A new study put ChatGPT to the test by asking it to judge whether hundreds of scientific hypotheses were true or false—and the results were far from reassuring. While the AI got it right about 80% of the time on the surface, its performance dropped significantly when accounting for random guessing, revealing only modest reasoning ability. Even more concerning, it frequently contradicted itself when asked the exact same question multiple times, sometimes flipping answers back and forth.
Category: Nevin Manimala
J Gerontol A Biol Sci Med Sci. 2026 Mar 17:glag076. doi: 10.1093/gerona/glag076. Online ahead of print.
ABSTRACT
BACKGROUND: Predicting Alzheimer’s disease (AD)-related cognitive impairment (CI) among cognitively normal (CN) adults enables meaningful disease modification through early intervention and enrichment of clinical trials.
METHODS: A deep survival model is trained to predict CI conversion risk in 1,415 CN adults from the National Alzheimer’s Coordinating Center. Converters’ (N = 212) and non-converters’ (N = 1,203) baseline clinical measures and magnetic resonance images are used to estimate their conversion probability up to 22 years after baseline observation.
RESULTS: After 20-fold cross-validation, the model predicts conversion probability with a c-index of 0.88, and classification accuracy of 75% and AUC ROC of 0.89, outperforming previous machine learning models.
CONCLUSIONS: This is one of few studies on the important challenge of predicting future CI among unimpaired subjects. Deep survival modelling can improve the identification of preclinical AD and suggests that uncertainty in AD risk estimation is due to potentially modifiable lifestyle factors.
PMID:41844537 | DOI:10.1093/gerona/glag076
Hematology. 2026 Dec;31(1):2642515. doi: 10.1080/16078454.2026.2642515. Epub 2026 Mar 17.
ABSTRACT
BACKGROUND: Relapsed immune thrombocytopenia (ITP) poses challenges in treatment. Thrombopoietin receptor agonists (TPO-RAs) are recommended as second-line therapy, but the efficacy of monotherapy is limited to about 60%. Our preliminary studies suggest sirolimus is effective for relapsed ITP. This study aimed to evaluate the efficacy and safety of TPO-RA combined with sirolimus in treating relapsed ITP.
METHODS: A retrospective analysis was conducted on 50 patients with relapsed ITP, all of whom received TPO-RAs (hetrombopag and eltrombopag) combined with sirolimus therapy. The primary endpoints were the overall response rate (ORR) at 4, 8, and 12 weeks of treatment. Secondary endpoints included safety of this regimen and subgroup analysis.
RESULTS: ORR at 4, 8, and 12 weeks was 61%, 76%, and 82%, respectively, with a median time to response of 7 days. Mean platelet counts at 2-, 4-, 8-, and 12-weeks post-treatment were 71 ± 12.7 × 109/L, 86 ± 10.9 × 109/L, 128 ± 19.7 × 109/L and 131 ± 17.2 × 109/L, respectively. Whether it is hetrombopag or eltrombopag, when combined with sirolimus, there is no statistically significant difference in efficacy between the two subgroups. TPORAs combined with sirolimus as second-line treatment and as third-line or beyond treatment showed no difference in efficacy. Among ANA-positive patients, the ORR shows no significant difference compared to the ANA-negative group (p > 0.05). Patients with ITP for less than one year are more likely to benefit from this treatment regimen (p = 0.003).
CONCLUSIONS: The combination of TPO-RAs and sirolimus significantly improved platelet counts and sustained response rates. The regimen demonstrated a manageable safety profile, offering a novel therapeutic option for relapsed ITP.
PMID:41844533 | DOI:10.1080/16078454.2026.2642515
J Clin Ultrasound. 2026 Mar 17. doi: 10.1002/jcu.70229. Online ahead of print.
ABSTRACT
BACKGROUND AND OBJECTIVES: Ultrasonography is routinely used after pediatric kidney transplantation, but the prevalence and longitudinal pattern, and prognostic value of abnormal findings are not well defined. We described the frequency and temporal trends of post-transplant ultrasound abnormalities in children and evaluated their association with graft outcomes.
