Frontiers in Artificial Intelligence

Background: Adjuvant therapy decision in early-stage node-negative breast cancer remains one of the most challenging problems in oncology. Although most patients in this population achieve favorable outcomes with endocrine therapy alone, a subset remains at risk of relapse and may benefit from chemotherapy. Accurate identification of high-risk and low-risk patients is therefore essential to prevent both undertreatment and overtreatment. Methods: In this study, machine learning models were assessed using MSK-IMPACT (n=645) and METABRIC (n=503) cohorts comprising node-negative stage I-II breast cancer patients. Three experimental settings were evaluated: training on MSK-IMPACT and testing on METABRIC, the reverse direction, and five-fold cross-validation on the combined cohort. Sensitivity analysis and data harmonization were applied to identify and mitigate cohort-specific bias. Survival analysis, SHapley Additive exPlanations (SHAP) interpretability, and ablation analyses were performed to validate clinical relevance and characterize feature contributions. Results: The combined cohort model achieved an area under the receiver operating characteristic curve (AUC) of 0.788{+/-}0.028 across five-fold cross-validation, while bidirectional cross-cohort validation yielded AUC of 0.741 when testing on MSK-IMPACT and AUC of 0.714 when testing on METABRIC. Survival validation on METABRIC out-of-fold predictions from the combined cohort model showed hazard ratios of 3.01 (95\% CI: 2.28-3.98) for disease-free survival and 2.35 (95\% CI: 1.79-3.09) for overall survival, consistent across Luminal A, Luminal B, and Triple-Negative subtypes. Conclusion: Validated across two geographically independent cohorts, the proposed model enables cost-effective, gene expression-free identification of patients at high-risk of distant relapse and may facilitate adjuvant chemotherapy decision-making in early-stage node-negative breast cancer.

Background: Neuroanatomy poses a significant challenge for Psychology students due to its spatial and conceptual complexity. Educational approaches that enhance the relevance and visualization of neuroanatomical content may improve students learning experiences. This study implemented a tractography-based activity focused on the virtual dissection of the arcuate fasciculus, a major white matter pathway, in undergraduate Psychology students and examined the relationships between perceived learning and students perceptions of utility, difficulty and handling, and organizational aspects of the activity. Methods: First-year undergraduate Psychology students participated in a two-session tractography-based activity combining instruction on white matter anatomy and diffusion tractography with a hands-on virtual dissection of the arcuate fasciculus using research-grade software routinely employed in neuroscience research. Following the activity, students completed an anonymous questionnaire assessing perceived learning, utility, difficulty and handling, and organizational aspects of the activity. Pearson correlations, multiple regression analyses, and relative importance analyses were performed. Results: Sixty-eight students completed the questionnaire. Students reported generally positive perceptions of the activity across the evaluated dimensions, with perceived learning receiving the highest mean score (M = 3.44, SD = .85). Perceived utility showed the strongest association with perceived learning (r = .64, p < .001). The regression model explained 41% of the variance in perceived learning (R2 = .41, adjusted R2 = .38, p < .001). Perceived utility was the only significant predictor in the model ({beta} = .59, p = .001), accounting for 67.9% of the explained variance. Conclusions: The findings support the feasibility of integrating authentic neuroimaging tools into undergraduate neuroanatomy teaching. Students who perceived the activity as more useful also reported higher perceived learning outcomes, with perceived utility emerging as the strongest predictor of perceived learning. In contrast, perceived difficulty and handling, and organizational aspects did not make significant independent contributions. These results suggest that students perceptions of educational relevance may play an important role in technology-enhanced STEM learning experiences.

