Category: Statistics

Int J Biostat. 2025 Nov 3. doi: 10.1515/ijb-2025-0016. Online ahead of print.

ABSTRACT

The binary classification problem (BCP) aims to correctly allocate subjects in one of two possible groups. The groups are frequently defined as having or not one characteristic of interest. With this goal, we are allowed to use different types of information. There is a huge number of methods dealing with this problem; including standard binary regression models, or complex machine learning techniques such as support vector machine, boosting, or perceptron, among others. When this information is summarized in a continuous score, we have to define classification regions (or subsets) which will determine whether the subjects are classified as positive, with the characteristic under study, or as negative, otherwise. The standard (or regular) receiver-operating characteristic (ROC) curve assumes that higher values of the marker are associated with higher probabilities of being positive and considers as positive those patients with values within the intervals [c, ∞) $\left(c\in R\right)$ , and plots the true- against the false- positive rates (sensitivity against one minus specificity) for all potential c. The so-called generalized ROC curve, gROC, allows that both higher and lower values of the score are associated with higher probabilities of being positive. The efficient ROC curve, eROC, considers the best ROC curve based on a transformation of the score. In this manuscript, we are interested in studying, comparing and approximating the transformations leading to the eROC and to the gROC curves. We will prove that, when the optimal transformation does not have relative maximum, both curves are equivalent. Besides, we investigate the use of the gROC curve on some theoretical models, explore the relationship between the gROC and the eROC curves, and propose two non-parametric procedures for approximating the transformation leading to the gROC curve. The finite-sample behavior of the proposed estimators is explored through Monte Carlo simulations. Two real-data sets illustrate the practical use of the proposed methods.

PMID:41174954 | DOI:10.1515/ijb-2025-0016

Psychiatr Serv. 2025 Nov 1;76(11):944-950. doi: 10.1176/appi.ps.20250086. Epub 2025 Aug 26.

ABSTRACT

OBJECTIVE: This study aimed to evaluate whether three popular chatbots powered by large language models (LLMs)-ChatGPT, Claude, and Gemini-provided direct responses to suicide-related queries and how these responses aligned with clinician-determined risk levels for each question.

METHODS: Thirteen clinical experts categorized 30 hypothetical suicide-related queries into five levels of self-harm risk: very high, high, medium, low, and very low. Each LLM-based chatbot responded to each query 100 times (N=9,000 total responses). Responses were coded as “direct” (answering the query) or “indirect” (e.g., declining to answer or referring to a hotline). Mixed-effects logistic regression was used to assess the relationship between question risk level and the likelihood of a direct response.

RESULTS: ChatGPT and Claude provided direct responses to very-low-risk queries 100% of the time, and all three chatbots did not provide direct responses to any very-high-risk query. LLM-based chatbots did not meaningfully distinguish intermediate risk levels. Compared with very-low-risk queries, the odds of a direct response were not statistically different for low-risk, medium-risk, or high-risk queries. Across models, Claude was more likely (adjusted odds ratio [AOR]=2.01, 95% CI=1.71-2.37, p<0.001) and Gemini less likely (AOR=0.09, 95% CI=0.08-0.11, p<0.001) than ChatGPT to provide direct responses.

CONCLUSIONS: LLM-based chatbots’ responses to queries aligned with experts’ judgment about whether to respond to queries at the extremes of suicide risk (very low and very high), but the chatbots showed inconsistency in addressing intermediate-risk queries, underscoring the need to further refine LLMs.

PMID:41174947 | DOI:10.1176/appi.ps.20250086

Psychiatr Serv. 2025 Nov 1;76(11):988-996. doi: 10.1176/appi.ps.20250042.

ABSTRACT

OBJECTIVE: People with serious mental illness (i.e., disabling psychotic, mood, and other disorders) develop chronic medical diseases early in life. This study aimed to examine the effects of integrating primary care into community mental health centers (CMHCs; reverse integrated care) on service use among young adults with serious mental illness who may benefit from early intervention.

METHODS: This retrospective cohort analysis used Medicaid claims of 945 people with serious mental illness (ages 18-40) in CMHC care from 2020 to 2022-315 in reverse integrated care and 630 propensity score matched participants in comparison care (i.e., not reverse integrated care). Logistic regression, adjusted for participant characteristics, enrollment quarter, and past service use, assessed outcomes in the 6 months after enrollment.

