Outdoors, and Games was associated with substantially reduced within-group depression, which was different from the other employment groups. In contrast, reserve-building activities were not implicated at all as buffers for employed participants. This study revealed a beneficial effect of reserve-building activities on buffering depression over time during the COVID-19 pandemic, particularly for disabled people. It documented that even if such individuals engaged in lesser amounts of such activities as compared to other employment groups, the buffering effect was substantial. Given the low-cost and accessible nature of reserve-building activities, it would be worthwhile to encourage such activities for disabled individuals.

Deep learning and tree-based models for earth skin temperature forecasting in Malaysian environments

Predicting the Earth Skin Temperature (TS) using artificial intelligence (AI) has the potential to offer valuable insights into environmental changes and their impacts. TS has significant nonlinearity due to several meteorological parameters, including average temperature, maximum temperature, minimum temperature, relative humidity, surface pressure, wind speed, and wind direction. This study introduced four machine learning (ML) models, namely solo Long Short-Term Memory (LSTM), Autoencoder-LSTM, Classification and Regression Tree (CART), and a hybrid Convolutional Neural Network-LSTM (CNN-LSTM), to predict daily averaged TS at different locations over the Malaysian region. In the first stage of the modeling development, Pearson Correlation (PC) was adopted to measure the strength and direction of the relationship between input and output variables. The statistical analysis and visual interpretations demonstrated that the Autoencoder-LSTM model outperformed the CNN-LSTM, LSTM, and CART models for each simulated city. The Autoencoder-LSTM model showed outstanding performance at Kuantan in comparison with the seven cities, achieving a coefficient of determination (R2 = 0.965), Root Mean Square Error (RMSE = 0.183), Mean Absolute Error (MAE = 0.142), and Nash-Sutcliffe Efficiency (NSE = 0.96). The research findings suggested that the coordinates of each station are crucial in determining the level of data randomness, which ultimately affects the learning process of the ML models.

County-level socio-environmental factors and obesity prevalence in the United States.

To investigate high-risk sociodemographic and environmental determinants of health (SEDH) potentially associated with adult obesity in counties in the United States using machine-learning techniques. We performed a cross-sectional analysis of county-level adult obesity prevalence (body mass index ≥30 kg/m2) in the United States using data from the Diabetes Surveillance System 2017. We harvested 49 county-level SEDH factors that were used in a classification and regression trees (CART) model to identify county-level clusters. The CART model was validated using a 'hold-out' set of counties and variable importance was evaluated using Random Forest. Overall, we analysed 2752 counties in the United States, identifying a national median (interquartile range) obesity prevalence of 34.1% (30.2%, 37.7%). The CART method identified 11 clusters with a 60.8% relative increase in prevalence across the spectrum. Additionally, seven key SEDH variables were identified by CART to guide the categorization of clusters, including Physically Inactive (%), Diabetes (%), Severe Housing Problems (%), Food Insecurity (%), Uninsured (%), Population over 65 years (%) and Non-Hispanic Black (%). There is significant county-level geographical variation in obesity prevalence in the United States, which can in part be explained by complex SEDH factors. The use of machine-learning techniques to analyse these factors can provide valuable insights into the importance of these upstream determinants of obesity and, therefore, aid in the development of geo-specific strategic interventions and optimize resource allocation to help battle the obesity pandemic.

Fabrication and optimization of PES/PVP hollow fiber membranes using Box–Behnken model and CART algorithm

AbstractThis study focuses on the optimization of polyethersulfone (PES) hollow fiber membranes fabricated with the phase inversion method. A Box–Behnken experimental design was employed with three different PES concentration ratios (11, 14, 17 wt.%), three polyvinylpyrrolidone (PVP) molecular weight ratios (K30/K90 ratios of 6:0, 3:3, 0:6 wt.%), three different bore fluid (BF) composition ratios (water/alcohol ratios of 20:80, 60:40, 100:0), and three different air gap values (24, 37, and 50 cm). The results were analyzed in terms of pure water permeability (PWP) and porosity as optimization parameters using response surface methodology and the classification and regression tree (CART) model. ANOVA results revealed significant effects of PES concentration, PVP molecular weight, and BF composition on the outcomes. After optimization, the maximum PWP and the maximum porosity were obtained as 360.15 L/m2 h bar, 60.57%, respectively. The CART model achieved sufficient accuracy in classifying samples.