Survival prediction frequently exhibit unstable and irreproducible performance, limiting their clinical applicability. Rather than proposing a new predictive architecture, this work presents a diagnostic investigation into the sources of this instability. Using a transformer-based deep fusion model combining RNA sequencing and histopathology data from the TCGA-BRCA cohort as a diagnostic probe, we demonstrate that global survival prediction is highly unstable across data splits. We then analyze the learned multimodal embeddings in an unsupervised manner, revealing stable latent patient subgroups with significantly divergent survival outcomes. These subgroups are not explained by canonical receptor status or immune infiltration, but instead align with tumor-intrinsic proliferative activity. Conditioning survival modeling on the largest discovered subgroup recovers strong and stable prognostic performance, substantially outperforming unimodal baselines. These findings indicate that survival predictability is conditional on latent heterogeneity and suggest that heterogeneity-aware modeling strategies are essential for robust multimodal prognostication. This work establishes a methodological framework for interpreting instability in multimodal survival learning.

In hypertension screening, previously diagnosed cases provide reliable positive labels, whereas individuals without a prior diagnosis comprise both undiagnosed hypertensive and non-hypertensive individuals, creating a canonical positive-unlabeled (PU) learning setting. We propose PU-Boost, a two-stage reliable-sample reconstruction framework that uses only readily obtainable demographic, behavioral, medical-history, and anthropometric variables. During training, only previously diagnosed hypertension cases are treated as labeled positives (P), while all remaining individuals are treated as unlabeled (U); no additional reference labels are supplied to the two-stage reconstruction procedure. Elastic Net first provides a global ranking of P versus U, after which a random forest refines the uncertain region and highly ambiguous observations are rejected. XGBoost is then trained on the reconstructed class-balanced pseudo-labeled sample. In the independent test set, PU-Boost achieved a sensitivity of 0.817 and a balanced accuracy of 0.682, the highest observed values among the compared models. It produced 729 false negatives, 41.1% fewer than standard XGBoost, although specificity decreased to 0.547. SHAP analysis indicated that established hypertension-related features, including age, body mass index, waist circumference, and family history, contributed strongly to model output. PU-Boost therefore provides a statistically motivated framework for reducing missed cases when screening for undiagnosed hypertension under incomplete diagnostic labeling.

Suicide risk assessment in clinical conversations relies on the sequential structure of therapist-patient dialogue, yet formal models of how suicide-related content emerges, transitions, and re-ciprocates across conversational turns are lacking. We applied network analysis to C-SSRS-coded utterances from psychotherapy sessions to characterize the topology and dynamics of suicide-related clinical communication. Transcripts of 110 psychotherapy sessions from the SPEAK-SAFE study were classified for suicidality relevance using the Qwen3-32B large language model with an adapted Columbia-Suicide Severity Rating Scale (C-SSRS) schema. From the suicidali-ty-relevant utterances, we constructed a weighted directed category transition network. Centrali-ty, reciprocity, clustering, assortativity, and path length metrics were computed. Independent human expert ratings for a session-level subsample of 30 sessions from this cohort showed sub-stantial agreement with the LLM classifications (Spearman {rho} = 0.885, ICC[2,1] = 0.708). The network was dominated by therapist inquiry. Reciprocity was high, and closeness centrality de-creased monotonically with C-SSRS severity. Active suicidal ideation functioned as a bridge node from which escalation pathways to method, plan, and attempt categories originated. The network exhibited a hub-and-spoke topology with small-world characteristics. Suicide-related clinical dialogue is structured as a therapist-centered hub-and-spoke network with high reciproci-ty and a consistent severity-peripherality gradient. The extension of network theory from intra-individual symptom structures to inter-individual communication dynamics represents a method-ological innovation demonstrated here as a proof of concept, with potential applications to clini-cal training, automated decision support, and multimodal risk assessment. clinical interviews clin-ical interviews

Game theory in biology started as an attempt to model animal contests by making assumptions about the fitness costs individuals pay in aggressive interactions. Among the different approaches, a notable one is to construct mechanistic models, using assumptions about behaviours and cognitive processes. Reinforcement learning is an important cognitive process, and we use it here to model pairwise and coalitional contests. We study situations where territorial neighbours become so-called dear enemies. A defending individual can get help from a neighbour through a defender-neighbour coalition against a challenger attempting territory takeover. In our model, coalition members have an advantage in contests in terms of costs of aggression, with the challenger being exposed to aggression from both members and each member only receiving part of the aggression from the challenger. We find that notable costs of border conflicts between neighbours, together with substantial advantages for coalitions, favour intervention. Neighbours intervene when they can help weaker defenders, in this way avoiding costs of border renegotiation with a new and potentially stronger territory owner. We also introduce different forms of perceived costs into the model, which we show correspond to the range from pure mutual assessment to partial self-assessment.