RESULTS: Participants’ mean±SD age was 32.56 ± 7.84 years; 29% had a diagnosis of schizophrenia, 40% had a co-occurring substance use disorder, 33% had a medical emergency department (ED) visit in the 6 months before enrollment, and all were enrolled in CMHC care at baseline. During follow-up, participants in reverse integrated care were more likely to have an outpatient medical visit (65% vs. 58%; adjusted odds ratio [AOR]=1.54, p=0.005) and were less likely to have a medical ED visit (26% vs. 33%; AOR=0.70, p=0.035) than those in comparison care.

CONCLUSIONS: Integrating primary care into CMHC services may increase access to outpatient medical care and reduce ED visits for medical reasons among young adults with serious mental illness. Future research should confirm these findings, assess longer-term outcomes, and examine implementation facilitators and barriers.

PMID:41174946 | DOI:10.1176/appi.ps.20250042

Psychiatr Serv. 2025 Nov 1;76(11):1027-1030. doi: 10.1176/appi.ps.20240487.

ABSTRACT

OBJECTIVE: Youths within the child welfare system have high rates of mental health needs and chronic barriers to service access. The cascade-of-care approach was used to explore this population’s use of mental health services.

METHODS: The mental health cascade was used with electronic medical record data to characterize service access among 97 youths served in a child welfare clinic.

RESULTS: Of the total sample, 82% of youths met criteria to identify those needing mental health services, 56% were referred for assessment and services, 33% completed an assessment, 24% had a treatment plan signed, and 17% received services. Time between steps averaged 64.52 days from screening to referral, 65.94 days from referral to assessment, 7.05 days from assessment to signed treatment plan, and 19.84 days from treatment plan to service initiation.

CONCLUSIONS: Significant gaps in the care cascade occurred, especially at earlier stages. Multilayered efforts to reduce service gaps in this population are needed.

PMID:41174942 | DOI:10.1176/appi.ps.20240487

Int J Gynaecol Obstet. 2025 Nov 1. doi: 10.1002/ijgo.70633. Online ahead of print.

ABSTRACT

OBJECTIVE: Our study aimed to evaluate the combined risk of macrosomia and nulliparity. We investigated whether macrosomia is independently associated with an increased rate of intrapartum cesarean delivery (CD) and adverse maternal and neonatal outcomes among nulliparous women delivering at term.

METHODS: We conducted a retrospective cohort study including nulliparous women with singleton, term (37-42 weeks) deliveries between 2005 and 2024 at two university-affiliated medical centers in Jerusalem, Israel. Women who delivered macrosomic neonates (birth weight ≥4000 g) were compared with those delivering neonates weighing 3000-3500 g. Exclusions included multifetal gestations, preterm deliveries, elective cesareans, fetal anomalies, and antepartum demise. The primary outcome was intrapartum cesarean delivery. Secondary outcomes included a range of maternal and neonatal complications. Multivariable logistic regression was used to adjust for potential confounders including maternal obesity, pre-gestational diabetes, labor induction, and gestational age at delivery.

RESULTS: Among 86 801 eligible nulliparous women, 2762 (3.2%) delivered macrosomic neonates and 40 963 (47.2%) served as the control group. The rate of intrapartum CD was significantly higher among macrosomic deliveries compared to controls (18.6% vs. 7.8%, P < 0.01), though this association was not statistically significant after adjustment (adjusted odds ratio [aOR] 1.81, 95% confidence interval [CI]: 0.91-3.58). Macrosomia was independently associated with increased odds of shoulder dystocia (aOR 33.42, 95% CI: 11.32-98.68), postpartum hemorrhage (aOR 2.13, 95% CI: 1.79-2.54), blood transfusion (aOR 2.45, 95% CI: 1.51-3.98), chorioamnionitis (aOR 2.03, 95% CI: 1.57-2.61), neonatal intensive care unit admission (aOR 1.62, 95% CI: 1.15-2.29), neonatal hypoglycemia (aOR 2.23, 95% CI: 1.32-3.77), and Erb’s palsy or clavicular fracture (aOR 9.43, 95% CI: 4.01-22.21). Stratification by birth weight categories revealed a dose-response relationship, with the highest complication rates among neonates >4500 g.