Identifying land use land cover dynamics using machine learning method and GIS approach in Karaivetti, Tamil Nadu

&lt;p&gt;An important analytical tool for tracking, mapping, and quantifying changes in land use and land cover (LULC) across time serves as the use of machine learning techniques. The environment and human activities both have the potential to change how land is used and covered. Classifying LULC types at different spatial scales has been effectively achieved by models like classification and regression trees (CART), support vector machines (SVM), extreme gradient boosting (XGBoost), and random forests (RF). To prepare images from Landsat before sending and analysis for an aspect of our research, we employed the Google Earth Engine. High-resolution imagery from Google Earth images were used to assess each kind of method and field data collection. Utilizing Geographic Information System (GIS) techniques, LULC fluctuations between 2015 and 2020 were assessed. According to our results, XGBoost, SVM, and CART models proved superior by the RF model regarding categorization precision. Considering the data, we collected between 2015 and 2020, from 11.57 hectares (1.74%) in 2015 to 184.19 hectares (27.65%) in 2020, the barren land experienced the greatest variation, that made an immense effect. Utilizing the support of satellite imagery from the Karaivetti Wetland, our work combines novel GIS techniques and machine learning strategies to LULC monitoring. The created land cover maps provide a vital benchmark that will be useful to authorities in formulating policies, managing for sustainability, and keeping track of degradation.&lt;/p&gt;

ESTIMATING THE POWER DRAW OF GRIZZLY FEEDERS USED IN CRUSHING–SCREENING PLANTS THROUGH SOFT COMPUTING ALGORITHMS

In this study, the power draw (P) of several grizzly feeders used in the Turkish Mining Industry (TMI) is investigated by considering the classification and regression tree (CART), random forest (RF) and adaptive neuro-fuzzy inference system (ANFIS) algorithms. For this purpose, a comprehensive field survey is performed to collect quantitative data, including power draw (P) of some grizzly feeders and their working conditions such as feeder width (W), feeder length (L), feeder capacity (Q), and characteristic feed size (F80). Before applying the soft computing methodologies, correlation analyses are performed between the input parameters and the output (P). According to these analyses, it is found that W and L are highly associated with P. On the other hand, Q is moderately correlated with P. Consequently, numerous soft computing models were run to estimate the P of the grizzly feeders. Soft computing analysis results demonstrate no superiority between the performances of RF and CART models. The RF analysis results indicate that the W is necessary for evaluating P for grizzly feeders. On the other hand, the ANFIS-based predictive model is found to be the best tool to estimate varying P values, and it satisfies promising results with a correlation of determination value (R2) of 0.97. It is believed that the findings obtained from the present study can guide relevant engineers in selecting the proper motors propelling grizzly feeders.

Predictors of flare-related inpatient or emergency department stay in systemic lupus erythematosus: A real-world analysis of Medicaid claims in the United States.

Systemic lupus erythematosus (SLE) is a chronic autoimmune disease characterized by multisystem inflammation. Medical management of SLE is based on reducing inflammation and tissue damage in the affected organs; however, medications used to treat SLE have been found to contribute to additional organ damage. Therefore, finding new ways to predict and prevent flares that require an inpatient (IP) stay or emergency department (ED) visit is critical for reducing the clinical and economic burden in patients with SLE. To identify risk factors of SLE flares requiring an IP/ED visit among a Medicaid-insured population with SLE. This retrospective study included patients from the Merative MarketScan Medicaid database (2013-2019). To capture patients at all stages of their SLE journey, all SLE claims for a patient were captured, and the index date was randomly selected among those claims that were at least 12 months after the first evidence of SLE. Patients were required to be continuously enrolled 1-year pre-index (year 1) and post-index (year 2). Demographics, clinical characteristics, and health care use and costs were measured in year 1, and flares requiring an IP/ED visit were identified in year 2 using the Garris algorithm. Multivariable logistic regression and classification and regression tree (CART) modeling were used to identify year 1 predictors and combination of factors, respectively, associated with flares-related IP/ED visits. Of the 8,083 patients included in the study, 37.6% of patients (n = 3,039) had a flare. Logistic regression identified ED visits in year 1 as one of the strongest predictors of flares-related IP/ED visits in year 2 (odds ratio = 2.19 [95% CI = 1.93-2.49]). SLE treatment progression to biologics (0.54 [0.42-0.70]) was the strongest predictor of decreased odds. Other strong predictors included other neurological disorders (1.63 [1.43-1.87]), Black race (1.49 [1.32-1.68]), chronic kidney disease/renal failure (1.35 [1.10-1.66]), and opioid use (1.30 [1.17-1.45]). CART modeling identified patients with an ED visit, an IP admission, and a diagnosis of Elixhauser Comorbidity Index-defined other neurological disorders in year 1 as having the highest probability of a flare-related IP/ED visit in year 2 (probability = 0.708), whereas patients without an ED visit had the lowest probability (probability = 0.185). Patients with the highest risk of a flare that required an IP/ED visit were those with a prior ED visit, IP admission, and other neurological disorders. Modeling also identified patients with prior opioid use, Black patients, and patients without SLE medications as subgroups with a high risk of a flare requiring an IP/ED visit.