Background and Objective: The volume of patient portal messages has grown substantially and artificial intelligence (AI) can support clinical staff and improve the timeliness of patient care. Our objective is to test if an AI tool for managing the electronic inbox In-Basket reduces response times to patient portal messages in neurology. Methods: We conducted an implementation study from January to May 2026 in the neurology department at two clinical sites within a regional medical system. A single AI tool was implemented to auto-categorize and auto-route all patient messages received in the electronic inbox. The tool also generates optional draft replies to messages. Primary process measures were the median weekly time (hours) from message creation to the first action and to closure. Impact was analyzed using a controlled interrupted time-series (ITS) design, comparing message-volume-matched control divisions. Results: Among 24,142 messages analyzed, pre-intervention median weekly time to first action was 11.3 hours [IQR: 5.6, 15.1] (Site A) and 16.1 hours [IQR: 13.8, 19.0] (Site B), falling post-intervention to 1.0 hour [IQR: 0, 3.1] (Site A) and 0.6 hours [IQR: 0.2, 2.4] (Site B). Controlled interrupted time-series analysis showed a significant immediate differential level change in median weekly hours to first action at Site A (-92.5%, 95% CI: -97.5% to -78.0%), and a significant downward trend at Site B (-43.5%, 95% CI: -65.2% to -8.2%). Changes in time to closure were directionally favorable but not statistically significant at either site. There was low use of the generated draft replies, only 0.5% at Site A and 3% at Site B. Discussion: Implementation of AI-assisted message categorization and routing was associated with a substantially reduced initial action time for patient In-Basket messages in Epilepsy. The utilization of AI-generated draft-replies to messages was low. These findings support the feasibility and effectiveness of AI-augmented workflows to improve inbox management.

BackgroundEvaluating fairness in artificial intelligence (AI) is essential before clinical implementation to prevent unintended harm across diverse patient populations. This study assesses performance differences of an image-based surgical site infection (SSI) detection model across age, gender, geography, race, and skin color. MethodsModel development/validation study using retrospective, multicenter data across surgical subspecialties from adults who submitted an image of their incision via the electronic health record within 30 days of their procedure between 2019-2024. Area under receiver operating curve (AUROC) was compared by age (<55 vs. [&ge;]55 years old), gender (male vs. female), geography (Rochester, Minnesota vs. Jacksonville, Florida vs. Scottsdale, Arizona vs. regional Minnesota/Wisconsin areas), race (White/Caucasian vs. non-White/Caucasian), and skin color (Fitpatrick I-III vs. IV-VI and Monk A-E vs. F-J). ResultsTraining data contained 13,702 images from 4,274 patients between 2019-2022 (50.1% [&ge;]55 years old; 63.2% female; 91.6% White/Caucasian; 58.5% Rochester, 17.3% Florida, 10.5% Arizona; 96.3% light skin tone), of which 1,348 images (9.8%) had an SSI. No differences in SSI detection existed for age, gender, race, or skin color (AUROC 0.83-0.87, p>0.09), however geographical differences were significant: lowest AUROC 0.83 (95% CI 0.77-0.88) vs. highest AUROC 0.87 (95% CI 0.84-0.89, p=0.01). Temporal validation using 4,242 images from 1,353 patients between 2022-2023 showed intact fairness across subgroups ({Delta}AUROC [&le;]0.06) except geography ({Delta}AUROC [&le;]0.16). ConclusionsThe model behaved fairly across age, gender, race, and skin color, however geographical differences persisted. These findings have practical implications underscoring the importance of evaluating bias in clinical AI before deployment.

