Predictive Logistic Regression Model Research Articles

Pathological features associated with metastasis in patients with early invasive (pT1) colorectal carcinoma in colorectal polyps.

Colorectal carcinoma (CRC) arising in a colorectal polyp with invasion limited to the submucosa is sufficiently treated by complete endoscopic resection alone in many cases. Histological features of the carcinoma including tumour size, vascular invasion and poor tumour differentiation or evidence of de-differentiation, such as tumour budding, are associated with a higher risk for metastasis such that oncological resection is recommended. However, most malignant polyps with these features do not have lymph node metastases at the time of resection, so there is a need for better refinement of the histological risk features. A total of 437 consecutive colorectal polyps with submucosal invasive carcinoma from a single centre, 57 of which had metastatic disease, were supplemented by 30 cases with known metastatic disease from two additional centres. Clinical and histological features of the polyp cancers were reviewed looking for differences between the 87 cancers with metastatic disease and the remaining cases without metastasis. A subgroup of 204 polyps removed intact was also analysed to ensure maximum histological accuracy. This study confirmed larger invasive tumour size, vascular invasion and poor tumour differentiation as adverse predictive features. Prominent peritumoral desmoplasia and high cytological grade were additional adverse features. A predictive logistic regression model comprised of (i) presence of any form of vascular invasion; (ii) presence of high tumour budding (BD3); (iii) width of invasive tumour component > 8 mm; (iv) depth of invasive tumour > 1.5 mm; and (v) the finding of prominent expansile desmoplasia located within and beyond the deep invasive edge of the carcinoma, showed excellent performance in predicting metastatic disease.

Identifying Postural Instability in Children with Cerebral Palsy Using a Predictive Model: A Longitudinal Multicenter Study.

Insufficient postural control and trunk instability are serious concerns in children with cerebral palsy (CP). We implemented a predictive model to identify factors associated with postural impairments such as spastic or hypotonic truncal tone (TT) in children with CP. We conducted a longitudinal, double-blinded, multicenter, descriptive study of 102 teenagers with CP with cognitive impairment and severe motor disorders with and without truncal tone impairments treated in two specialized hospitals (60 inpatients and 42 outpatients; 60 males, mean age 16.5 ± 1.2 years, range 12 to 18 yrs). Clinical and functional data were collected between 2006 and 2021. TT-PredictMed, a multiple logistic regression prediction model, was developed to identify factors associated with hypotonic or spastic TT following the guidelines of "Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis". Predictors of hypotonic TT were hip dysplasia (p = 0.01), type of etiology (postnatal > perinatal > prenatal causes; p = 0.05), male gender, and poor manual (p = 0.01) and gross motor function (p = 0.05). Predictors of spastic TT were neuromuscular scoliosis (p = 0.03), type of etiology (prenatal > perinatal > postnatal causes; p < 0.001), spasticity (quadri/triplegia > diplegia > hemiplegia; p = 0.05), presence of dystonia (p = 0.001), and epilepsy (refractory > controlled, p = 0.009). The predictive model's average accuracy, sensitivity, and specificity reached 82%. The model's accuracy aligns with recent studies on applying machine learning models in the clinical field.

Providence Bracing: Predicting the Progression to Surgery in Patients With Braced Idiopathic Scoliosis.

