Frontiers in Artificial Intelligence

Machine learning workflows frequently incorporate data preprocessing to enhance predictive performance. However, the need for Super Learner ensembles made up only of preprocessing-invariant tree-based algorithms remains unexplored. Using three benchmark clinical classification datasets, this study examined how preprocessing affected the Super Learner's prediction performance, learner weight distribution, and oracle behavior. The Heart Disease (207 observations), Indian Liver Patient Dataset (583 observations), and Pima Indians Diabetes (768 observations) datasets were used to create a Super Learner ensemble model that included Classification and Regression Trees (CART), Random Forest, Ranger, and Extreme Gradient Boosting (XGBoost). Models were evaluated under raw and preprocessed data conditions using repeated cross-validation. Predictive performance was assessed using the area under the receiver operating characteristic curve (AUC), Matthews correlation coefficient (MCC), and Brier score. Learner weight allocation and Oracle Gap were compared using paired Wilcoxon signed-rank tests with Benjamini-Hochberg adjustment. Preprocessing produced negligible changes in predictive performance for the Heart Disease and Pima datasets. For the ILPD dataset, preprocessing significantly improved AUC (0.746 to 0.752; adjusted p = 0.0017) and reduced the Brier score (0.177 to 0.175; adjusted p < 0.001). Learner weights remained largely stable, although Random Forest replaced Ranger as the dominant learner for the Heart Disease dataset. Oracle Gaps remained extremely small (<0.002) across all datasets and did not differ significantly between preprocessing conditions. Preprocessing provides limited benefit for Super Learner ensembles composed of preprocessing-invariant learners and does not materially alter their oracle behavior. Preprocessing decisions should therefore be guided by dataset characteristics rather than adopted as a universal modelling practice.

Predicting and understanding behaviour is a primary objective of many disciplines, especially human behaviour, as it is the cause of many of the worlds most pressing problems. Although it is a fundamental concept in multiple disciplines, there is no agreed operational definition of what it is. Neither is there a generally agreed theoretical framework for predicting it. Here we propose a data-driven approach, using the "Conductome" -- the complete set of factors that both predict and explain a behaviour -- to operationalise a discipline-neutral definition of behaviour that is based on an ensemble of stimulus/response measurements of a system, showing that it must be determined through a process of statistical inference. As the prediction of behaviour can be characterised as a classification problem, we argue that Bayesian classifiers offer a promising framework in which explainable prediction models that can approximate the Conductome can be developed. We show the efficacy of the framework using a dataset of 1075 persons, with over 3000 features, constructing a model for predicting sedentariness, a behaviour that is a known risk factor for obesity and metabolic disease. We analyse the effect size, coverage, statistical significance and potential causality of a subset of 396 features associated with 58 variables.of different types.

Alzheimer s disease (AD) is a critical neurological disorder that causes the brain to shrink and leads to the eventual death of brain cells, adversely affecting a person s ability to function. AD is a fast-growing disease in the United States and was the fifth leading cause of death among Americans 65 years of age or older in 2023. In the United States 6.9 million people aged 65 or older were diagnosed with AD, along with a high rate of undiagnosed patients. Thus, the objective of our study is to develop a real data-driven predictive model to identify a patient with AD based on eight risk factors: Age, Gender, ADAS-Cog13, Entorhinal, Fusiform, Intracranial Volume (ICV), Amyloid-Beta, and Tau Protein, with a high degree of accuracy. The quality of the model was evaluated using well-established and sophisticated statistical measures: the area under the receiver operating characteristic curve, calibration plot, Hosmer-Lemeshow goodness-of-fit test, and K-fold cross-validation. If a patient is given information on the above risk factors, our proposed binary logistic regression model can classify the patient as having AD or not with at least 98% accuracy.