Background: Conventional performance metrics do not indicate whether artificial intelligence (AI)-platforms and radiologists are right or wrong on the same examinations, nor whether their non-agreement is more often beneficial or harmful. Objective: To develop, illustrate, and apply directional metrics quantifying non-agreement between radiologists and AI-platforms against a common reference standard, and to adapt a fragility-based framework for quantifying the statistical robustness of these metrics. Methods: We defined Positive Non-Agreement (PNA, AI-platform correct, radiologist incorrect), Negative Non-Agreement (NNA, radiologist correct, AI-platform incorrect), and Net Non-Agreement (NENA, their difference). Metrics were illustrated in three hypothetical scenarios of 1,000 examinations each and applied to three previously published AI-radiologist studies on pulmonary embolism and mammography. Bootstrap confidence intervals were calculated, and a reclassification fragility index (FI) quantified the minimum case reclassification needed to remove statistical support for NENA's direction. Results: In illustration, NENA was positive in two scenarios and negative in one. In the three published studies, NENA ranged from -4.18% to +10.74% and decreased consistently with increasing radiologist experience (junior residents, +6.27%; consultants, +1.45%), a gradient driven primarily by declining PNA with modest increase in NNA. The reclassification FI was defined for all three illustrative scenarios (range, 0.10%-1.70%) and for six of the seven comparisons from these studies (range, 0.34%-4.31%), but not for the weakest comparison, whose confidence interval for NENA included zero. Conclusions: PNA, NNA, and NENA provide a directional, case-level complement to conventional performance metrics for AI-radiologist non-agreement. The accompanying reclassification FI offers a statistically grounded, generalizable method for quantifying their robustness.

Background: Artificial intelligence (AI) models for mammography can achieve high diagnostic performance when training and test data originate from similar populations and imaging environments, but performance often deteriorates under cross- cohort domain shift. Differences in acquisition systems, image characteristics, preprocessing, population composition, and disease prevalence can introduce distributional variation that is poorly captured by conventional within- dataset validation. Recent studies have further shown that Foundation Model representations can retain substantial dataset-specific information. However, it remains unclear how image representation influences diagnostic transfer to completely unseen cohorts and whether domain-adversarial learning provides comparable benefits across different representation types. Methods: We systematically evaluated cross-cohort mammography classification using seven independent datasets spanning diverse geographic, institutional, scanner, and imaging environments. Three image representations were compared: handcrafted radiomic features, EfficientNet-B0-derived features, and embeddings from the medical vision-language model MedSigLIP-448. Each was evaluated under three preprocessing conditions: unprocessed images, Otsu-based breast-region extraction, and Otsu extraction combined with Contrast Limited Adaptive Histogram Equalization (CLAHE). Patient-level classification was performed using a mean-pooled multilayer perceptron (MLP) baseline and an attention-based Domain-Adversarial Neural Network (DANN). Generalization was evaluated using a multi-source Leave-One-Domain-Out (LODO) framework in which six datasets were used for model development and the seventh was completely withheld as an unseen test cohort, iteratively across all seven datasets. Performance was primarily assessed by area under the receiver operating characteristic curve (AUC-ROC). Results: Cross-cohort performance depended strongly on image representation and model architecture. MedSigLIP consistently provided the highest performance on unseen cohorts. With MedSigLIP features, mean LODO AUC-ROC increased from 0.598 with MLP to 0.685 with DANN for unprocessed images, from 0.607 to 0.678 after Otsu preprocessing, and from 0.603 to 0.659 after CLAHE+Otsu preprocessing. When DANN-MLP differences were evaluated across held-out datasets and preprocessing conditions, a significant gain was observed for MedSigLIP after correction for multiple comparisons, whereas corresponding gains were not significant for Radiomics or EfficientNet-B0. A mixed-effects analysis similarly showed a positive MedSigLIP X DANN interaction, although this narrowly missed conventional statistical significance. Unprocessed images produced the highest numerical mean performance for MedSigLIP-DANN, but preprocessing itself had no significant overall effect. Notably, linear probing showed strong dataset discriminability in MedSigLIP and EfficientNet representations but little linearly recoverable dataset information in Radiomics, indicating that low dataset separability alone did not predict better diagnostic generalization. Conclusion: Cross-cohort mammography generalization depends strongly on the choice of image representation and its interaction with the classification strategy. MedSigLIP provided the most transferable representation in our multi-source unseen-domain evaluation, and the performance gain associated with DANN was concentrated in MedSigLIP rather than consistently observed across representation types. These findings suggest that the effectiveness of domain-adversarial learning is representation-dependent and that neither conventional preprocessing nor low dataset discriminability alone ensures robust transfer to unseen clinical cohorts.

