Interactive calculator to estimate insulin sensitivity in type 1 diabetes

The gold standard for measuring insulin sensitivity (IS) is the hyperinsulinemic-euglycemic clamp, a time, costly, and labor-intensive research tool. A low insulin sensitivity is associated with a complication-risk in type 1 diabetes. Various formulae using clinical data have been developed and correlated with measured IS in type 1 diabetes. We consolidated multiple formulae into an online calculator (bit.ly/estimated-GDR), enabling comparison of IS and its probability of IS <4.45 mg/kg/min (low) or >6.50 mg/kg/min (high), as measured in a validation set of clamps in 104 adults with type 1 diabetes. Insulin sensitivity calculations using different formulae varied significantly, with correlations (R2 ) ranging 0.005-0.87 with agreement in detecting low and high glucose disposal rates in the range 49-93% and 89-100%, respectively. We demonstrate that although the calculated IS varies between formulae, their interpretation remains consistent. Our free online calculator offers a user-friendly tool for individual IS calculations and also offers efficient batch processing of data for research.

Development and internal validation of machine learning models for personalized survival predictions in spinal cord glioma patients

BACKGROUND CONTEXT

Numerous factors have been associated with the survival outcomes in patients with spinal cord gliomas (SCG). Recognizing these specific determinants is crucial, yet it is also vital to establish a reliable and precise prognostic model for estimating individual survival outcomes.

OBJECTIVE

The objectives of this study are twofold: first, to create an array of interpretable machine learning (ML) models developed for predicting survival outcomes among SCG patients; and second, to integrate these models into an easily navigable online calculator to showcase their prospective clinical applicability.

STUDY DESIGN

This was a retrospective, population-based cohort study aiming to predict the outcomes of interest, which were binary categorical variables, in SCG patients with ML models.

PATIENT SAMPLE

The National Cancer Database (NCDB) was utilized to identify adults aged 18 years or older who were diagnosed with histologically confirmed SCGs between 2010 and 2019.

OUTCOME MEASURES

The outcomes of interest were survival outcomes at three specific time points post-diagnosis: 1, 3, and 5 years. These outcomes were formed by combining the "Vital Status" and "Last Contact or Death (Months from Diagnosis)" variables. Model performance was evaluated visually and numerically. The visual evaluation utilized receiver operating characteristic (ROC) curves, precision-recall curves (PRCs), and calibration curves. The numerical evaluation involved metrics such as sensitivity, specificity, accuracy, area under the PRC (AUPRC), area under the ROC curve (AUROC), and Brier Score.

METHODS

We employed five ML algorithms-TabPFN, CatBoost, XGBoost, LightGBM, and Random Forest-along with the Optuna library for hyperparameter optimization. The models that yielded the highest AUROC values were chosen for integration into the online calculator. To enhance the explicability of our models, we utilized SHapley Additive exPlanations (SHAP) for assessing the relative significance of predictor variables and incorporated partial dependence plots (PDPs) to delineate the influence of singular variables on the predictions made by the top performing models.

RESULTS

For the 1-year survival analysis, 4,913 patients [5.6% with 1-year mortality]; for the 3-year survival analysis, 4,027 patients (11.5% with 3-year mortality]; and for the 5-year survival analysis, 2,854 patients (20.4% with 5-year mortality) were included. The top models achieved AUROCs of 0.938 for 1-year mortality (TabPFN), 0.907 for 3-year mortality (LightGBM), and 0.902 for 5-year mortality (Random Forest). Global SHAP analyses across survival outcomes at different time points identified histology, tumor grade, age, surgery, radiotherapy, and tumor size as the most significant predictor variables for the top-performing models.

CONCLUSIONS

This study demonstrates ML techniques can develop highly accurate prognostic models for SCG patients with excellent discriminatory ability. The interactive online calculator provides a tool for assessment by physicians (https://huggingface.co/spaces/MSHS-Neurosurgery-Research/NCDB-SCG). Local interpretability informs prediction influences for a given individual. External validation across diverse datasets could further substantiate potential utility and generalizability. This robust, interpretable methodology aligns with the goals of precision medicine, establishing a foundation for continued research leveraging ML's predictive power to enhance patient counseling.