Glioblastoma (GBM) is the most common and aggressive primary malignant brain tumor in adults with extremely poor prognosis. Complete surgical treatment is practically impossible, as the true extent of GBM infiltration cannot be fully delineated using currently available in vivo neuroimaging methods, leading to frequent recurrences and low overall survival. Consequently, mathematical models are being developed to estimate the GBM expanse beyond the visible tumor mass, providing additional information for treatment planning and patient prognosis. Here, a novel graph-based stochastic mathematical model of GBM invasion using patient-specific structural brain connectome data is proposed. The model is assessed using publicly available UCSF-PDGM dataset to demonstrate GBM invasion dynamics across multiple patients and anatomical locations. Additional scaling using fractional anisotropy (FA) is tested and evaluated. Parameter sensitivity analysis is provided to explore model's behavior under different settings. Ablation testing is performed to suppress model mechanisms utilizing the structural connectome, showing that the tentacle-like extrusions from the tumor core emerge only if the patient-specific connectome is utilized. The model seems to capture GBM micro-infiltration along white matter tracts to a very high degree, making it a potential tool for studying distant recurrences farther from the resection cavity and GBM invasion dynamics in relation to the structural connectome. Full source code is publicly available, ensuring complete transparency of the study.

Clinical oncology datasets often store systemic therapy as a regimen label with a start date and an end date. Those records are clinically recognizable but can be analytically incomplete when the research question concerns whether a patient was exposed to a concurrent CNS-active drug (cCNS-aD) or an adjuvant CNS-active drug (aCNS-aD) around radiotherapy. Contemporary CNS-oncology studies usually define CNS activity by empiric drug lists and define concurrency by fixed calendar windows, although the literature shows substantial heterogeneity across both concepts. This paper proposes a generalizable framework for converting raw systemic therapy records into reproducible cCNS-aD and aCNS-aD variables, useful in subgrouping for clinical studies. The framework uses a transparent CNS scoring model based on three clinical evidence components: intracranial objective response rate, consensus CNS endorsement, and intrathecal route of administration. It then defines a pharmacokinetic exposure proxy as the recorded end date plus five half-lives. Concurrent exposure is classified by overlap with the radiotherapy interval, while post-radiotherapy exposure is classified by overlap with a prespecified post-RT attribution window. The framework separately identifies post-RT pharmacokinetic persistence and post-RT treatment initiation, allowing investigators to distinguish continued exposure from true adjuvant initiation. This is a methodological framework and reference implementation. Implementation audits and endpoint-specific sensitivity analyses remain necessary before use as a definitive exposure classifier

Survival analysis is a fundamental technique in biomedical research for modeling time-to-event data. It enables the identification of prognostic factors in disease, compares survival outcomes across treatment groups, and performs targeted treatment selection. A variety of machine learning (ML) approaches to survival analysis have emerged to complement classical statistical methods, especially for high-dimensional datasets with complex, nonlinear interactions between features. However, using survival ML methods requires addressing challenges such as censoring-unaware evaluation, overfitting, selecting performance metrics, and data leakage. To address these and other difficulties in using survival ML models, we developed the mlsurv software package. mlsurv is an open-source Python package built around three major design principles: 1) methodological rigor, including evidence-based model selection, leakage-free pipelines, and multi-metric evaluation, 2) multi-scale evaluation and interpretation, including population and subpopulation evaluation, patient-level explanations, and feature analysis, and 3) automated trust and transparency, including limitation flagging and TRIPOD+AI-aligned reporting. mlsurv bundles ten models spanning linear, ensemble, kernel, and deep learning families within a unified software package. We demonstrate mlsurv on the Chowell immunotherapy cohort (n=1,479). The survival-trained models achieve a test concordance index of 0.73 for overall survival prediction. Further, risk scores strongly correlate with the response-trained LORIS clinical score (|{rho}| up to 0.84), reflecting the overlap between prognostic and predictive signal. mlsurv enables biomedical researchers to conduct rigorous, multi-model survival analysis and benchmarking using minimal code with default best practices rather than implementing custom scripts and methodological safeguards from scratch.