Interception of moving objects is a universal pattern of animal behavior with many industrial applications. Inspired by the behavior of the characid fish Brycon guatemalensis, which catches fruits falling onto the surface of a river, we study one-shot (i.e., open-loop) interceptions using both stochastic optimal control theory and reinforcement learning. In the first instance, an agent actuates a control (wait time before launch and heading angle) to meet its target, given a noisy estimate of the target initial state. In a minimal setting, where launch velocity and actuated control are constants, we show that the optimal wait time is attenuated by noise in both the target measurement and control actuation, as well as agent energy conservation. We then derive an upper bound for the heading angle variance which scales as the ratio of the non-vanishing optimal interception error to the initial planar separation. Complementing our control framework, our deep reinforcement learning approach shows that the neural network representing the fish first learns the value function for the interception, then learns how to aim, and finally learns the optimal wait time. Beyond the specific problem at hand, our results may also have applications to problems with little time for feedback or corrections such as perching, landing, and docking.

Objective A diversity of approaches and cohorts have been published in research literature for suicide risk estimation, leading to some uncertainty around which factors and conditions are most strongly contributing to risk. Methods exist for weighting estimates by propensity score, removing simplifying the interpretation of model effects. Few studies have explored the use of these tools to assess when the effect of certain factors is uncertain or inconsistent. Method A mix of 1,001,737 suicide decedents and living patients were evaluated for suicide risk, relying on a subset of commonly identified factors: demographics, psychiatric conditions, substance use disorders, and social needs. Care utilization patterns were then assessed controlling for these factors in a hospital care setting. Four methods were considered for controlling multivariate confounding of odds predicting death by suicide: multivariate logistic regression, 1-to-1 sample matching, inverse propensity weighting (IPTW), and entropy balancing (EBAL). Results Test statistics showed suicide decedents were much more likely to have various psychiatric and substance use diagnoses, social needs and more likely to heightened care utilization across sampling intervals. Greater utilization was found close to date of death. Effect sizes after IPTW and EBAL weighting were also generally greater than would have been documented by regression alone. Conclusions Propensity weighting is a useful alternative to exploring nuanced effects, especially when the direction and strength of an association differ from traditional statistics. By weighting, the overall imbalance of cases remains consistent and effect terms might reflect more nuanced associations than regression or sample matching alone.

