Explore the vast range of titles available.

This evaluation of 36,000 clinical vignettes found that next-generation reasoning large language models, o3-mini and DeepSeek-R1, frequently perpetuate racial and gender stereotypes for common medical conditions, indicating that advancements in reasoning do not inherently improve representational fairness.

Background Multiple studies have indicated that patients with high body mass index (BMI) may have favourable survival outcomes following treatment with an immune checkpoint inhibitor (ICI). However, this evidence is limited by several factors, notably the minimal evidence from randomised controlled trials (RCTs), the use of categorised BMI with inconsistent cut point definitions, and minimal investigation of contemporary combination ICI therapy. Moreover, whether overweight and obese patients gain a larger benefit from contemporary frontline chemoimmunotherapy in non-small cell lung cancer (NSCLC) is unclear. Methods This secondary analysis pooled individual patient data from the intention-to-treat population of the IMpower130 and IMpower150 RCTs comparing chemoimmunotherapy versus chemotherapy. Co-primary outcomes were overall survival (OS) and progression-free survival (PFS). The potentially non-linear relationship between BMI and chemoimmunotherapy treatment effect was evaluated using Multivariable Fractional Polynomial Interaction (MFPI). As a sensitivity analysis, chemoimmunotherapy treatment effect (chemoimmunotherapy versus chemotherapy) on survival was also estimated for each BMI subgroup defined by World Health Organisation classification. Exploratory analyses in the respective chemoimmunotherapy and chemotherapy cohort were undertaken to examine the survival outcomes among BMI subgroups. Results A total of 1282 patients were included. From the MFPI analysis, BMI was not significantly associated with chemoimmunotherapy treatment effect with respect to either OS (p = 0.71) or PFS ( p  = 0.35). This was supported by the sensitivity analyses that demonstrated no significant treatment effect improvement in OS/PFS among overweight or obese patients compared to normal weight patients (OS: normal BMI HR = 0.74 95% CI 0.59–0.93, overweight HR = 0.78 95% CI 0.61–1.01, obese HR = 0.84 95% CI 0.59–1.20). Exploratory analyses further highlighted that survival outcomes were not significantly different across BMI subgroups in either the chemoimmunotherapy therapy cohort (Median OS: normal BMI 19.9 months, overweight 17.9 months, and obese 19.5 months, p  = 0.7) or the chemotherapy cohort (Median OS: normal 14.1 months, overweight 15.9 months, and obese 16.7 months, p  = 0.7). Conclusion There was no association between high BMI (overweight or obese individuals) and enhanced chemoimmunotherapy treatment benefit in front-line treatment of advanced non-squamous NSCLC. This contrasts with previous publications that showed a superior treatment benefit in overweight and obese patients treated with immunotherapy given without chemotherapy.

Background The impact of commonly used non‐cancer medications on breast cancer outcomes remains underexplored in large datasets. Aims To evaluate the associations between commonly used non‐cancer medications and survival as well as adverse events in patients with breast cancer. Materials & Methods Individual participant data from 19 breast cancer clinical trials (n = 23,211) were pooled. Cox proportional hazards models and logistic regression analyses were used to assess associations between medication use and overall survival, progression‐free survival and grade ≥ 3 adverse events. Analyses were adjusted for demographic, cancer and comorbidity factors. Results Proton pump inhibitor use was associated with poorer overall survival (HR 1.19, 95% CI: 1.08–1.30), progression‐free survival (HR 1.11, 95% CI: 1.02–1.21) and an increased risk of grade ≥ 3 adverse events (OR 1.36, 95% CI: 1.21–1.53). Beta‐blockers, ACE inhibitors/ARBs and calcium channel blockers were linked with higher adverse event rates but showed no significant impact on survival. Statins and metformin demonstrated no significant associations with either survival or adverse events. Conclusion These findings emphasise the need for careful management of concomitant medications in breast cancer care and support ongoing research to optimise treatment safety and efficacy.

Most ambient AI medical scribes process audio only, omitting clinically important visual details. We developed a vision-enabled AI scribe using Google’s Gemini model and Ray-Ban Meta smart glasses to document medication histories—a task requiring both audio and visual input. Ten clinical pharmacists video-recorded 110 simulated medication history interviews. Following iterative prompt engineering on 10 training recordings, the scribe was evaluated on 100 test recordings (2160 data points) across patient details and medication-specific fields. The vision-enabled scribe achieved 98% overall accuracy (2114/2,160 data points), ranging from 96% for patient details to 99% for dosing directions and indication. Video input significantly outperformed audio-only processing (98% vs 81%, P  < 0.001), primarily through reduced omissions (10 vs 358 errors). Vision-enabled AI scribes substantially improved documentation accuracy for tasks requiring visual input, demonstrating potential to markedly reduce omission errors in clinical documentation.