METHODS: We retrospectively included patients (< 18 years) who underwent kidney transplantation at Children’s Medical Center Hospital between 2018 and 2024. Sonographic features (parenchymal changes, peritransplant collections, urinary tract abnormalities, and Doppler indices) were recorded within six predefined follow-up windows. We used descriptive statistics and Kaplan-Meier survival curves stratified by the presence of any abnormal ultrasound finding at each window. Analyses were performed in R (version 4.3).
RESULTS: Sixty-three patients (mean age 12.0 ± 3.9 years) were included. Abnormal findings were present in 43%-56% of scans across follow-up windows. Peritransplant collections (33.9%) and hematomas (24.6%) were common early findings, whereas hydronephrosis and increased echogenicity became more frequent over time. Resistive index elevation (> 0.8) was uncommon (< 3.5%). Three-year graft survival was 55.7%. Any abnormality on baseline ultrasound was associated with lower 36 month survival (40.9% vs. 73.1%, log-rank p = 0.025), while abnormalities detected at later windows were not associated with survival differences.
CONCLUSION: Abnormal ultrasound findings were common after pediatric kidney transplantation, particularly in the early postoperative period. Baseline abnormalities were associated with poorer long-term graft survival, which may reflect perioperative complications or early allograft injury. This supports the clinical value of baseline ultrasound for early risk stratification and guiding the follow-up intensity.
PMID:41844519 | DOI:10.1002/jcu.70229
Clin Oncol (R Coll Radiol). 2026 Jan 27;53:104059. doi: 10.1016/j.clon.2026.104059. Online ahead of print.
ABSTRACT
AIMS: Healthcare systems data (HSD) can potentially improve the efficiency and costs of clinical trials (CTs) and reduce the burden of follow-up. To be used in CTs, HSD need to be accessible, accurate and complete. We compared data from trial-specific case-report forms (CRFs) with relevant HSD to assess whether HSD could replace hospital-based follow-up within an ongoing CT.
MATERIALS AND METHODS: Add-Aspirin (NCT02804815) is a multicentre, randomised, basket trial evaluating aspirin after potentially curative cancer therapy in four tumour types. The primary outcome measure is disease-free survival. HSD obtained from the English National Cancer Registration and Analysis Service (NCRAS) included initial cancer diagnosis and staging, recurrence, specific adverse events, and mortality for English participants during the first 3 years of the trial and was compared to CRF data collected at participating centres. Tabulations and descriptive statistics were used to assess agreement and explore differences between the datasets.
RESULTS: HSD was obtained for 3188/3538 (90%) of English participants. Tumour staging for all four tumour groups generally had good concordance, though this was variable (ranging from 85% (Cohen’s weighted kappa = 0.32) to 99% (kappa = 0.84)) and was lower with neoadjuvant therapy and/or radical radiotherapy. HSD identified all CRF-reported deaths and some further deaths not in the trial database due to loss to follow-up or lag time in data submission. Cancer registry recurrence data captured <50% of recurrences. However, most recurrences could be identified by utilising a combination of NCRAS datasets, including treatment and hospital events. Specific trial-defined adverse events were challenging to identify within HSD.
CONCLUSION: In this example, we found that HSD could not replace CRFs for later trial follow-up as anticipated, because at that time it could not capture all key outcomes accurately . Focussed, hospital-based follow-up was continued for a further 5 years. Selective use of HSD to improve trial efficiency retains potential.
TRIAL REGISTRATION NUMBER: NCT02804815.
PMID:41844499 | DOI:10.1016/j.clon.2026.104059
Gynecol Oncol. 2026 Mar 16;207:66-73. doi: 10.1016/j.ygyno.2026.02.035. Online ahead of print.
ABSTRACT
OBJECTIVE: To compare the long-term survival outcomes of robotic-assisted versus traditional laparoscopic surgery for treatment of stage I endometrial cancer.
METHODS: We performed a retrospective cohort study using the National Cancer Database, including patients with stage I endometrial cancer diagnosed from 2010 to 2017 (with follow-up through 2020) who underwent treatment with either robotic-assisted or traditional laparoscopic surgery. Demographics, clinical characteristics, and outcomes were summarized using descriptive statistics. Overall survival (OS) was compared using a Cox proportional hazard model, adjusting for demographic and clinical factors; hazard ratios and 95% confidence intervals reported.