In recent years, the issue of reproducibility in scientific studies has been at the forefront of scientific discussion. Many of these issues arise due to poor employment of statistical methods, making it crucial to address these knowledge gaps for frequently used methods. Elastic net regression (ENET) is one such algorithm used frequently in biology and public health due to its ability to select important predictors (also known as "features") from large data sets. However, ENET struggles with consistent feature selection and hyperparameter choice in these applications due to low observation counts and noisy data. This limitation hinders ENET's effectiveness in biological research due to low model reproducibility and predictor interpretability. We propose the Instability Quotient Criterion (IQC) as a metric to quantify the consistency of selected features for feature selection models such as ENET, with a specific emphasis on biological applications. IQC leverages ensemble modeling and Shannon's entropy in order to determine the stability of selected features derived from ENET models. We demonstrate IQC's robustness via simulation studies, showing that IQC is capable of detecting significant decreases in model stability with lower observation counts, higher dimensionality, and higher proportions of noisy signal. We additionally utilize simulation to provide a naive threshold for low versus high stability. Finally, we applied IQC to a case study to illustrate how our method can be integrated into a larger analysis pipeline when looking at microarray gene expression data. IQC indicated stability issues for differentially-expressed genes when classifying acute leukemia subtypes in spite of high classification success. Lastly, we discuss how model instability impedes interpretation of biologically significant genes for discerning cancer subtypes. Overall, we present IQC as a robust evaluation metric that can easily be incorporated into high-dimensional biological analyses.

Objective: We explored the implementation considerations of machine learning (ML) suicide risk identification in a health system that serves primarily American Indian patients in the Southwestern United States. Previous work developed and reported the performance of suicide risk models, but traditional performance metrics do not directly reflect clinical workflows. Materials and Methods: Using EHR data collected for Emergency Department (ED) visits between 10/01/2019 and 10/02/2021, we conducted Decision Curve Analysis (DCA) to identify the standardized net benefit (SNB) of all models previously evaluated and plotted the potential workload implications of the model with the greatest SNB. Finally, we evaluated suicide screening practices during the study period to inform implementation barriers. Results: Among 9,244 patients, 139 (1.5%) had a suicide attempt or death within 90 days of an included visit. The population-specific logistic regression model had the greatest SNB compared to other approaches. When operationalized, the 95th percentile cutoff did not substantially increase clinical workloads, while the 75th percentile cutoff substantially impacted workloads. The secondary analysis on suicide screening depicted variations in screening administration over time. Discussion: While DCA identified advantages of ML suicide risk identification over current standards, the complexities of suicide-related care limited its clinical interpretability. A stepped-care framework may mitigate potential workload burdens. Training initiatives and dynamic implementation strategies will support high-quality implementation of ML suicide risk identification. Conclusion: Our analyses support the feasibility and utility of augmenting existing suicide screening practices with ML suicide risk identification, and demonstrate the importance of nuanced analyses to inform implementation considerations.

Depression in Alzheimer's disease (AD) is clinically challenging to characterise because affective and cognitive manifestations frequently overlap with the underlying neurodegenerative condition. This study investigates depression-related speech patterns in AD using an explainable multimodal framework that integrates linguistic and acoustic features from elicited picture-description speech. To address the absence of directly observed depression labels in the target AD cohort, we employ a source-informed cross-dataset transfer strategy. Unsupervised clustering identified two distinct speech sub-phenotypes within the AD cohort. By comparing these unsupervised representations to a depression-labelled source corpus, one cluster demonstrated mathematical proximity (measured via cosine similarity in a shared latent space) to the source depression centroid. We explicitly acknowledge that this proximity reflects geometric similarity in a shared latent space, which may encapsulate overlapping domain characteristics alongside affective variance. To validate the structural integrity and separability of these discovered clusters, a proxy supervised classification task was employed. The retained multimodal artificial neural network (ANN) integrated information from both modalities and achieved 92.73\% cluster separability accuracy. This demonstrates that the representation-derived two-cluster partition was highly learnable on the independent internal hold-out test set; it does not establish clinical validity, longitudinal stability, or diagnostic validity of the identified sub-phenotypes. For interpretability, Shapley Additive exPlanations (SHAP) analysis was utilised to extract feature attributions and organise influential acoustic and linguistic features into a multimodal taxonomy. Furthermore, a Multimodal SHAP Attribution Magnitude (MSAM) score was derived as an exploratory model-based attribution score rather than a clinically validated severity measure. Ultimately, this study provides an interpretable computational baseline for characterising depression-related speech variance in AD, establishing a structural foundation for future prospective validation using independently labelled clinical cohorts.