Emerging as a significant global health challenge, Alzheimer's Disease (AD) is a progressive neurodegenerative disorder that causes memory loss and cognitive decline. Despite the ever-increasing waiting time for a specialist diagnosis, the need for a cost-effective and fast diagnostic technique is evident. This study explores the development of an explainable deep learning model to diagnose AD using only routine and low-cost clinical data, including demographic information, patient history, and results of neuropsychological tests (limited to those that can be automatically acquired). The analysis was carried out using a dataset provided by the National Alzheimer's Coordinating Center, comprising 167,364 observations and 1,024 features. The findings demonstrate diagnostic performance comparable, and slightly superior, to that of clinicians when evaluated under similar informative constraints. This study introduces two classification models to discriminate whether the presumptive etiological cause of cognitive impairment is Alzheimer's disease. The deep neural network achieved an accuracy of 90\% with an area under the receiver operating characteristic curve (ROC-AUC) of 0.96, whereas the Light Gradient Boosting Machine reached the same accuracy with a ROC-AUC of 0.97.

The process of medical diagnostics is challenging, especially since patients can simultaneously suffer from several diseases with similar, contradictory, or even opposing diagnoses. Statistical prediction can support physicians in this task; however, the quality of data used for predicition as well as the chosen statistical model can affect the reliability of data-driven decision support. Data quality can, in particular, be reduced by incomplete medical diagnoses, that is, the termination of the diagnostic process once a patient has tested positive for one disease that explains the symptoms. When interpreting missing diagnoses as negative, this leads to potentially false negative health data. Another source of low data quality lies in diagnoses being made through a principle of elimination, i.e., after several negative results, one opts for the seemingly last remaining possibility. This may lead to false positive health data. In our work, we investigate how such inaccurately labeled data affects the predictive ability of multi-label classification (MLC) for disease recognition. Unlike single-label classification (SLC), MLC allows the simultaneous assignment of multiple diseases to a patient and can therefore describe clinical conditions more holistically. To that end, we conduct a synthetic-data simulation study as well as a real-data case study on the example of chronic pain patients. In this regard, we compare MLC performance on accurately and inaccurately labeled data. We manipulate the data such that it corresponds to different diagnostic test sensitivities as well as to different examination sequences, thus paying special attention to resulting uncertainty within the process of medical diagnostics. Our results show that inaccurate labeling substantially decreases MLC prediction ability. Furthermore, low diagnostic test-sensitivity, the order of disease examination and covariate effects have a strong impact on MLC performance. These findings contribute to a better understanding of the interplay and impact of diagnostic procedures, data documentation and interpretation, and statistical modeling. This underlines the need for careful data collection as a basis for model development; special consideration should be given to the extensive examination of patients as well as the targeted collection of covariates. This is particularly crucial when models are transferred into everyday clinical practice.

Patients suffering from sepsis need to be treated with an effective antibiotic therapy within the first hour after sepsis onset to decrease their risk of death. Microbiological data that provide information about the suitability of antibiotic therapies, however, is usually available only after 72 hours. Consequently, the treating physicians need to judge a therapy's effectiveness based on the patients' measured health records and their general health condition. This medical assessment is complex and requires years of experience. In our study, we investigate how statistical modeling can contribute to assessing the effectiveness of antibiotic therapies. To that purpose, we describe the effectiveness of antibiotic therapies by modeling sepsis patients' health conditions using a three-state continuous-time hidden Markov model (ctHMM). In literature, procalcitonin (PCT) and lactate have proven to be helpful for deriving the health condition in this context. The state probabilities obtained by the ctHMM are subsequently used to quantify the effectiveness of antibiotic therapies. To this end, we apply two different approaches, namely (i) averaging of the state probabilities and (ii) a logistic regression model. For (i), we calculate the average of the state probabilities for the state indicating a sepsis-free condition over an antibiotic administration period of 48 hours. For (ii), we use the information about antibiotic susceptibility testings as dependent variable in the logistic regression model; as independent variables, we calculate the difference between state probabilities at the start of antibiotic administration and 48 hours later. With this work, we are able to better understand the relationship between laboratory values, in particular PCT and lactate, and the patients' health condition. We further provide approaches for quantifying the effectiveness. Therefore, our work contributes to developing a clinical decision support system which helps physicians assess the effectiveness of antibiotic therapies in patients with sepsis. Supported by such a system, a physician is able to quickly adjust an ineffective therapy which avoids antibiotic resistances and increases a patient's chance to survive a sepsis.