RESULTS: A total of 127,342 patients with stage I endometrial cancer who underwent minimally invasive surgery were included. The median follow-up among those still known to be alive was 69.7 months. Almost three-quarters (74.5%) received robotic-assisted surgery, while 25.5% received traditional laparoscopic surgery. Demographic and clinical factors were similar between groups. No differences in OS were observed by surgery modality (5-year OS: 91.7% for robotic vs. 91.4% for traditional laparoscopic surgery) and this remained after adjustment for demographic and clinical factors (HR = 1.00, 95% CI 0.96-1.04).
CONCLUSIONS: There was no difference in OS for patients with stage I endometrial cancer treated with robotic-assisted versus traditional laparoscopic surgery. These data demonstrate excellent survival outcomes in this cohort and do not favor a robotic-assisted or traditional laparoscopic approach. However, the full extent of a route-of-surgery decision may not be detected given the potential for late recurrences and deaths, highlighting the need for large, prospective studies with long duration of follow-up.
PMID:41844495 | DOI:10.1016/j.ygyno.2026.02.035
JACC Adv. 2026 Mar 16;5(4):102653. doi: 10.1016/j.jacadv.2026.102653. Online ahead of print.
ABSTRACT
BACKGROUND: Peripartum cardiomyopathy (PPCM) is a leading cause of heart failure in pregnancy and contributes significantly to maternal morbidity and mortality. Black women are disproportionately affected and experience worse outcomes compared with other groups.
OBJECTIVES: This study aimed to quantify differences in risk factors and outcomes between Black and White women in the United States diagnosed with PPCM.
METHODS: We conducted a systematic review and meta-analysis of observational studies published after 2002 including U.S. women with PPCM and race-stratified risk factors and outcomes. Investigated outcomes included mortality, major adverse cardiac events, and recovery of left ventricular ejection fraction. Random-effects meta-analysis estimated the pooled prevalence of risk factors and outcomes. Logistic regression, forest plots, and I2 statistics were utilized for analysis.
RESULTS: Compared with controls, cases had higher rates of obesity, preeclampsia, hypertension, diabetes, multiple gestations, and tobacco use. Compared to White cases, Black cases had higher prevalence of diabetes (14% vs 5%; P = 0.027) and utilization of public payer (72% vs 30%; P < 0.001). At presentation, mean left ventricular ejection fraction was 26% in Black women and 29% in White women. White women experienced higher rates of recovery in ejection fraction (63% vs 40%; P < 0.0001). Mortality rates were higher among Black women (8% vs 2%; P = 0.013).
CONCLUSIONS: Black women with PPCM experienced lower recovery and higher mortality rates compared with White women. With the exception of a significant difference in payer status, modest differences in previously identified risk factors were observed between racial groups to account for worse outcomes (Disparities in risk factors and outcomes between Black and White US women with peripartum cardiomyopathy: A systematic review and meta-analysis; CRD42023439228).
PMID:41844488 | DOI:10.1016/j.jacadv.2026.102653
Maturitas. 2026 Feb 25;208:108895. doi: 10.1016/j.maturitas.2026.108895. Online ahead of print.
ABSTRACT
OBJECTIVES: To investigate the controversial association between exogenous hormone use (EHU) and dementia, with a focus on subtype-specific risks.
STUDY DESIGN: This prospective cohort study followed 273,069 women in the UK Biobank over 3,802,608 person-years, identifying 4,710 dementia cases.
MAIN OUTCOME MEASURES: Cox models assessed use of oral contraceptive (OC) and hormone replacement therapy (HRT) in relation to all-cause dementia, Alzheimer’s disease (AD), vascular dementia (VaD), and frontotemporal dementia (FTD) across treatment durations. Subgroup analyses were stratified by age, ethnicity, APOE status, education, income, and reproductive factors. A systematic review was conducted to synthesize existing evidence.