Idiopathic scoliosis (IS) is a common spinal abnormality, in which orthotic management can reduce progression to surgery. However, predictors of bracing success are still not fully understood. We studied a large patient population treated with the nighttime Providence orthosis, utilizing multivariable logistic regression to assess results and predict future spine surgery. We retrospectively reviewed patients with IS meeting Scoliosis Research Society inclusion and assessment criteria presenting from April 1994 to June 2020 at a single institution and treated with a Providence orthosis. A predictive logistic regression model was developed utilizing the following candidate features: age, sex, body mass index, Risser classification, Lenke classification, curve magnitude at brace initiation, percentage correction in a brace, and total months of brace use. Model performance was assessed using the area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity. The importance of individual features was assessed using the variable importance score. There were 329 consecutive patients with IS with a mean age of 12.8 ± 1.4 years that met inclusion and assessment criteria. Of these, 113 patients (34%) ultimately required surgery. The model's area under the curve (AUC) was 0.72 on the testing set, demonstrating good discrimination. The initial curve magnitude (Importance score: 100.0) and duration of bracing (Importance score: 82.4) were the 2 most predictive features for curve progression leading to surgery. With respect to skeletal maturity, Risser 1 (Importance score: 53.9) had the most predictive importance for future surgery. For the curve pattern, Lenke 6 (Importance score: 52.0) had the most predictive importance for future surgery. Out of 329 patients with IS treated with a Providence nighttime orthosis, 34% required surgery. This is similar to the findings of the BrAist study of the Boston orthosis, in which 28% of monitored braced patients required surgery. In addition, we found that predictive logistic regression can evaluate the likelihood of future spine surgery in patients treated with the Providence orthosis. The severity of the initial curve magnitude and total months of bracing were the 2 most important variables when assessing the probability of future surgery. Surgeons can use this model to counsel families on the potential benefits of bracing and risk factors for curve progression.

Validating the use of machine-learning cancer staging algorithms for Medicare cost analyses.

e13548 Background: Administrative claims provide valuable real-world insight into the care of cancer patients; however, claims data lacks cancer stage information. This limitation constrains research on the value of early diagnosis and treatment as well as on the costs and savings associated with increased cancer screening. In prior work, our team used the SEER-Medicare data to develop machine learning (ML) algorithms to stage non-small cell lung (NSCLC), colon (CC), and rectal (RC) cancer patients using clinical flags derived from claims data. These algorithms were 69% (RC), 78% (NSCLC), and 83% (CC) accurate at matching incident cancer patients with their SEER-recorded AJCC stage (SEER-stage) at diagnosis. This work sought to test whether these ML algorithms are sufficiently accurate for use in claim cost analyses. Methods: Incident NSCLC, CC, and RC patients were identified using 2016-2017 SEER-Medicare data and assigned a cancer stage using a claims-based predictive multinomial logistic regression model (R Statistical Software - v4.1.2 R Core Team 2021; nnet package - Venables and Ripley 2002). Patients’ cumulative medical and pharmacy costs were summarized for 12 months starting with patients’ index month. Patients’ Medicare index month was set equal to the month of their first claim with a cancer diagnosis. Patients’ SEER index month was set equal to the diagnosis month associated with their incident tumor record in SEER. Patients with each cancer type were then grouped two ways - by ML-stage and by SEER-stage. Median patient costs were compared between stage groups for each cancer type and differences tested for statistical significance using Wilcoxon Rank-Sum Testing. Results: For NSCLC and CC, raw differences in median 12-month claim costs between the ML- and SEER-stage cohorts were small (1%-4%). Cost differences for RC were higher (7%-17%). ML and SEER costs were not significantly different (p &gt; 0.05) between later-stage cohorts (NSCLC stages 3 and 4, CC stages 2C/3 and 4, and RC stage 4); however, early-stage groups were always significantly different (p &lt; 0.05). Conclusions: Although costs were not statistically equivalent across all stage groups, the similarity of ML and SEER costs across higher-stage cohorts and small raw differences in median costs for each NSCLC and CC group suggests that ML algorithms with higher accuracy may be used to develop costs from administrative data for stage shift modeling and cost tradeoff analyses. [Table: see text]

Risk Factors Analysis of Post Anesthesia Induction Hypotension in Patients with Recurrent Nasopharyngeal Carcinoma

This study is to investigate the risk factors of post anesthesia induction(PIH) in patients with recurrent nasopharyngeal carcinoma. A retrospective analysis was conducted on the clinical data of 262 patients who underwent the nasal endoscopic surgery from January 2021 to May 2023 were into this study. These patients were divided into group A(patients of non-PIH) and group B(patients of PIH). Univariate analysis and multivariate Logistic regression analysis were performed to analyze the independent risk factors of PIH. The overall incidence rate of PIH was 17.56%. According to the univariate analyses, anesthetics used for induction (dexmetomidine) and Preoperative serum sodium level were significantly correlated with PIH in patients with recurrent nasopharyngeal carcinoma(p&lt;0.05). Furthermore, multivariate logistic regression prediction model identified the two indicators above are independent risk factors for PIH in patients. The model fit well, and the area under the receiver operating characteristic curve was 0.676, and the predictive sensitivity and specificity were 78% and 53.2%, respectively. The results of this paper show that, anesthetics used for induction (dexmetomidine) and Preoperative serum sodium level were independent risk factors for PIH. It provides theoretical foundation for predicting PIH after splenectomy.