Background Hepatocellular carcinoma (HCC) treatment selection demands nuanced integration of heterogeneous patient data, yet prevailing predictive models rely on restricted data modalities and oversimplified therapeutic frameworks, compromising clinical translation. Objective We developed and validated a multimodal artificial intelligence framework to guide optimal treatment strategy selection across the full spectrum of HCC interventions. Methods This retrospective study comprised 1,043 HCC patients (development cohort, January 2017-December 2023) and 55 external validation patients (2023) from Wuxi Peoples Hospital. We engineered Embedding-Augmented Extra Trees (ET-Emb), a novel model fusing structured clinical variables with contextual text embeddings derived from medical histories and radiology reports. ET-Emb quantifies probabilities for five primary treatments: open/laparoscopic resection, transarterial chemoembolization, radiofrequency ablation (RFA), and chemotherapy. Model performance was rigorously assessed via 10-fold cross-validation and external validation using ROC-AUC and PR-AUC metrics. Results ET-Emb demonstrated robust performance in the development cohort (ROC-AUC: 0.84 {+/-} 0.04; PR-AUC: 0.55 {+/-} 0.06), significantly outperforming established benchmarks. This generalizability was preserved in external validation (ROC-AUC: 0.77 {+/-} 0.02; PR-AUC: 0.47 {+/-} 0.03). SHAP analysis identified textual clinical narratives and socioeconomic determinants as critical predictive drivers. Conclusions By unifying structured and unstructured data modalities, ET-Emb delivers accurate, multi-treatment strategy prediction for HCC. Its clinical validity and the demonstrated significance of textual features establish multimodal AI as an essential paradigm for simulating complex oncological decision-making, positioning ET-Emb as a transformative tool for precision HCC management.

"Revealing the structure of pharmacobehavioral space through motion sequencing" by Wiltschko et al. (2020) has been highly influential in behavioral phenotyping research. In a cohort of nearly 700 mice, the authors demonstrated that Motion Sequencing (MoSeq) could distinguish behavioral effects across a large and diverse set of neuroactive and psychoactive compounds. A central conclusion of the study is that MoSeq syllable features substantially outperform more traditional scalar behavioral features in treatment classification tasks. Although this comparison is not emphasized outside the Results section, the reported advantage corresponds to an increase in classification performance exceeding 50% relative to scalar feature representations. While reproducing parts of the analysis using the publicly available dataset, we found that much of this apparent performance difference can be attributed to differences in preprocessing, classifier selection, and hyperparameter optimization. Under alternative, but comparably standard, analytical choices, the performance gap between scalar features and MoSeq syllables was reduced to approximately 11%. Furthermore, in our reanalysis, the performance advantage of MoSeq syllables became statistically significant primarily in highly dense pharmacobehavioral spaces. These findings do not contradict the utility of MoSeq syllables. Rather, they suggest that the magnitude and generality of their advantage over simpler scalar features may depend strongly on analytical methodology and dataset structure. This distinction is practically relevant, as scalar feature approaches are substantially less computationally demanding and often easier to interpret biologically. Consequently, for laboratories with limited computational resources or for studies focused on specific treatment effects, conventional scalar representations may provide a competitive and more accessible alternative. Our findings highlight the importance of analytical standardization and reproducibility in comparative behavioral representation studies.