ImportanceCardiovascular disease remains the leading cause of mortality worldwide. Although machine learning models have demonstrated strong predictive performance, limited attention has been given to uncertainty quantification and gender-specific risk factor identification, which are essential for clinically trustworthy artificial intelligence. ObjectiveTo develop and evaluate an interpretable and uncertainty-aware machine learning framework for cardiovascular disease prediction that integrates an association study on biomarkers, explainable artificial intelligence, and gender-stratified risk assessment. Design, Setting, and ParticipantsThis retrospective observational study utilized the UCI Heart Disease dataset comprising 918 participants, including 725 males and 193 females. Demographic, clinical, laboratory, and exercise-related variables were analyzed. Missing values were addressed using a hybrid imputation strategy combining MICE and supervised Random Forest imputation. MethodsAn ensemble machine learning imputation strategy was developed to address missingness. Then advanced machine learning models were developed and evaluated using stratified training/testing partitions and 5-fold cross-validation. Predictive performance was assessed using several evaluation metrics. Uncertainty was quantified through bootstrap resampling, predictive entropy, and mutual information. Linear Double machine learning and Causal Forest double machine learning were applied to estimate gender-specific averages and association studies for risk factors. ResultsRandom Forest achieved the strongest and most stable predictive performance across male and female cohorts, with ROC-AUC values exceeding 0.90 and robust cross-validation performance. Hybrid imputation improved classification performance, increasing the F1-score from 0.87 to 0.89 and ROC-AUC from 0.90 to 0.92. The causal association study analyzes identified the number of major vessels, ST-segment slope, thalassemia, and exercise-induced angina as the strongest determinants of heart disease, with notable differences observed between male and female subgroups. Conclusions and RelevanceIntegrating predictive modeling, uncertainty quantification, and association study machine learning provides a robust and clinically interpretable framework for cardiovascular disease risk assessment. The proposed approach improves prediction reliability while identifying gender-specific causally important risk factors, supporting precision cardiovascular medicine and trustworthy clinical artificial intelligence.

Identifying progressive dementia during its prodromal stage is a challenging yet critical task for healthcare providers. Traditional clinical assessments are often not sensitive enough to detect warning signs of cognitive decline. Machine learning (ML) methods have shown promise in modeling clinical trajectories associated with neurodegenerative disorders such as Alzheimer's disease (AD). Validation of these models' forecasting abilities, however, is widely lacking in the literature. To address this gap, we established a longitudinal cohort from the National Alzheimer's Coordinating Center (n = 15,727) comprising individuals who were cognitively normal (CN) or had mild cognitive impairment (MCI) at baseline. Cognitive status and clinical observations collected over up to 10 visits were used to train several ML classifiers. We employed the gradient-boosted tree ensemble method XGBoost with a custom feature-engineering pipeline. On non-imputed data, our models achieved an AUC-ROC of 0.9447 [0.9417-0.9476] and PR-AUC of 0.8554 [0.8497-0.8611] for the CN to MCI task, and an AUC-ROC of 0.9059 [0.8983-0.9134] and PR-AUC of 0.9214 [0.9156-0.9272] for the MCI to AD task. In a lead-time forecasting analysis spanning mean follow-up periods of 56.1 months (range, 9.1-214.1) for CN to MCI and 42.7 months (range, 9.1-126.1) for MCI to AD, the non-imputed models detected future progression before diagnosis in 29.58% [28.29%-30.87%] of CN to MCI progressors and 22.41% [21.36%-23.46%] of MCI to AD progressors. Among these individuals, the mean lead time was 33.04 months [32.52-33.56] and 17.95 months [17.71-18.19], respectively. False alarm rates among non-progressors were 12.6% [11.5%-13.7%] and 20.0% [18.4%-21.6%], respectively. These findings suggest that longitudinal machine learning models can identify future cognitive decline before diagnosis, indicating clinical utility for early dementia risk stratification.