Determinants and reference values of the 6-min walk distance in the general population-results of the population-based STAAB cohort study

AIMS

The 6-min walk test is an inexpensive, safe, and easy tool to assess functional capacity in patients with cardiopulmonary diseases including heart failure (HF). There is a lack of reference values, which are a prerequisite for the interpretation of test results in patients. Furthermore, determinants independent of the respective disease need to be considered when interpreting the 6-min walk distance (6MWD).

METHODS

The prospective Characteristics and Course of Heart Failure Stages A-B and Determinants of Progression (STAAB) cohort study investigates a representative sample of residents of the City of Würzburg, Germany, aged 30 to 79 years, without a history of HF. Participants underwent detailed clinical and echocardiographic phenotyping as well as a standardized assessment of the 6MWD using a 15-m hallway.

RESULTS

In a sample of 2762 participants (51% women, mean age 58 ± 11 years), we identified age and height, but not sex, as determinants of the 6MWD. While a worse metabolic profile showed a negative association with the 6MWD, a better systolic and diastolic function showed a positive association with 6MWD. From a subgroup of 681 individuals without any cardiovascular risk factors (60% women, mean age 52 ± 10 years), we computed age- and height-specific reference percentiles.

CONCLUSION

In a representative sample of the general population free from HF, we identified determinants of the 6MWD implying objective physical fitness associated with metabolic health as well as with cardiac structure and function. Furthermore, we derived reference percentiles applicable when using a 15-m hallway.

Establishing normative data for the evaluation of cognitive performance in Huntington's disease considering the impact of gender, age, language, and education

BACKGROUND

A declining cognitive performance is a hallmark of Huntington's disease (HD). The neuropsychological battery of the Unified HD Rating Scale (UHDRS'99) is commonly used for assessing cognition. However, there is a need to identify and minimize the impact of confounding factors, such as language, gender, age, and education level on cognitive decline.

OBJECTIVES

Aim is to provide appropriate, normative data to allow clinicians to identify disease-associated cognitive decline in diverse HD populations by compensating for the impact of confounding factors METHODS: Sample data, N = 3267 (60.5% females; mean age of 46.9 years (SD = 14.61, range 18-86) of healthy controls were used to create a normative dataset. For each neuropsychological test, a Bayesian generalized additive model with age, education, gender, and language as predictors was constructed to appropriately stratify the normative dataset.

RESULTS

With advancing age, there was a non-linear decline in cognitive performance. In addition, performance was dependent on educational levels and language in all tests. Gender had a more limited impact. Standardized scores have been calculated to ease the interpretation of an individual's test outcome. A web-based online tool has been created to provide free access to normative data.

CONCLUSION

For defined neuropsychological tests, the impact of gender, age, education, and language as factors confounding disease-associated cognitive decline can be minimized at the level of a single patient examination.

Development and Validation of a Machine-Learning Model to Predict Early Recurrence of Intrahepatic Cholangiocarcinoma

BACKGROUND

The high incidence of early recurrence after hepatectomy for intrahepatic cholangiocarcinoma (ICC) has a detrimental effect on overall survival (OS). Machine-learning models may improve the accuracy of outcome prediction for malignancies.

METHODS

Patients who underwent curative-intent hepatectomy for ICC were identified using an international database. Three machine-learning models were trained to predict early recurrence (< 12 months after hepatectomy) using 14 clinicopathologic characteristics. The area under the receiver operating curve (AUC) was used to assess their discrimination ability.