CONCLUSION: In nulliparous women delivering at term, macrosomia is independently associated with a higher risk of multiple adverse maternal and neonatal outcomes. These findings underscore the compounded risk faced by nulliparous women with macrosomic fetuses and highlight the need for enhanced prenatal surveillance and individualized delivery planning in this population.

PMID:41174933 | DOI:10.1002/ijgo.70633

J Diabetes Sci Technol. 2025 Nov;19(6):1457-1463. doi: 10.1177/19322968251371046. Epub 2025 Oct 14.

ABSTRACT

BACKGROUND: Type 1 diabetes is treated with exogenous insulin using multiple daily injections or insulin pumps. However, both strategies require carbohydrate counting for prandial insulin dosing, which is both burdensome and error prone.

METHODS: We conducted a pilot, randomized, controlled study to eliminate carbohydrate counting in adults (n = 12, 7 females, age 39.5 [15.1], HbA1c 7.4% [0.6]) using an automated insulin and pramlintide fully closed-loop system. The interventions included five arms during which participants underwent 14 hours of outpatient, free-living, supervised experiments of (1) faster aspart with carbohydrate counting (control), faster aspart and pramlintide without carbohydrate counting at (2) 8 µg/U and (3) 10 µg/U ratios, and aspart and pramlintide without carbohydrate counting at (4) 8 µg/U and (5) 10 µg/U ratios.

RESULTS: The median time in target range (3.9-10.0 mmol/L) with the control arm was 78.6 [65.3-92.9], compared with 76.2 [64.6-86.9] and 78.8 [68.8-86.0] with the fully closed-loop faster aspart and pramlintide systems at 8 and 10 µg/U ratios, respectively, and compared with 65.9 [59.9-83.6] and 77.4 [72.1-82.7] with the fully closed-loop aspart and pramlintide systems at 8 and 10 µg/U ratios, respectively. Times spent below 3.9 and 3.0 mmol/L were numerically higher with the fully closed-loop aspart and pramlintide systems than the control arm. None of the differences were statistically significant.

CONCLUSIONS: This study suggests that automated insulin and pramlintide systems have the potential to alleviate carbohydrate counting without degrading time in range. A longer and larger study is underway.

PMID:41174925 | DOI:10.1177/19322968251371046

Int J Phytoremediation. 2025 Oct 31:1-12. doi: 10.1080/15226514.2025.2579150. Online ahead of print.

ABSTRACT

The application of engineered humic Fe/Zn complexes in remediating lead (Pb) biotoxicity, as well as their impact on CO₂ efflux, soil carbon distribution, and spinach growth, remains unexplored. This study revealed the impact of engineered Fe- and Zn-enriched humate complexes on Pb immobilization, geochemical fractionation, and translocation in a spinach crop, and the effects on soil biochemical health and CO₂-C efflux from tillage and no-tillage Pb-contaminated soils in closed chambers. Advanced statistical models like PLS-PM were employed to determine the direct and total effects of the applied amendments under different tillage regimes. Results demonstrated that Zn- and Fe-humate applications decreased Pb contents by 48% and 72% in tillage soil, while under no-tillage soil Pb concentration was reduced by 35%; 8.7 mg kg^-1 (Zn-humate) vs 33%; 5.9 mg kg^-1 (Fe-humate) compared to respective controls. Soil C distribution showed dramatically varying trends in tilled and no-tilled soils. Soil extracellular enzyme activity was enhanced under both tillage and no-tillage operations with Zn- and Fe-humate complexes. Human-associated health risk was reduced by 3-fold by the application of Fe-humate, and it was 2-fold with Zn-humate. Overall, findings revealed that the tillage-driven application of Fe/Zn humate complexes significantly enhanced spinach growth, reducing Pb bioavailability, highlighting their potential for sustainable soil management in contaminated farmlands.

PMID:41174889 | DOI:10.1080/15226514.2025.2579150

Am J Biol Anthropol. 2025 Nov;188(3):e70146. doi: 10.1002/ajpa.70146.