Evaluation of test-day milk somatic cell count to predict intramammary infection in late lactation grazing dairy cows

Use of selective dry cow antimicrobial therapy requires to precisely differentiate cows with an intramammary infection (IMI) from uninfected cows close to drying-off to enable treatment allocation. Milk somatic cell count (SCC) is an indicator of an inflammatory response in the mammary gland and is usually associated with IMI. However, SCC can also be influenced by cow-level variables such as milk yield, lactation number and stage of lactation. In recent years, predictive algorithms have been developed to differentiate cows with IMI from cows without IMI based on SCC data. The objective of this observational study was to explore the association between SCC and subclinical IMI, taking cognizance of cow-level predictors on Irish seasonal spring calving, pasture-based systems. Additionally, the optimal test-day SCC cut-point (maximized sensitivity and specificity) for IMI diagnosis was determined. A total of 2,074 cows, across 21 spring calving dairy herds with an average monthly milk weighted bulk tank SCC of ≤200,000 cells/mL were enrolled in the study. Quarter-level milk sampling was carried out on all cows in late lactation (interquartile range = 240-261 d in milk) for bacteriological culturing. Bacteriological results were used to define cows with IMI, when ≥1 quarter sample resulted in bacterial growth. Cow-level test-day SCC records were provided by the herd owners. The ability of the average, maximum and last test-day SCC to predict infection were compared using receiver operator curves. Predictive logistic regression models tested included parity (primiparous or multiparous), yield at last test-day and a standardized count of high SCC test-days. In total, 18.7% of cows were classified as having an IMI, with first parity cows having a higher proportion of IMI (29.3%) compared with multiparous cows (16.1%). Staphylococcus aureus accounted for the majority of these infections. The last test-day SCC was the best predictor of infection with the highest area under the curve. The inclusions of parity, yield at last test-day, and a standardized count of high SCC test-days as predictors did not significantly improve the ability of last test-day SCC to predict IMI. The cut-point for last test-day SCC which maximized sensitivity and specificity was 64,975 cells/mL. This study indicates that in Irish seasonal pasture-based dairy herds,with low bulk tank SCC control programs, the last test-day SCC (interquartile range days in milk = 221-240) is the best predictor of IMI in late lactation.

A Metabolism-Based Interpretable Machine Learning Prediction Model for Diabetic Retinopathy Risk: A Cross-Sectional Study in Chinese Patients with Type 2 Diabetes.

The burden of diabetic retinopathy (DR) is increasing, and the sensitive biomarkers of the disease were not enough. Studies have found that the metabolic profile, such as amino acid (AA) and acylcarnitine (AcylCN), in the early stages of DR patients might have changed, indicating the potential of metabolites to become new biomarkers. We are amid to construct a metabolite-based prediction model for DR risk. This study was conducted on type 2 diabetes (T2D) patients with or without DR. Logistic regression and extreme gradient boosting (XGBoost) prediction models were constructed using the traditional clinical features and the screening features, respectively. Assessing the predictive power of the models in terms of both discrimination and calibration, the optimal model was interpreted using the Shapley Additive exPlanations (SHAP) to quantify the effect of features on prediction. Finally, the XGBoost model incorporating AA and AcylCN variables had the best comprehensive evaluation (ROCAUC = 0.82, PRAUC = 0.44, Brier score = 0.09). C18 : 1OH lower than 0.04 μmol/L, C18 : 1 lower than 0.70 μmol/L, threonine higher than 27.0 μmol/L, and tyrosine lower than 36.0 μmol/L were associated with an increased risk of developing DR. Phenylalanine higher than 52.0 μmol/L was associated with a decreased risk of developing DR. In conclusion, our study mainly used AAs and AcylCNs to construct an interpretable XGBoost model to predict the risk of developing DR in T2D patients which is beneficial in identifying high-risk groups and preventing or delaying the onset of DR. In addition, our study proposed possible risk cut-off values for DR of C18 : 1OH, C18 : 1, threonine, tyrosine, and phenylalanine.