Impact of land use/land cover changes on evapotranspiration and model accuracy using Google Earth engine and classification and regression tree modeling

This research uses a Classification and Regression Tree (CART) model with Google Earth Engine (GEE) to assess the winter season’s land cover and change detection mapping impact on the evapotranspiration (crop water requirement) parameters. Winter seasons, crucial for agricultural planning, and irrigation water requirement challenges in accurately mapping land cover and detecting changes due to the dynamic nature of farming practices during this period. In this study, Landsat-8 OLI images have been combined to map Land use and Land cover (LULC) and other change detection mapping in Akola Block, Maharashtra, India, during the 2018–2022 winter season. As an discoverer researcher that found detailed information of LULC classes during last 2018 to 2022 winter seasons, the use of the CART model in combination with a cloud-computing GEE demonstrates to be a practical approach for accurate land cover classification and change detection maps to create a pixel-based winter seasons information of study area. The novelty of this study lies in its innovative use of GEE, a powerful platform for remote sensing and geospatial analysis, to create LULC maps with remarkable accuracy. Achieving a 100% training accuracy across the four years under consideration is an exceptional feat, highlighting the reliability and stability of the methodology. Furthermore, the validation accuracy values, ranging from 89 to 94% for the winter seasons of 2018 to 2022, underscore the robustness of this approach. Such consistently high accuracy in mapping LULC over time is a groundbreaking achievement and offers a new dimension to the field of hydrology. For the hydrological community, the implications of this study are profound. Accurate LULC mapping and change detection provide critical data for modeling and analyzing the effects of land use changes on water resources, watershed management, and water quality. The User, Kappa, and Producer accuracy metrics used in this research highlight the model’s performance and its suitability for hydrological applications. These accurate LULC maps can aid in the development of hydrological models, forecasting, and decision-making processes, ultimately contributing to more effective water resource management and environmental conservation. In summary, this study’s innovative use of GEE, its remarkable accuracy in LULC mapping, and its relevance to the hydrological community demonstrate the potential for advanced remote sensing and geospatial tools to significantly improve our understanding of land use changes and their implications for water resources and environmental management.

Abstract A128: A novel approach to understanding disparities in cancer clinical trial enrollment

Abstract Background: Cancer patients who participate in clinical trials have access to novel therapies, increased monitoring, and potentially improved survival. However, this access is not universal. Striking disparities in clinical trial enrollment have been linked to race, sex, and geography. However, studies on this topic are limited due to a lack of patient-specific socioeconomic data. Methods: This analysis used a novel linkage of administrative patient data from University Hospitals Seidman Cancer Center (Cleveland, Ohio) linked to the LexisNexis Socioeconomic Health Attributes, a robust dataset containing 442 patient-specific attributes gathered from public records regarding income, education, housing stability, property ownership and value, social support, and many others. We included cancer patients aged 18-75 years at the time of the latest cancer diagnosis (before trial enrollment if patient was enrolled) during 2007-2022. Because of the substantial differences in the demographic and clinical characteristics between the trial and non-trial patients, we applied the propensity score matching technique to one-to-one match all trial patients to non-trial patients by age, sex, race/ethnicity, cancer type, and year of the latest cancer diagnosis. Next, we applied the classification and regression tree (CART) machine learning algorithm to identify “phenotypes”, or combinations of patient socioeconomic characteristics that predict their trial enrollment status. We additionally performed the random forest algorithm to determine whether our CART model captured the most important variables in predicting trial enrollment. Results: We identified 28,671 cancer patients with a mean and median age of 60 and 62 years. Of those, 2,479 patients were enrolled in trials, and a corresponding 2,479 non-trial patients were matched. We identified 9 phenotypes of trial enrollment from the results of CART. The percentage of patients enrolled in trials ranges from 30% to 80%. Patients belonging to the highest-enrollment phenotype were identified by their estimated annual income exceeding $50,000, higher fraud-risk level, and age under 61 years. Patients falling into the lowest-enrollment phenotype were characterized by an estimated annual income below $30,000, an estimated home value below $243,000, and no college attendance. Variables identified by CART were also among the top-ranked variables in the random forest regarding their variable. Other important variables identified by random forest include, whether the patient ever registered to vote, risk of not being motivated to manage one's own health, and business records. Conclusion: Compared to community-level data, the use of patient-specific socioeconomic data provides a more granular understanding of clinical trial disparities. The use of machine learning methods of CART and random forest offers a promising avenue for future disparities research. This novel approach will deepen our understanding of cancer disparities and improve our ability to intervene and improve care for our most vulnerable patients. Citation Format: Weichuan Dong, Jamie Shoag, Tamila Kindwall-Keller, Ali Kara, John Shanahan, Debora Bruno, Johnie Rose, Siran Koroukian, Richard Hoehn. A novel approach to understanding disparities in cancer clinical trial enrollment [abstract]. In: Proceedings of the 16th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2023 Sep 29-Oct 2;Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(12 Suppl):Abstract nr A128.