Objectives The cost-effectiveness of new amyloid-targeting therapies (ATTs) for patients with mild cognitive impairment (MCI) or mild dementia due to Alzheimer's disease (AD) is influenced by assumptions about treatment effectiveness beyond the trial durations. To assess the cost-effectiveness of ATTs over a lifetime horizon, an AD microsimulation model was applied. Methods The AD microsimulation model is based on statistical joint models, which link cognitive decline (Mini-Mental State Examination [MMSE]) and states of functional independence (MCI, dementia, institutionalisation), fitted to the Amsterdam Dementia Cohort. The time from MCI to death was simulated under care-as-usual (CAU) and two ATT scenarios, assuming an ATT duration of eighteen months and treatment effect waning of 0% (no-waning) or 20% per year. The main outcome was the incremental cost-effectiveness ratio (ICER), defined as incremental costs per quality-adjusted life year (QALY) gained. A societal perspective was taken for costs and effects. Results The ATT scenario without waning resulted in 0.72 additional QALYs and {euro}28,502 additional costs per person compared to CAU. At a list price of {euro}22,600/year for the ATT, the ICER was {euro}39,745/QALY for the no-waning and {euro}151,016/QALY for the 20%-waning scenario. At a willingness-to-pay threshold of 20,000 EUR/QALY, the corresponding threshold prices were 10,500 EUR (95% CI: <{euro}0 to {euro} 23,000; no-waning) and <{euro}0 (95% CI: <{euro}0 EUR to {euro} 2,250; 20%-waning), respectively. Conclusions The cost-effectiveness of ATTs is strongly influenced by the waning of the treatment effect. A favourable cost-effectiveness profile was only achieved when the treatment effect did not wane after the eighteen-month treatment period.

Standard uncertainty-informed rejection can unexpectedly trigger severe performance collapse, exposing localized vulnerabilities that common machine learning metrics typically do not show. We systematically diagnose this failure dynamic using Levodopa-Induced Dyskinesia prediction in Parkinson's Disease as a proof-of-concept. By training a heterogeneous ML ensemble, decomposing Aleatoric and Epistemic uncertainty and applying unsupervised subgroup discovery, we isolated the precise drivers of these atypical errors. Stratified error analysis revealed two divergent predictive regimes previously hidden by a global evaluation. While the models successfully extracted a predictive signal for one subgroup, the baseline features of a second subgroup lacked discriminative capacity, resulting in a high rate of confident misclassifications. Operating entirely below rejection thresholds, this single subgroup flatlined predictive metrics, driving the collapse of the global rejection curve. Ultimately, we demonstrate that atypical rejection failures stem from subgroup-specific data ambiguity rather than algorithmic deficiencies, making localized uncertainty-aware evaluation a critical methodological requirement prior to real-world deployment.