Abstract 515: A Prediction Model For Left Ventricular Thrombus Persistence/recurrence: Based On A Prospective Study And A Retrospective Study

Background: This study aimed to develop a prediction model to forecast thrombus persistence or recurrence in patients with left ventricular (LV) thrombus. Methods: We enrolled patients prospectively from 2020 to 2022 and retrospectively from 2013 to 2019 at the National Center of Cardiovascular Diseases of China. The two cohorts were combined to derive predictive thrombus persistence/recurrence models. The primary study comprised patients who had been consistently taking systemic oral anticoagulants and had imaging records available at the end of a three-month follow-up period. The Lasso regression algorithm was performed to select independent predictors and a nomogram risk prediction model was applied as a risk stratification tool. Results: A total of 172 patients were included, with 124 patients in a training set and 48 patients in a validation set. Six predictors were incorporated into the multivariate logistic regression prediction model. The area under the receiving operating characteristic was 0.852 in the training set and 0.631 in the validation set. Patients with protuberant thrombus and higher baseline D-dimer levels had a reduced risk of persistence/recurrence (OR 0.17, 95% CI 0.03-0.69, P = 0.025; OR 0.67, 95% CI 0.43-0.91, P = 0.030, separately), whereas thicker thrombus was linked to an increased rate of persistent thrombus (OR 1.11, 95% CI 1.05-1.20, P = 0.002). Additionally, thrombus persistence/recurrence rate was greater in patients with nonischemic cardiomyopathy and antiplatelet therapy. Conclusions: This prediction model provides tools and allows the identification of characteristics associated with unresolved thrombus.

Cerebral collateral circulation as an independent predictor for in-stent restenosis after carotid artery stenting.

In-stent restenosis is a crucial problem after carotid artery stenting, but the exact predictors of in-stent restenosis remain unclear. We aimed to evaluate the effect of cerebral collateral circulation on in-stent restenosis after carotid artery stenting and to establish a clinical prediction model for in-stent restenosis. This retrospective case-control study enrolled 296 patients with severe carotid artery stenosis of C1 segment (≥70%) who underwent stent therapy from June 2015 to December 2018. Based on follow-up data, the patients were divided into the in-stent restenosis and no in-stent restenosis groups. The collateral circulation of the brain was graded according to the criteria of the American Society for Interventional and Therapy Neuroradiology/Society for Interventional Radiology (ASITN/SIR). Clinical data were collected, such as age, sex, traditional vascular risk factors, blood cell count, high-sensitivity C-reactive protein, uric acid, stenosis degree before stenting and residual stenosis rate, and medication after stenting. Binary logistic regression analysis was performed to identify potential predictors of in-stent restenosis, and a clinical prediction model for in-stent restenosis after carotid artery stenting was established. Binary logistic regression analysis showed that poor collateral circulation was an independent predictor of in-stent restenosis (P=0.003). We also found that a 1% increase in residual stenosis rate was associated with a 9% increase in the risk of in-stent restenosis (P=0.02). Ischemic stroke history (P=0.03), family history of ischemic stroke (P<0.001), in-stent restenosis history (P<0.001), and nonstandard medication after stenting (P=0.04) were predictors of in-stent restenosis. The risk of in-stent restenosis was lowest when the residual stenosis rate was 12.5% after carotid artery stenting. Furthermore, we used some significant parameters to construct a binary logistic regression prediction model for in-stent restenosis after carotid artery stenting in the form of a nomogram. Collateral circulation is an independent predictor of in-stent restenosis after successful carotid artery stenting, and the residual stenosis rate tends to be below 12.5% to reduce restenosis risk. The standard medication should be strictly carried out for patients after stenting to prevent in-stent restenosis.