Bayesian CART models for insurance claims frequency

The accuracy and interpretability of a (non-life) insurance pricing model are essential qualities to ensure fair and transparent premiums for policy-holders, that reflect their risk. In recent years, classification and regression trees (CARTs) and their ensembles have gained popularity in the actuarial literature, since they offer good prediction performance and are relatively easy to interpret. In this paper, we introduce Bayesian CART models for insurance pricing, with a particular focus on claims frequency modelling. In addition to the common Poisson and negative binomial (NB) distributions used for claims frequency, we implement Bayesian CART for the zero-inflated Poisson (ZIP) distribution to address the difficulty arising from the imbalanced insurance claims data. To this end, we introduce a general MCMC algorithm using data augmentation methods for posterior tree exploration. We also introduce the deviance information criterion (DIC) for tree model selection. The proposed models are able to identify trees which can better classify the policy-holders into risk groups. Simulations and real insurance data will be used to illustrate the applicability of these models.

A Pork Price Prediction Model Based on a Combined Sparrow Search Algorithm and Classification and Regression Trees Model

The frequent fluctuation of pork prices has seriously affected the sustainable development of the pork industry. The accurate prediction of pork prices can not only help pork practitioners make scientific decisions but also help them to avoid market risks, which is the only way to promote the healthy development of the pork industry. Therefore, to improve the prediction accuracy of pork prices, this paper first combines the Sparrow Search Algorithm (SSA) and traditional machine learning model, Classification and Regression Trees (CART), to establish an SSA-CART optimization model for predicting pork prices. Secondly, based on the Sichuan pork price data during the 12th Five-Year Plan period, the linear correlation between piglet, corn, fattening pig feed, and pork price was measured using the Pearson correlation coefficient. Thirdly, the MAE fitness value was calculated by combining the validation set and training set, and the hyperparameter “MinLeafSize” was optimized via the SSA. Finally, a comparative analysis of the prediction performance of the White Shark Optimizer (WSO)-CART model, CART model, and Simulated Annealing (SA)-CART model demonstrated that the SSA-CART model has the best prediction of pork price (compared with a single decision tree, R2 increased by 9.236%), which is conducive to providing support for pork price prediction. The accurate prediction of pork prices with an optimized machine learning model is of great practical significance for stabilizing pig production, ensuring the sustainable growth of farmers’ income, and promoting sound economic development.