RESULTS: In the cohort study, OC use was associated with reduced risks of all-cause dementia (HR 0.90, 95%CI 0.84-0.95), AD (HR 0.87, 95%CI 0.79-0.95), and VaD (HR 0.81, 95%CI 0.70-0.93), particularly after 4-14 years of use. HRT showed no significant association with increased dementia risk. Synthesized results largely corroborated these findings: OC use was associated with reduced risks of dementia (HR 0.90, 95%CI 0.89-0.92); and although four European studies reported a moderately increased AD risk after post-menopausal HRT use, neither cohort-based studies (HR 0.98, 95%CI 0.90-1.06) nor traditional case-control studies (OR 1.00, 95%CI 0.90-1.11) found an association between HRT and dementia.
CONCLUSIONS: Our combined evidence does not support an increased risk of dementia associated with OC use; similarly, no clear association was observed between HRT and increased dementia risk. Clinical decisions on EHU should be individualized, balancing overall benefits against potential risks.
PMID:41844465 | DOI:10.1016/j.maturitas.2026.108895
J Trace Elem Med Biol. 2026 Mar 11;95:127858. doi: 10.1016/j.jtemb.2026.127858. Online ahead of print.
ABSTRACT
BACKGROUND: Endocrine disrupting metals such as cadmium (Cd), lead (Pb) and mercury (Hg) may have significant influence on fertility issues among otherwise healthy women.
METHODS: In this study, fifty women with preserved ovarian function, undergoing in vitro fertilization (IVF) procedure were tested for heavy metals (Cd, Hg and Pb) in their morning urine samples using inductively coupled plasma mass spectrometry after microwave digestion.
RESULTS: None of the women had measurable lead in the urine samples while cadmium and mercury were quantified in 28% and 60% of the samples, respectively. Serum LH levels were significantly lower (p = 0.046) among women with measurable urinary cadmium values compared to those with cadmium below the limit of quantification (LOQ). The number of immature oocytes retrieved during transvaginal aspiration was significantly higher (p = 0.030) in women with mercury above LOQ in comparison to those below LOQ. The elevation of urinary mercury concentrations was followed by the increment AST and ALT (p = 0.035 and p = 0.001, respectively). Moreover, higher urinary mercury levels were positively associated with fasting and 2 h post-load serum glucose levels during oral glucose tolerance testing (p = 0.035 and p = 0.002, respectively). For the first time, statistically significant (p = 0.009) correlation was obtained between the urinary mercury concentration and serum LDH values. All the associations between mercury and the observed parameters remained significant after adjustments for age, BMI and time attempting to conceive were made.
CONCLUSIONS: Women undergoing IVF have considerable amount of cadmium and mercury in the urine samples, despite the absence of occupational exposure.
PMID:41844457 | DOI:10.1016/j.jtemb.2026.127858
Psychiatry Res. 2026 Mar 9;360:117082. doi: 10.1016/j.psychres.2026.117082. Online ahead of print.
ABSTRACT
OBJECTIVE: Mental health in adolescents has been a growing problem in recent decades. However, comprehensive discussion of the research on the distal and proximal factors related to adolescent mental health were limited and the evidence was lacking. This study aimed to explore the result of adverse childhood experiences and biological rhythms for depressive and anxiety symptoms in adolescents.
METHODS: The data used in this study were obtained from China in 2021 (as baseline) and 2023 (as follow-up). Participants provided self-reported information through an anonymous questionnaire. Logistic regression, cluster analysis, restriction cubic spline plots and receiver operating characteristic curves were used for statistical analysis.
RESULTS: The study involved 2103 teenagers (mean age 15 years old), consisting of 1089 females and 1014 males. Multiple logistic regression results based on this subgroup showed that higher scores on either ACEs or biorhythms were associated with a greater risk of concurrent depressive-anxiety symptoms at baseline and follow-up, the adjusted ORs (95% CI) were 14.46 (7.14-29.26), 12.09 (5.83-22.08), respectively. The area under the receiver operating characteristic curve of the joint predictive effect of the ACEs and biorhythms were 0.664 for depressive symptoms, 0.656 for anxiety symptoms, and 0.667 for co-morbidity.
CONCLUSION: These findings emphasize the necessity of balancing past risk factors with emerging risk factors (biorhythms) in the precaution and management of adolescent depression and anxiety symptoms problems.
PMID:41844454 | DOI:10.1016/j.psychres.2026.117082