A prediction model for left ventricular thrombus persistence/recurrence: based on a prospective study and a retrospective study

BackgroundIt remains unknown whether anticoagulation for persistent left ventricular (LV) thrombus should be continued indefinitely. Identifying patients with a high risk of thrombus unresolved may be helpful to determine the optimum anticoagulation duration. This study aimed to develop a prediction model to forecast thrombus persistence or recurrence in patients with LV thrombus.MethodsWe enrolled patients prospectively from 2020 to 2022 and retrospectively from 2013 to 2019 at the National Center of Cardiovascular Diseases of China. The two cohorts were then combined to derive predictive models of thrombus persistence/recurrence. The primary study comprised patients who received systemic oral anticoagulants and had imaging records available at the end of a 3-month follow-up period. The Lasso regression algorithm and the logistic regression were performed to select independent predictors. The calibration curve was generated and a nomogram risk prediction model was applied as a risk stratification tool.ResultsA total of 172 (64 in the prospective cohort and 108 in the retrospective cohort) patients were included, with 124 patients in a training set and 48 patients in a validation set. Six predictors were incorporated into the multivariate logistic regression prediction model. The area under the receiving operating characteristic was 0.852 in the training set and 0.631 in the validation set. Patients with protuberant thrombus and higher baseline D-dimer levels had a reduced risk of persistence/recurrence (OR 0.17, 95% CI 0.03–0.69, P = 0.025; OR 0.67, 95% CI 0.43–0.91, P = 0.030, separately), whereas thicker thrombus was linked to an increased rate of persistent thrombus (OR 1.11, 95% CI 1.05–1.20, P = 0.002). Additionally, patients with diverse diagnoses or receiving different antiplatelet treatments had different rates of LV thrombus persistence/recurrence at 3 months.ConclusionsThis prediction model provides tools to forecast the occurrence of persistent/recurrent thrombus and allows the identification of characteristics associated with unresolved thrombus. To validate the model and determine the duration of anticoagulation in patients with persistent thrombus, prospective randomized trials are necessary.

Development of a Predictive Model of Cardiovascular Risk in a Male Population from the Peruvian Amazon.

The coexistence of malnutrition due to over- and under-nutrition in the Peruvian Amazon increases chronic diseases and cardiovascular risk. A cross-sectional study of a male population where anthropometric, clinical, and demographic variables were obtained to create a binary logistic regression predictive model of cardiovascular risk. We compared two methods with good predictive results, finally choosing Model 4 (r2 = 0.57, sensitivity 73.68%, specificity 95.35%, Youden index 0.69, and validity index 94.21), with non-invasive variables such as blood pressure (p < 0.001), hip circumference (p < 0.001), and FINDRISC test result (p < 0.05); Conclusions: We developed a cheap, fast, and non-invasive tool to determine cardiovascular risk in the population of this endemic area.

Comparison of Machine Learning Algorithms for Predicting Hospital Readmissions and Worsening Heart Failure Events in Patients With Heart Failure With Reduced Ejection Fraction: Modeling Study.