ABSTRACT

OBJECTIVES: During the Paleolithic-Neolithic transition, modern human femoral diaphyses underwent significant structural changes, primarily driven by shifts in subsistence patterns including decreased mobility and increased sedentism. However, femoral remains from East Asia during this period are inadequately reported and studied. This study investigates the femoral diaphyseal structures across East Asia during this transition, exploring their variation, evolutionary processes, and links to subsistence patterns reflected in the archeological record.

MATERIALS AND METHODS: Human femora from Qihe Cave, Donghulin, and Taipinghu, representing South, North, and Northeast China during the transition, were analyzed. Midshaft cross-sectional shapes were compared with Early Upper Paleolithic (EUP), Late Upper Paleolithic (LUP), and recent sedentary agricultural (RSA) samples. Morphometric maps illustrating cortical bone thickness, external radius, and bending rigidity along the entire diaphysis were compared with Late Pleistocene early modern humans from South and North China and RSA specimens.

RESULTS: Analysis of midshaft cross-sectional shapes revealed that DHL 4 and Qihe M2 align with the LUP group, whereas DHL M1 and TPH 45 show close affinities with the RSA group. Statistical analyses based on morphometric maps further reveal that DHL 4 and Qihe M2 share key features with Late Pleistocene early modern humans, whereas DHL M1 and TPH 45 fall within the RSA variation range.

DISCUSSION: Two distinct femoral diaphyseal patterns are identified among East Asian modern humans during the transition, reflecting regional variations and intrapopulation divisions of labor, primarily associated with hunting and gathering strategies shaped by local environmental conditions and corresponding archeological cultures.

PMID:41174884 | DOI:10.1002/ajpa.70146

Head Neck. 2025 Oct 31. doi: 10.1002/hed.70050. Online ahead of print.

ABSTRACT

OBJECTIVE: High-frequency ultrasound was used to observe and measure the width of the normal adult glottal-level paraglottic space (PGS).

METHODS: One thousand healthy Han Chinese adults (694 women; 18-79 years) were enrolled between October 2023 and April 2024. PGS ultrasound characteristics were observed, and the width was measured at the midpoint of the anterior, middle, and posterior 1/3 points. The data were stratified by age group and sex. Spearman correlation analysis was used to analyze the correlations between PGS width and sex, age, height, weight, BMI, and BSA.

RESULTS: The glottic-level PGS is a triangular hyperechoic structure that is narrow in the front and wide in the back of the parathyroid cartilage section. The PGS width reference ranges (5th and 95th percentiles) were as follows: Male: W₁ (0.45-1.16) mm, W₂ (0.55-1.71) mm, W₃ (1.30-2.35) mm; Female: W₁ (0.55-0.95) mm, W₂ (0.90-1.35) mm, W₃ (1.55-2.00) mm. There were no statistically significant differences in W₁, W₂, and W₃ between males and females or between different age groups (all p > 0.05).

CONCLUSIONS: This study summarized normal adult PGS ultrasound characteristics at the glottal level and established references for PGS width, which can provide reference values for clinical practice.

PMID:41174874 | DOI:10.1002/hed.70050

Health Informatics J. 2025 Oct-Dec;31(4):14604582251383804. doi: 10.1177/14604582251383804. Epub 2025 Oct 31.

ABSTRACT

Objectives: Digital patient-reported outcome (PRO) tools, though beneficial for managing inflammatory bowel disease (IBD), remain underutilized in Australia. This study aimed to investigate a group of Australian patients’ readiness to engage with digital PRO tools and identify potential barriers to their implementation. Methods: We assessed 58 patients from a tertiary IBD clinic in Melbourne, Australia, using the Readiness and Enablement Index for Health Technology (ReadHy) tool, and compared the results to those from a Danish study. Results: Compared to the Danish cohort, our patients were younger with more frequent users of electronic devices, showed higher readiness across most ReadHy dimensions, except in the “heiQ8 Emotional Distress” dimension. Conclusion: These findings suggest a generally favourable environment for implementing digital PRO tools at an Australian tertiary IBD clinic, though attention should be paid to emotional well-being to improve adoption. This study also provides a framework for other centres to evaluate their patients’ readiness for digital PRO engagement.

PMID:41174870 | DOI:10.1177/14604582251383804