Economic stability and progress in modern technological societies depend on vigorous and independent public funding of science and engineering research. When peer review or funding decisions are perceived as politically directed, scientists, funding agencies, and the public react in coupled and conflicting ways. We describe an evolutionary game-theoretic model to analyze how perceived political interference in science funding affects the interrelated behaviors of scientists, funding agencies, and the public. The model simulates scientists choosing to refuse peer reviews and retaliate, agencies responding by adopting AI-assisted review and altering reviewer pay, and the public accepting or rejecting these AI systems. Through numerical simulations, five principal findings are identified: (1) Operational capacity and institutional legitimacy are governed by separate conditions and can fail independently. (2) Legitimacy of the process is bistable, meaning final states are determined by the publics acceptance of AI. (3) Since the career cost for researchers refusing to review is generally low, resistance/retaliation cascades can readily ignite, leading identical institutions to entirely opposite fates. (4) Increasing reviewer pay only stabilizes participation within a strict budget-solvency frontier, and emergency pay can paradoxically erode the legitimacy it aims to protect. (5) Finally, finite-population simulations reveal that baseline scenarios partition into either legitimacy recovery without capacity or joint failure, confirming that the fundamental separation of capacity and legitimacy outcomes is a dominant structural feature driven primarily by initial scientific resistance and politicization levels. This theoretical work quantifies issues for future work in science policy. SignificanceModern industrial nations rely on public funding of science, certified through expert peer reviews whose authority rest as much on perceived institutional independence as on processing capacity. We model what happens when review or funding decisions are seen as politically directed, and artificial intelligence (AI) substitutes for resistant human reviewers. Treating scientists, the public, and an adaptive funding agency as interacting populations, the model shows that operational capacity and institutional legitimacy obey separate dynamics and can fail independently: automation can sustain a review pipeline while its authority collapses, or legitimacy can recover while backlogs persist. Under finite populations, identical institutions can reach markedly different fates by chance alone, and how often chance favors survival depends on the noise process assumed, not on a fixed institutional probability. This framework identifies which conditions govern whether resistance ignites or whether legitimacy, once threatened, is repaired, as explicit future priorities for empirical calibration and policy design. Lay AbstractModern societies depend on public funding for scientific research, a system that only functions if the public believes the process is fair. Using evolutionary game theory, this mathematical work explores what happens when scientific peer review is viewed as politically compromised, and funding agencies turn to artificial intelligence (AI) to replace human experts who protest the system. The mathematical model reveals several critical warnings for policymakers: O_LIAutomation Does Not Equal Trust: Keeping grant money flowing and maintaining public trust are two separate problems. An agency can successfully use AI to process grants while the public completely loses faith in its decisions, creating a "zombie" institution. C_LIO_LIThe Threat of "Proposal Flooding": When scientists protest, in principle they refuse to undertake peer review, and can overwhelm the funding agency by submitting many proposals. This retaliatory flooding and refusal to review, within a limited budget, can outpace any AI systems ability to keep up, causing the entire pipeline to collapse. C_LIO_LIEmergency Pay Can Backfire: Trying to secure scientists cooperation by raising reviewer pay during a highly politicized crisis can make things worse, as the public may view this emergency pay as a "bribe," which rapidly destroys whatever institutional trust remains. C_LIO_LIEarly Action is Critical: Human behavior is unpredictable, small early differences in how people react can push identical institutions toward completely different fates. This model tracks individual scientists and members of the public to capture their behavior, which appears to be model-dependent, thus producing sensitivity to early demographic fluctuations and chance factors. Under the standard assumption, every simulated agency ended up with an unmanageable backlog; the only thing chance decided was whether public trust recovered. Under a coarser and more commonly used shortcut, roughly half the agencies appeared to recover fully, and a "zombie" agency that kept processing grants while trust collapsed appeared in about one in six. Policymakers therefore must urgently focus on building trust through transparent, explainable AI rollouts before public rejection becomes permanent. C_LI

Deep learning reads 12 lead electrocardiograms at close to expert level on public benchmarks, yet most reports give one accuracy figure for the whole test set and stop there. We trained three architectures that are standard in this field, a 1D ResNet, a convolutional network with a bidirectional LSTM, and a convolutional network with a bidirectional LSTM followed by a transformer encoder, on the PTB-XL dataset to classify the five diagnostic superclasses, and then looked at how each one performed across sex and age. On the held out fold all three reached a macro AUC near 0.92, in line with the strongest published results on this benchmark, and the simplest model, the 1D ResNet, was marginally the best at 0.9241. The averages hid a steady pattern. Every model scored lower for female patients than for male patients, and every model scored lowest for patients aged 80 and over, where the 1D ResNet fell to 0.8878 and the transformer to 0.8693. Adding complexity did not close either gap and slightly widened the gap by age. Overall accuracy on PTB-XL is close to solved for these model families, but the benefit is not shared evenly, and a single headline number hides the patients a model serves worst. We release the full stratified evaluation to support fairness aware reporting.