Heart failure (HF) is highly prevalent in the United States. Approximately one-third to one-half of HF cases are categorized as HF with reduced ejection fraction (HFrEF). Patients with HFrEF are at risk of worsening HF, have a high risk of adverse outcomes, and experience higher health care use and costs. Therefore, it is crucial to identify patients with HFrEF who are at high risk of subsequent events after HF hospitalization. Machine learning (ML) has been used to predict HF-related outcomes. The objective of this study was to compare different ML prediction models and feature construction methods to predict 30-, 90-, and 365-day hospital readmissions and worsening HF events (WHFEs). We used the Veradigm PINNACLE outpatient registry linked to Symphony Health's Integrated Dataverse data from July 1, 2013, to September 30, 2017. Adults with a confirmed diagnosis of HFrEF and HF-related hospitalization were included. WHFEs were defined as HF-related hospitalizations or outpatient intravenous diuretic use within 1 year of the first HF hospitalization. We used different approaches to construct ML features from clinical codes, including frequencies of clinical classification software (CCS) categories, Bidirectional Encoder Representations From Transformers (BERT) trained with CCS sequences (BERT + CCS), BERT trained on raw clinical codes (BERT + raw), and prespecified features based on clinical knowledge. A multilayer perceptron neural network, extreme gradient boosting (XGBoost), random forest, and logistic regression prediction models were applied and compared. A total of 30,687 adult patients with HFrEF were included in the analysis; 11.41% (3184/27,917) of adults experienced a hospital readmission within 30 days of their first HF hospitalization, and nearly half (9231/21,562, 42.81%) of the patients experienced at least 1 WHFE within 1 year after HF hospitalization. The prediction models and feature combinations with the best area under the receiver operating characteristic curve (AUC) for each outcome were XGBoost with CCS frequency (AUC=0.595) for 30-day readmission, random forest with CCS frequency (AUC=0.630) for 90-day readmission, XGBoost with CCS frequency (AUC=0.649) for 365-day readmission, and XGBoost with CCS frequency (AUC=0.640) for WHFEs. Our ML models could discriminate between readmission and WHFE among patients with HFrEF. Our model performance was mediocre, especially for the 30-day readmission events, most likely owing to limitations of the data, including an imbalance between positive and negative cases and high missing rates of many clinical variables and outcome definitions. We predicted readmissions and WHFEs after HF hospitalizations in patients with HFrEF. Features identified by data-driven approaches may be comparable with those identified by clinical domain knowledge. Future work may be warranted to validate and improve the models using more longitudinal electronic health records that are complete, are comprehensive, and have a longer follow-up time.

Hepatitis C Screening and Antibody Prevalence Among Newly Arrived Refugees to the United States, 2010–2017

Six refugee screening sites collaborated to estimate the prevalence of hepatitis C virus (HCV) antibodies among newly arrived refugees in the United States from 2010 to 2017, identify demographic characteristics associated with HCV antibody positivity, and estimate missed HCV antibody-positive adults among unscreened refugees. We utilized a cross-sectional study to examine HCV prevalence among refugees (N = 144,752). A predictive logistic regression model was constructed to determine the effectiveness of current screening practices at identifying cases. The prevalence of HCV antibodies among the 64,703 refugees screened was 1.6%. Refugees from Burundi (5.4%), Moldova (3.8%), Democratic Republic of Congo (3.2%), Burma (2.8%), and Ukraine (2.0%) had the highest positivity among refugee arrivals. An estimated 498 (0.7%) cases of HCV antibody positivity were missed among 67,787 unscreened adults. The domestic medical examination represents an opportunity to screen all adult refugees for HCV to ensure timely diagnosis and treatment.

Logistic regression prediction models and key influencing factors analysis of diabetes based on algorithm design

Logistic regression prediction models and key influencing factors analysis of diabetes based on algorithm design

Development and Validation of a Diabetic Retinopathy Risk Stratification Algorithm.

Although diabetic retinopathy is a leading cause of blindness worldwide, diabetes-related blindness can be prevented through effective screening, detection, and treatment of disease. The study goal was to develop risk stratification algorithms for the onset of retinal complications of diabetes, including proliferative diabetic retinopathy, referable retinopathy, and macular edema. Retrospective cohort analysis of patients from the Kaiser Permanente Northern California Diabetes Registry who had no evidence of diabetic retinopathy at a baseline diabetic retinopathy screening during 2008-2020 was performed. Machine learning and logistic regression prediction models for onset of proliferative diabetic retinopathy, diabetic macular edema, and referable retinopathy detected through routine screening were trained and internally validated. Model performance was assessed using area under the curve (AUC) metrics. The study cohort (N = 276,794) was 51.9% male and 42.1% White. Mean (±SD) age at baseline was 60.0 (±13.1) years. A machine learning XGBoost algorithm was effective in identifying patients who developed proliferative diabetic retinopathy (AUC 0.86; 95% CI, 0.86-0.87), diabetic macular edema (AUC 0.76; 95% CI, 0.75-0.77), and referable retinopathy (AUC 0.78; 95% CI, 0.78-0.79). Similar results were found using a simpler nine-covariate logistic regression model: proliferative diabetic retinopathy (AUC 0.82; 95% CI, 0.80-0.83), diabetic macular edema (AUC 0.73; 95% CI, 0.72-0.74), and referable retinopathy (AUC 0.75; 95% CI, 0.75-0.76). Relatively simple logistic regression models using nine readily available clinical variables can be used to rank order patients for onset of diabetic eye disease and thereby more efficiently prioritize and target screening for at risk patients.