2905. Development of a Desirability of Outcome Ranking (DOOR) for Skin Infection Clinical Trials

Abstract Background Infectious disease clinical trials treatment outcomes have almost exclusively been dichotomized into cure and failure. The novel Desirability of Outcome Ranking (DOOR) method uses a hierarchical ordinal outcome of intervention outcomes. DOOR measures are more sensitive for detection of differences between groups compared to dichotomous outcomes and thus can lead to dramatic reduction of the number of required trial participants. Additionally, the DOOR design addresses patient-centered outcomes that are ignored in traditional infectious diseases clinical trial outcomes. DOOR methodology has neither been used for skin and soft tissue infection (SSTI) clinical trial outcomes nor incorporated patients about their experiences with SSTIs. Methods We conducted 6 focus groups of providers, patients, and parents at two medical centers in Los Angeles and St. Louis. Focus groups were composed of clinicians who treat SSTIs (n=2), patients who had SSTIs (n=3) and parents of children who had an SSTI (n=1). Focus group participants were given a series of SSTI clinical scenarios with various outcomes and asked to rank these 10 outcomes on a numerical scale from most to least desirable. We fitted 4 CART (classification and regression tree) models for overall, provider, patient and parent responders, respectively. Results In total, we had 62 focus group participants (24 providers, 30 patients, and 8 parents). Overall, the most desirable outcome was no complications (rank value = 1.0), followed by rash or nausea (mean rank values = 4.9 - 5.1), diarrhea or recurrent infection (= 5.9), dizziness or stomach ache (= 6.1 - 6.2), recurrent drainage procedure or household transmission (= 7.0 - 7.2), and hospitalization (= 8.3). Least desirable outcomes differed between groups, with providers choosing hospitalization while patient/parents indicated household transmission. Variance in outcome desirability differed more widely between patients compared to providers. (Figure). Conclusion We developed a DOOR outcome for SSTIs clinical trials that incorporates both provider and patient perspectives. This metric could be tested in clinical trials as a means to reduce sample size and to incorporate patient-focused outcomes. Disclosures Loren G. Miller, MD MPH, ContraFect: Grant/Research Support|GSK: Grant/Research Support|Medline: Grant/Research Support|Merck: Grant/Research Support|Paratek: Grant/Research Support

Habitat modelling of native freshwater mussels distinguishes river specific differences in the Detroit and St. Clair rivers of the Laurentian Great Lakes

Native freshwater mussels (Bivalvia: Unionidae) were seemingly pushed to extirpation following the establishment of dreissenid mussels in the St. Clair (SCR)–Detroit River (DR) System and the current state of unionids in the main channels of the SCR remains unknown. To assess remnant unionid populations, the DR and SCR were surveyed in 2019 and 2021, respectively, using a mixture of stratified random, historical, potential refuge, and model-selected (SCR sampling only) sites (n = 56 DR sites and 51 SCR sites). Data collected from the DR (2019) were used to create unionid species distribution models in MaxEnt, which produced the model-selected sites for sampling the SCR. A total of 14 live unionids representing 9 species were found among 7 sites in the SCR; however, the DR model projected to the SCR failed to be predictive for unionid presence in the SCR (0 % success). Additional models were created for the DR and SCR using MaxEnt and classification and regression tree (CART) Analysis to elucidate the characteristic differences between the two rivers. MaxEnt modelling showed the highest contributing variable for the DR model was water velocity (m/s) while the highest contributing variable for the SCR model was river distance to the outlet which could be one explanation for why the DR model did not accurately predict unionid presence in the SCR. Additionally, when DR and SCR data were combined in the CART model, the model failed to predict any live unionid sites from the SCR. Variability in model predictions demonstrate the characteristic differences between the DR and SCR. Therefore, model verification for both rivers specific models took place in 2022 (n = 8 DR sites and 6 SCR sites). The DR unionid species distribution model identified 3 additional sites with live unionids (38 % of predicted sites), but the SCR model did not locate any additional live unionid locations (0 % of predicted sites). This research updated our knowledge on native unionid distributions and habitat use in the DR and SCR which could contribute towards unionid conservation in the future.

Association between High Bleeding Risk and 2-Year Mortality in Patients with Chronic Limb-Threatening Ischemia.

Patients with chronic limb-threatening ischemia (CLTI) have a high bleeding risk (HBR) and mortality rate. The 2-year life expectancy is an important factor in deciding the appropriate treatment strategy. This study aimed to assess the influence of HBR on the prognosis of patients with CLTI. A total of 259 patients with CLTI who underwent endovascular therapy (EVT) (mean age, 76.2 years; male, 62.9%) between January 2018 and December 2019 were evaluated. The Academic Research Consortium for HBR (ARC-HBR) criteria were applied to each patient, and the ARC-HBR scores were calculated. The cut-off score for predicting all-cause mortality within two years was derived using a survival classification and regression tree (CART) model. Causes of death and the association between ARC-HBR scores and major bleeding events within two years were also investigated. Based on the CART model, patients were divided into three groups (low HBR score 0-1.0, 48 patients; moderate HBR score 1.5-3.0, 176 patients; and high HBR score ≥ 3.5, 35 patients). During the study period, 82 patients (39.6%) died due to cardiac (n=23) and non-cardiac causes (n=59). All-cause mortality increased significantly with increasing ARC-HBR scores. Cox multivariate analysis revealed a significant association between high ARC-HBR scores and the risk of all-cause mortality within two years. Major bleeding events increased significantly with increasing ARC-HBR scores. The ARC-HBR score could predict 2-year mortality in patients with CLTI who underwent EVT. Thus, this score can help determine the best revascularization strategy for patients with CLTI.