RESULTS

In this study, 536 patients were randomly assigned to training (n = 376, 70.1%) and testing (n = 160, 29.9%) cohorts. Overall, 270 (50.4%) patients experienced early recurrence (training: n = 150 [50.3%] vs testing: n = 81 [50.6%]), with a median tumor burden score (TBS) of 5.6 (training: 5.8 [interquartile range {IQR}, 4.1-8.1] vs testing: 5.5 [IQR, 3.7-7.9]) and metastatic/undetermined nodes (N1/NX) in the majority of the patients (training: n = 282 [75.0%] vs testing n = 118 [73.8%]). Among the three different machine-learning algorithms, random forest (RF) demonstrated the highest discrimination in the training/testing cohorts (RF [AUC, 0.904/0.779] vs support vector machine [AUC, 0.671/0.746] vs logistic regression [AUC, 0.668/0.745]). The five most influential variables in the final model were TBS, perineural invasion, microvascular invasion, CA 19-9 lower than 200 U/mL, and N1/NX disease. The RF model successfully stratified OS relative to the risk of early recurrence.

CONCLUSIONS

Machine-learning prediction of early recurrence after ICC resection may inform tailored counseling, treatment, and recommendations. An easy-to-use calculator based on the RF model was developed and made available online.

Development and validation of an online calculator to predict the pathological nature of colorectal tumors

BACKGROUND

No single endoscopic feature can reliably predict the pathological nature of colorectal tumors (CRTs).

AIM

To establish and validate a simple online calculator to predict the pathological nature of CRTs based on white-light endoscopy.

METHODS

This was a single-center study. During the identification stage, 530 consecutive patients with CRTs were enrolled from January 2015 to December 2021 as the derivation group. Logistic regression analysis was performed. A novel online calculator to predict the pathological nature of CRTs based on white-light images was established and verified internally. During the validation stage, two series of 110 images obtained using white-light endoscopy were distributed to 10 endoscopists [five highly experienced endoscopists and five less experienced endoscopists (LEEs)] for external validation before and after systematic training.

RESULTS

A total of 750 patients were included, with an average age of 63.6 ± 10.4 years. Early colorectal cancer (ECRC) was detected in 351 (46.8%) patients. Tumor size, left semicolon site, rectal site, acanthosis, depression and an uneven surface were independent risk factors for ECRC. The C-index of the ECRC calculator prediction model was 0.906 (P = 0.225, Hosmer–Lemeshow test). For the LEEs, significant improvement was made in the sensitivity, specificity and accuracy (57.6% vs 75.5%; 72.3% vs 82.4%; 64.2% vs 80.2%; P < 0.05), respectively, after training with the ECRC online calculator prediction model.

CONCLUSION

A novel online calculator including tumor size, location, acanthosis, depression, and uneven surface can accurately predict the pathological nature of ECRC.

A web-based interface to calculate phonological neighborhood density for words and nonwords in Modern Standard Arabic

The availability of online databases (e.g., Balota et al., 2007) and calculators (e.g., Storkel & Hoover, 2010) has contributed to an increase in psycholinguistic-related research, to the development of evidence-based treatments in clinical settings, and to scientifically supported training programs in the language classroom. The benefit of online language resources is limited by the fact that the majority of such resources provide information only for the English language (Vitevitch, Chan & Goldstein, 2014). To address the lack of diversity in these resources for languages that differ phonologically and morphologically from English, the present article describes an online database to compute phonological neighborhood density (i.e., the number of words that sound similar to a given word) for words and nonwords in Modern Standard Arabic (MSA). A full description of how the calculator can be used is provided. It can be freely accessed at https://calculator.ku.edu/density/about .

A Web-based interface to calculate phonotactic probability for words and nonwords in Modern Standard Arabic

A number of databases (Storkel Behavior Research Methods, 45, 1159-1167, 2013) and online calculators (Vitevitch & Luce Behavior Research Methods, Instruments, and Computers, 36, 481-487, 2004) have been developed to provide statistical information about various aspects of language, and these have proven to be invaluable assets to researchers, clinicians, and instructors in the language sciences. The number of such resources for English is quite large and continues to grow, whereas the number of such resources for other languages is much smaller. This article describes the development of a Web-based interface to calculate phonotactic probability in Modern Standard Arabic (MSA). A full description of how the calculator can be used is provided. It can be freely accessed at http://phonotactic.drupal.ku.edu/ .