Voxel-level prediction of treatment response from longitudinal fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) can enable spatially adaptive dose planning in advanced non-small cell lung cancer (NSCLC), but point predictions alone lack the uncertainty quantification needed for safe clinical decision-making. Standard conformal prediction (CP) provides distribution-free coverage but yields spatially uniform intervals and assumes exchangeability, an assumption violated by intra-tumoral spatial correlation. We propose a multiscale CP framework built on a re-optimized VoxelForecast generalized least squares model that explicitly captures voxel-level covariance. Variogram sensitivity analysis identified the Stable model as significantly superior to the previously used Matern family. The framework combines hierarchical nested leave-one-patient-out calibration, which restores exchangeability for valid voxel-level coverage, with residual-variance CP (VarCP), which learns a heteroscedastic, location-dependent scale to produce adaptive intervals while preserving finite-sample marginal coverage guarantees. On two prospective NSCLC trial cohorts, FLARE-RT (N=25 locally advanced, chemoradiotherapy; 11,100 voxels) and BRIGHT (N=19 metastatic, chemoimmunotherapy; 26,980 voxels), multiscale feature integration reduced voxel-level prediction error by 35% and 31%, respectively. VarCP maintained nominal coverage while producing 7-18% (FLARE-RT) and 4-8% (BRIGHT) narrower voxel-level intervals than a hierarchically calibrated standard CP baseline (paired Wilcoxon signed-rank, significant at most miscoverage levels), with comparable lesion-level gains (7-18% and 6-10%) via median-based aggregation of voxel residuals. VarCP delivers tighter yet mathematically guaranteed spatial uncertainty maps, supporting both voxel-level dose adjustment and whole-lesion treatment planning in advanced NSCLC.

While Artificial intelligence (AI) is increasingly being integrated into psychotherapy, little is known about how prospective patients evaluate different levels of AI involvement. Participants (n = 1,003) with depression evaluated vignettes describing psychotherapy options, from treatment delivered without AI integration (Human therapist) to treatments involving different degrees of AI integration (Assistive, Collaborative, Fully Autonomous AI). Participants consistently evaluated Human therapists more favorably, reporting a greater likelihood of seeking treatment, less hesitancy, and greater treatment acceptability. Economically, compared with a Human therapist, participants were willing to pay 31.6% less for therapists using Assistive or Collaborative AI and 57.2% less for Fully Autonomous AI. 18-19% of participants indicated that they would refuse treatment involving Assistive or Collaborative AI, compared with 31% for Fully Autonomous AI. Preference for Human over AI therapists was particularly pronounced at higher levels of therapist experience (doctoral/masters vs. peer/trainee). Overall, participants valued psychotherapy delivered by Human therapists more highly than AI integrated psychotherapy. These findings suggest that, despite increasing AI integration in mental health care, prospective patients continue to view psychotherapy as a distinctly human endeavor.

Biomedical Machine Learning (ML) is increasingly evaluated through Independent and Identically Distributed (IID) benchmarks, even when deployment involves irregular observation, missingness, distribution shift, and high failure costs. This position paper argues for a stability-oriented reporting standard for safety-relevant biomedical ML. The central requirement is not a specific model class or optimizer, but auditable evidence: performance should be reported under clinically plausible perturbations such as thinning, timestamp jitter, bursty missingness, channel dropout, and Signal-to-Noise Ratio (SNR) shift. When authors invoke training-stability diagnostics, those diagnostics should be hazard-linked: their assumptions, operational meaning, and relationship to stress-test degradation should be stated explicitly. We use controlled mechanism probes to illustrate two hazards that IID scores can hide: observation-process dependence under irregular sampling, and perturbation sensitivity associated with high-curvature training regimes such as Edge of Stability (EoS). We do not propose Neural ODEs, continuous-time models, Sharpness-Aware Minimization (SAM), or EoS metrics as default prescriptions. Instead, we propose a minimum reporting standard that makes robustness claims testable and supports escalation to more complex modelling or optimization only when stress tests justify the cost.