Abstract Background Monotherapy immune checkpoint inhibitor (ICI) used in second- or later-line settings has been reported to induce hyperprogression. This study evaluated hyperprogression risk with ICI (atezolizumab) in the first-, second-, or later-line treatment of advanced non–small cell lung cancer (NSCLC), and provides insights into hyperprogression risk with contemporary first-line ICI treatment. Methods Hyperprogression was identified using Response Evaluation Criteria in Solid Tumours (RECIST)-based criteria in a dataset of pooled individual-participant level data from BIRCH, FIR, IMpower130, IMpower131, IMpower150, OAK, and POPLAR trials. Odds ratios were computed to compare hyperprogression risks between groups. Landmark Cox proportional-hazard regression was used to evaluate the association between hyperprogression and progression-free survival/overall survival. Secondarily, putative risk factors for hyperprogression among second- or later-line atezolizumab-treated patients were evaluated using univariate logistic regression models. Results Of the included 4644 patients, 119 of the atezolizumab-treated patients (n = 3129) experienced hyperprogression. Hyperprogression risk was markedly lower with first-line atezolizumab—either chemoimmunotherapy or monotherapy—compared to second/later-line atezolizumab monotherapy (0.7% vs. 8.8%, OR = 0.07, 95% CI, 0.04-0.13). Further, there was no statistically significant difference in hyperprogression risk with first-line atezolizumab-chemoimmunotherapy versus chemotherapy alone (0.6% vs. 1.0%, OR = 0.55, 95% CI, 0.22-1.36). Sensitivity analyses using an extended RECIST-based criteria including early death supported these findings. Hyperprogression was associated with worsened overall survival (HR = 3.4, 95% CI, 2.7-4.2, P < .001); elevated neutrophil-to-lymphocyte ratio was the strongest risk factor for hyperprogression (C-statistic = 0.62, P < .001). Conclusions This study presents first evidence for a markedly lower hyperprogression risk in advanced NSCLC patients treated with first-line ICI, particularly with chemoimmunotherapy, as compared to second- or later-line ICI treatment. This study evaluated hyperprogression risk with the use of immune checkpoint inhibitor (ICI) in the first-, second-, or later-line treatment of advanced non-small cell lung cancer, providing insight into hyperprogression risk with contemporary first-line ICI treatment.

Prognostic models can enhance clinician-patient communication and guide treatment decisions. Numerous machine learning (ML) algorithms are available and offer a novel approach to predicting survival in patients treated with immune checkpoint inhibitors. However, large-scale benchmarking of their performances—particularly in terms of calibration—has not been evaluated across multiple independent cohorts. This study aimed to develop, evaluate, and compare statistical and ML models regarding discrimination, calibration, and variable importance for predicting overall survival across seven clinical trial cohorts of advanced non–small cell lung cancer (NSCLC) undergoing immune checkpoint inhibitor treatment. This study included atezolizumab-treated patients with advanced NSCLC from seven clinical trials. We compared two statistical models: Cox proportional-hazard (Coxph) and accelerated failure time models, and 6 ML models: CoxBoost, extreme gradient-boosting (XGBoost), gradient-boosting machines (GBMs), random survival forest, regularized Coxph models (least absolute shrinkage and selection operator [LASSO]), and support vector machines (SVMs). Models were evaluated on discrimination and calibration using a leave-one-study-out nested cross-validation (nCV) framework. Discrimination was assessed using Harrell's concordance index (Cindex), while calibration was assessed using integrated calibration index (ICI) and plot. Variable importance was assessed using Shapley Additive exPlanations (SHAP) values. In a cohort of 3203 patients, the two statistical models and 5 of the 6 ML models demonstrated comparable and moderate discrimination performances (aggregated Cindex: 0.69–0.70), while SVM exhibited poor discrimination (aggregated Cindex: 0.57). Regarding calibration, the models appeared largely comparable in aggregated plots, except for LASSO, although the XGBoost models demonstrated superior calibration numerically. Across the evaluation cohorts, individual performance measures varied and no single model consistently outperforming the others. Pretreatment neutrophil-to-lymphocyte ratios (NLRs) and Eastern Cooperative Oncology Group Performance Status (ECOGPS) were ranked among the top five most important predictors across all models. There was no clear best-performing model for either discrimination or calibration, although XGBoost models showed possible superior calibration numerically. Performance of a given model varied across evaluation cohorts, highlighting the importance of model assessment using multiple independent datasets. All models identified pretreatment NLR and ECOGPS as the key prognostic factors. [Display omitted] •Machine learning (ML) may not outperform statistical models in discrimination.•Model calibration varied across cohorts, offering insights for model selection.•Evaluating models on multiple independent datasets from target population is crucial.•Black box ML and statistical models can be consistently explained by SHAP.•There is a need for comprehensive guidelines on evaluation of survival (ML) models.