Comparison of Some Prediction Models and their Relevance in the Clinical Research

In healthcare research, predictive modeling is commonly utilized to forecast risk variables and enhance treatment procedures for improved patient outcomes. Enormous quantities of data are being created as a result of recent advances in research, clinical trials, next-generation genomic sequencing, biomarkers, and transcriptional and translational studies. Understanding how to handle and comprehend scientific data to offer better treatment for patients is critical. Currently, multiple prediction models are being utilized to investigate patient outcomes. However, it is critical to recognize the limitations of these models in the research design and their unique benefits and drawbacks. In this overview, we will look at linear regression, logistic regression, decision trees, and artificial neural network prediction models, as well as their advantages and disadvantages. The two most perilous requirements for building any predictive healthcare model are feature selection and model validation. Typically, feature selection is done by a review of the literature and expert opinion on that subject. Model validation is also an essential component of every prediction model. It characteristically relates to the predictive model's performance and accuracy. It is strongly recommended that all clinical parameters should be thoroughly examined before using any prediction model.

Serum TSH Levels are Associated with Hyperactivity Behaviors in Children with Attention Deficit/Hyperactivity Disorder.

Previous studies indicated that maternal thyroid dysfunction increase the offspring's risk for attention deficit/hyperactivity disorder (ADHD). However, the relationship between thyroid function and symptoms in children with ADHD remains unclear. A total of 49 children with ADHD were enrolled. The Conners 3 scale was used to estimate the symptoms associated with ADHD. Correlation between thyroid hormones and the scores of the Conners 3 scale was evaluated by Pearson correlation analysis. Then, ADHD children were divided into two groups according to the hyperactivity index (HI) of the Conners 3 scale: ADHD children with hyperactivity behaviors (HB) (HI > 1.5) and ADHD children without HB (HI < 1.5). The demographic characteristics, thyroid hormones, and routine laboratory parameters between the two groups were collected. To distinguish HI-related factors, a univariate analysis and a binary logistic regression predictive model were used. The discriminative ability of thyroid stimulating hormone (TSH) in predicting ADHD children with HB from ADHD children without HB was investigated using the receiver operating characteristic (ROC) curve method. The levels of TSH were positively correlated to the scores of the Conners 3 scale (r = 0.338, P = 0.033) and HI (r = 0.371, P = 0.019). Moreover, the levels of TSH, serum ferritin, and lactic acid were significantly increased in ADHD children with HB compared to ADHD children without HB (all P < 0.05). Furthermore, the results of binary logistic regression found that TSH (OR 2.243 (CL 1.052-4.783)) and lactic acid (OR 1.018 (CI 1.003-1.032)) were independently associated with HI. Additionally, ROC analysis indicated the potential diagnostic value of TSH in discriminating ADHD children with HB from ADHD children without HB with an AUC of 0.684. These results suggested that the serum TSH levels may be related to the HB in children with ADHD.

A Clinical Decision Tool to Calculate Pretest Probability of Sentinel Lymph Node Metastasis in Primary Cutaneous Melanoma.