The association between falls and depressive symptoms among older adults: evidence from the China Health and Retirement Longitudinal Study.

Falls place a heavy burden on older adults and families, and there was little research on the relationship between falls and depressive symptoms among older adults in China. This study is designed to examine the association between falls and depressive symptoms in Chinese older adults. This study was based on 9,539 data sets from the China Health and Retirement Longitudinal Study (CHARLS) in 2018. The 10-item Center for Epidemiologic Studies-Depression Scale (CESD-10) was used to access depressive symptoms in older adults. A logistic regression model was used to calculate multivariate odds ratios (ORs) and 95% confidence intervals (CIs) for falls and depressive symptoms, adjusted for possible confounders. The Classification and regression tree (CART) demonstrates the prediction of the target variable values based on other variables. In this study, 9,539 older people were selected: 60-69 years old accounted for 63.0%, 70-79 years old accounted for 29.7%, and 80 years old and above accounted for 7.3%. Male accounted for 49.7% and female for 50.3%. The rate of falls among older adults was 21.4%, and the rate of depressive symptoms was 33.9%. Adjusted ORs (OR = 1.37, 95% CI: 1.23, 1.53) showed a significant association between falls and depressive symptoms among older adults. Subgroup analysis revealed that this association was statistically significant across male (OR = 1.29, 95% CI: 1.23, 1.53) and female (OR = 1.42, 95% CI: 1.23, 1.64), 60-69 aged (OR = 1.38, 95% CI: 1.19, 1.60) and 70-79 aged (OR =1.42, 95% CI: 1.16, 1.74), rural (OR = 1.42, 95% CI: 1.25, 1.61), <15,000 CNY (OR = 1.35, 95% CI: 1.19, 1.54) and more than 25,000 CNY (OR = 1.42, 95% CI: 1.09, 1.85). Additionally, The CART model showed that the probability (73.0%) of falls was highest among older adults with depressive symptoms who self-rated poor health and female gender. This cross-sectional study demonstrated a significant association between falls and depressive symptoms in Chinese older adults. The findings provide some evidence and support for risk monitoring, screening for depressive symptoms, and early prevention in the high-risk older population.

Prediction of corporate credit ratings with machine learning: Simple interpretative models

This study utilizes machine learning techniques, notably classification and regression trees (CART) and support vector regression (SVR), to predict corporate credit ratings. While SVR marginally outperforms in accuracy, CART offers interpretability. However, unconstrained models can produce non-monotonic relationships between credit ratings and core features, an undesired outcome. To circumvent this, we recommend restricted CART models that ensure interpretable, theory-consistent results. We underscore the importance of company size in credit rating prediction with an ideal model integrating size, interest coverage, and dividends. Although being a large-cap company is crucial, it doesn't guarantee high ratings, and small-cap companies rarely secure investment-grade ratings.

Validity of Diagnostic Support Model for Attention Deficit Hyperactivity Disorder: A Machine Learning Approach.

An accurate and early diagnosis of attention deficit hyperactivity disorder can improve health outcomes and prevent unnecessary medical expenses. This study developed a diagnostic support model using a machine learning approach to effectively screen individuals for attention deficit hyperactivity disorder. Three models were developed: a logistic regression model, a classification and regression tree (CART), and a neural network. The models were assessed by using a receiver operating characteristic analysis. In total, 74 participants were enrolled into the disorder group, while 21 participants were enrolled in the control group. The sensitivity and specificity of each model, indicating the rate of true positive and true negative results, respectively, were assessed. The CART model demonstrated a superior performance compared to the other two models, with region values of receiver operating characteristic analyses in the following order: CART (0.848) > logistic regression model (0.826) > neural network (0.67). The sensitivity and specificity of the CART model were 78.8% and 50%, respectively. This model can be applied to other neuroscience research fields, including the diagnoses of autism spectrum disorder, Tourette syndrome, and dementia. This will enhance the effect and practical value of our research.