Abstract Background Treatment response to immune checkpoint inhibitors varies considerably, a phenomenon known as heterogeneity of treatment effect. Heterogeneity of treatment effect is explored via one-variable-at-a-time subgroup analyses in randomized controlled trials (RCTs), however, this method has limitations, which the risk-modeling approach seeks to address. Methods Applying the risk-modeling approach, individual patient data from 10 RCTs (6 supporting US Food and Drug Administration’s atezolizumab label: OAK, IMpower130, IMpower150, IMpower133, IMbrave150, IMspire150; 4 unlabeled indications: IMpower131, IMpower132, IMmotion151, and IMvigor211) were analyzed by an extreme gradient-boosting algorithm to predict pretreatment prognosis for overall survival. The predicted risk scores were evaluated as efficacy modifiers categorically (high-, intermediate-, low-risk groups) and continuously in Cox models with treatment-by-risk-group interaction terms. Sensitivity and exploratory analyses investigated absolute and meta-analyzed treatment effect and compared the results with established prognostic tools and treatment effect predictors. Statistical significance tests are 2-sided. Results Among the 10 RCTs (n = 7053), one trial (IMvigor211) showed statistically significant heterogeneity of treatment effect by pretreatment prognosis across all evaluations (risk groups, risk scores, sensitivity analyses: P < .001). Among other trials, no statistically significant heterogeneity of treatment effect was detected (risk group and risk score analysis interaction test: OAK P = .61 and P = .77; IMpower130 P = .13 and P = .52; IMpower131 P = .21 and P = .02; IMpower150 P = .14 and P = .36; IMpower133 P = .38 and P = .12; IMbrave150 P = .15 and P = .08; IMspire150 P = .24 and P = .6; IMpower132 P = .15 and P = .81; IMmotion151 P = .48 and P = .21, respectively). Conclusions The risk-modeling approach showed no clear link between pretreatment prognosis and immune checkpoint inhibitor efficacy in most RCTs, particularly those supporting atezolizumab’s Food and Drug Administration label. In IMvigor211, patients with better pretreatment prognosis were more likely to benefit from atezolizumab treatment for platinum-refractory metastatic urothelial carcinoma.

Tumour mutational burden (TMB) is an established biomarker for patients treated with immune checkpoint inhibitors (ICIs). The optimal TMB cut-off is uncertain. It is also uncertain whether there is a sharp TMB threshold or a more graduated change in clinical outcomes as TMB increases. We aimed to determine the relationship between TMB and ICI treatment outcomes using alternative statistical approaches in patients with non-small cell lung cancer. Tumour mutational burden was evaluated as a prognostic and predictive biomarker in advanced non-small cell lung cancer utilising data from two real-world cohorts of ICI use (n = 968) and three randomised controlled trials evaluating ICIs (n = 1588). The non-linear relationship between continuous TMB and response/survival/efficacy outcomes was evaluated using statistical methods that do not require specifying a TMB cut-off. Median TMB for all cohorts was seven mutations/megabase, excluding MYSTIC, where the median was 13 mutations/megabase. Progressively higher TMB was significantly associated with a progressively higher objective response rate and progression-free survival in ICI-treated patients in Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets [MSK-IMPACT] (objective response rate: p < 0.001, progression-free survival: p < 0.001), Strata Clinical Molecular Database [SCMD] (progression-free survival: p = 0.023) and OAK/POPLAR (objective response rate: p = 0.017, progression-free survival: p < 0.001) This relationship was not apparent for patients treated with chemotherapy. There was no obvious TMB threshold for ICI response. The relationship between TMB and overall survival was more complex and heterogeneous. Using a single cut-off to analyse a continuous biomarker may hide important information. Methods that provide more nuance to the underlying relationship between TMB and outcomes enable readers to judge for themselves the value and limitations of TMB cut-offs proposed for clinical practice.

Background Tumour mutational burden (TMB) is an established biomarker for patients treated with immune checkpoint inhibitors (ICIs). The optimal TMB cut-off is uncertain. It is also uncertain whether there is a sharp TMB threshold or a more graduated change in clinical outcomes as TMB increases. Objective We aimed to determine the relationship between TMB and ICI treatment outcomes using alternative statistical approaches in patients with non-small cell lung cancer. Methods Tumour mutational burden was evaluated as a prognostic and predictive biomarker in advanced non-small cell lung cancer utilising data from two real-world cohorts of ICI use ( n  = 968) and three randomised controlled trials evaluating ICIs ( n  = 1588). The non-linear relationship between continuous TMB and response/survival/efficacy outcomes was evaluated using statistical methods that do not require specifying a TMB cut-off. Results Median TMB for all cohorts was seven mutations/megabase, excluding MYSTIC, where the median was 13 mutations/megabase. Progressively higher TMB was significantly associated with a progressively higher objective response rate and progression-free survival in ICI-treated patients in Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets [MSK-IMPACT] (objective response rate: p  < 0.001, progression-free survival: p  < 0.001), Strata Clinical Molecular Database [SCMD] (progression-free survival: p  = 0.023) and OAK/POPLAR (objective response rate: p  = 0.017, progression-free survival: p  < 0.001) This relationship was not apparent for patients treated with chemotherapy. There was no obvious TMB threshold for ICI response. The relationship between TMB and overall survival was more complex and heterogeneous. Conclusions Using a single cut-off to analyse a continuous biomarker may hide important information. Methods that provide more nuance to the underlying relationship between TMB and outcomes enable readers to judge for themselves the value and limitations of TMB cut-offs proposed for clinical practice.