Although sentinel lymph node biopsy (SLNB) status is a strong prognostic indicator for cutaneous melanoma, unnecessary SLNBs have substantial cost and morbidity burden. This study was designed to develop, validate, and present a personalized, clinical, decision-making tool using nationally representative data with clinically actionable probability thresholds (Expected Lymphatic Metastasis Outcome [ELMO]). Data from the Surveillance, Epidemiology, and End Results (SEER) Registry from 2000 to 2017 and the National Cancer Database (NCDB) from 2004 to 2015 were used to develop and internally validate a logistic ridge regression predictive model for SLNB positivity. External validation was done with 1568 patients at a large tertiary referral center. The development cohort included 134,809 patients, and the internal validation cohort included 38,518 patients. ELMO (AUC 0.85) resulted in a 29.54% SLNB reduction rate and greater sensitivity in predicting SLNB status for T1b, T2a, and T2b tumors than previous models. In external validation, ELMO had an accuracy of 0.7586 and AUC of 0.7218. Limitations of this study are potential miscoding, unaccounted confounders, and effect modification. ELMO ( https://melanoma-sentinel.herokuapp.com/ ) has been developed and validated (internally and externally) by using the largest publicly available dataset of melanoma patients and was found to have high accuracy compared with other published models and gene expression tests. Individualized risk estimates for SLNB positivity are critical in facilitating thorough decision-making for healthcare providers and patients with melanoma.

Environmental, climatic, and situational factors influencing the probability of fatality or injury occurrence in flash flooding: a rare event logistic regression predictive model.

Flash flooding is considered one of the most lethal natural hazards in the USA as measured by the ratio of fatalities to people affected. However, the occurrence of injuries and fatali- ties during flash flooding was found to be rare (about 2% occurrence rate) based on our analysis of 6,065 flash flood events that occurred in Texas over a 15-year period (2005 to 2019). This article identifies climatic, environmental, and situational factors that affect the occurrence of fatalities and injuries in flash flood events and provides a predictive model to estimate the likelihood of these occurrences. Due to the highly imbalanced dataset, three forms of logit models were investigated to achieve unbiased estimations of the model coef- ficients. The rare event logistic regression (Relogit) model was found to be the most suit- able model. The model considers ten independent situational, climatic, and environmental variables that could affect human safety in flash flood events. Vehicle-related activities dur- ing flash flooding exhibited the greatest effect on the probability of human harm occur- rence, followed by the event's time (daytime vs. nighttime), precipitation amount, location with respect to the flash flood alley, median age of structures in the community, low water crossing density, and event duration. The application of the developed model as a simula- tion tool for informing flash flood mitigation planning was demonstrated in two study cases in Texas.

Induced abortions of women living with HIV in Finland 1987–2019: a national register study

BackgroundRecent data on the rate and risk factors of induced abortion among women living with HIV (WLWH) are limited. Our aim was to use Finnish national health register data to 1) determine the nationwide rate of induced abortions of WLWH in Finland during 1987–2019, 2) compare the rates of induced abortions before and after HIV diagnosis over different time periods, 3) determine the factors associated with terminating a pregnancy after HIV diagnosis, and 4) estimate the prevalence of undiagnosed HIV at induced abortions to see whether routine testing should be implemented.MethodsA retrospective nationwide register study of all WLWH in Finland 1987–2019 (n = 1017). Data from several registers were combined to identify all induced abortions and deliveries of WLWH before and after HIV diagnosis. Factors associated with terminating a pregnancy were assessed with predictive multivariable logistic regression models. The prevalence of undiagnosed HIV at induced abortion was estimated by comparing the induced abortions among WLWH before HIV diagnosis to the number of induced abortions in Finland.ResultsRate of induced abortions among WLWH decreased from 42.8 to 14.7 abortions/1000 follow-up years from 1987–1997 to 2009–2019, more prominently in abortions after HIV diagnosis. After 1997 being diagnosed with HIV was not associated with an increased risk of terminating a pregnancy. Factors associated with induced abortion in pregnancies that began after HIV diagnosis 1998–2019 were being foreign-born (OR 3.09, 95% CI 1.55–6.19), younger age (OR 0.95 per year, 95% CI 0.90–1.00), previous induced abortions (OR 3.36, 95% CI 1.80–6.28), and previous deliveries (OR 2.13, 95% CI 1.08–4.21). Estimated prevalence of undiagnosed HIV at induced abortion was 0.008–0.029%.ConclusionsRate of induced abortions among WLWH has decreased. Family planning should be discussed at every follow-up appointment. Routine testing of HIV at all induced abortions is not cost-effective in Finland due to low prevalence.