Treatment response to immune checkpoint inhibitors (ICIs) varies considerably, a phenomenon known as treatment effect heterogeneity (HTE). While HTE is explored via one-variable-at-a-time subgroup analyses in randomised controlled trials (RCTs), this method has limitations, which the risk-modelling approach seeks to address.BACKGROUNDTreatment response to immune checkpoint inhibitors (ICIs) varies considerably, a phenomenon known as treatment effect heterogeneity (HTE). While HTE is explored via one-variable-at-a-time subgroup analyses in randomised controlled trials (RCTs), this method has limitations, which the risk-modelling approach seeks to address.Applying the risk-modelling approach, individual patient data from ten RCTs (six supporting FDA atezolizumab label: OAK, IMpower130, IMpower150, IMpower133, IMbrave150, IMspire150; four unlabelled indications: IMpower131, and IMpower132, IMmotion151, IMvigor211) were analysed by an extreme gradient-boosting algorithm to predict pre-treatment prognosis for overall survival. The predicted risk scores were evaluated as efficacy modifiers categorically (high, intermediate, low risk groups) and continuously in Cox models with treatment-by-risk-group interaction terms. Sensitivity and exploratory analyses investigated absolute and meta-analysed treatment effect and compared the results with established prognostic tools and treatment effect predictors. Statistical significance tests are two-sided.METHODSApplying the risk-modelling approach, individual patient data from ten RCTs (six supporting FDA atezolizumab label: OAK, IMpower130, IMpower150, IMpower133, IMbrave150, IMspire150; four unlabelled indications: IMpower131, and IMpower132, IMmotion151, IMvigor211) were analysed by an extreme gradient-boosting algorithm to predict pre-treatment prognosis for overall survival. The predicted risk scores were evaluated as efficacy modifiers categorically (high, intermediate, low risk groups) and continuously in Cox models with treatment-by-risk-group interaction terms. Sensitivity and exploratory analyses investigated absolute and meta-analysed treatment effect and compared the results with established prognostic tools and treatment effect predictors. Statistical significance tests are two-sided.Among the ten RCTs (N = 7053), one trial-IMvigor211 - showed significant HTE by pre-treatment prognosis across all evaluations (risk groups, risk scores, sensitivity analyses: p < .001). Amongst other trials, no significant HTE was detected (risk group and risk score analysis interaction test: OAK p = .61, 0.77; IMpower130 p = .13, 0.52; IMpower131: p = .21, 0.02; IMpower150: p = .14, 0.36; IMpower133: p = .38, 0.12; IMbrave150: p = .15, 0.08; IMspire150: p = .24, 0.6; IMpower132: p = .15, 0.81; IMmotion151: p = .48, 0.21).RESULTSAmong the ten RCTs (N = 7053), one trial-IMvigor211 - showed significant HTE by pre-treatment prognosis across all evaluations (risk groups, risk scores, sensitivity analyses: p < .001). Amongst other trials, no significant HTE was detected (risk group and risk score analysis interaction test: OAK p = .61, 0.77; IMpower130 p = .13, 0.52; IMpower131: p = .21, 0.02; IMpower150: p = .14, 0.36; IMpower133: p = .38, 0.12; IMbrave150: p = .15, 0.08; IMspire150: p = .24, 0.6; IMpower132: p = .15, 0.81; IMmotion151: p = .48, 0.21).The risk-modelling approach showed no clear link between pre-treatment prognosis and ICI efficacy in most RCTs, particularly those supporting atezolizumab's FDA label. In IMvigor211, patients with better pre-treatment prognosis were more likely to benefit from atezolizumab treatment for platinum-refractory metastatic urothelial carcinoma.CONCLUSIONThe risk-modelling approach showed no clear link between pre-treatment prognosis and ICI efficacy in most RCTs, particularly those supporting atezolizumab's FDA label. In IMvigor211, patients with better pre-treatment prognosis were more likely to benefit from atezolizumab treatment for platinum-refractory metastatic urothelial carcinoma.