Large vision-language models (VLMs) have demonstrated remarkable perfor- mance on computational pathology benchmarks, yet their reliability under adversarial or vacuous inputs remains poorly understood. This paper examines the visual grounding behaviour of two Gemini models Gemini 3.0 Flash Pre- view (gemini-flash) and Gemini 3.1 Pro Preview (gemini-pro) on a well known histopathology classification task, and probes for confabulation using a adver- sarial blank-image set. On the real histopathology dataset both models achieve near-perfect accuracy (98.75% - 100%) across three temperatures (0.0, 0.5, 1.0) and three independent runs. On a controlled adversarial set of blank white images labelled as either benign or malignant, however, a stark divergence emerges. Gemini-flash consistently acknowledges the absence of visual content and assigns zero confidence, while Gemini-pro fabricates detailed, clinically plausible histo- logical descriptions and reports high confidence (mean {approx} 0.95) across the same blank inputs, a behaviour we term confident confabulation. The confabulation rate of gemini-pro reaches 77.8% image-responses at temperature 0.0, dropping to 44.4% at temperature 0.5 and rising to 66.7% at temperature 1.0, while gemini- flash records 0% at all temperatures. These findings raise important questions about the safety and trustworthiness of VLMs in clinical decision-support con- texts, and underscore the need for comprehensive evaluation beyond standard accuracy metrics.

Machine-learning survival models are increasingly proposed for intensive-care mortality prediction and are almost always selected and reported using the concordance index, a ranking metric averaged over follow-up. Yet most bedside decisions hinge on a probability at a specific time, such as 60- or 180-day mortality. We asked whether ranking-optimized models remain competitive at fixed clinical horizons against two reference points clinicians actually rely on: unaided attending-physician judgment and the original 1995 SUPPORT logistic model. Reanalyzing the SUPPORT2 cohort (9,105 critically ill adults from five United States centers, 1989-1994) under a stratified 70/15/15 split, we compared a gradient-boosted survival model, the physician's recorded prognosis, and the 1995 model at 60 and 180 days, alongside several alternative learners. The survival model achieved competitive ranking concordance (0.705) yet underperformed both comparators at fixed horizons: at 60 days its area under the ROC curve was 0.750, against 0.808 for physicians on the matched sample and 0.827 for the 1995 model, a gap that held across eight independent data splits and remained statistically reliable after multiplicity correction. The shortfall was not miscalibration, since post-hoc recalibration left discrimination unchanged, nor limited capacity, since neural networks, a deep ranking model, and two timepoint-aware discrete-time models also failed to close it; replacing the ranking objective with timepoint-matched binary training recovered roughly half the gap, pointing to an objective-horizon mismatch. Discrimination was equitable across sex, race, and age, but leave-one-disease-out validation exposed severe failure for disease groups absent from training, and the physician advantage was conditional on a physician electing to provide an estimate. We recommend reporting timepoint-specific discrimination alongside concordance, timepoint-matched training when fixed-horizon predictions drive care, leave-one-subgroup validation, and distribution-free prediction intervals to support selective deployment.

Background: Parametric time-to-event models require specification of a baseline hazard function, which may influence prediction when the underlying hazard shape is uncertain. This study compared conventional joint longitudinal time-to-event models with mechanistic Multi-Task Logistic Regression, which directly models the survival distribution without selecting a continuous parametric hazard family. Methods: A simulated dataset of 100 individuals with longitudinal sum of longest diameters and event outcomes was analyzed using a shared mechanistic tumor shrinkage regrowth model. Event submodels comprised exponential, Gompertz, Weibull, log-normal, log-logistic, and circadian hazards, mechanistic Multi-Task Logistic Regression, and a hybrid neural-mechanistic extension. All models were estimated jointly using shared patient-specific random effects and longitudinal data. Models were evaluated using longitudinal goodness-of-fit, visual predictive checks, five-fold cross-validated inverse-probability-of-censoring-weighted dynamic area under the curve and Brier scores, integrated Brier score, calibration, and event-interval negative log score. Results: Longitudinal parameter estimates and diagnostics were comparable across models. All conventional hazard models produced identical dynamic area under the curve values within prediction windows, although probabilistic accuracy differed. The log-normal hazard achieved the lowest overall integrated Brier score (0.1928). Mechanistic Multi-Task Logistic Regression achieved the highest later landmark discrimination (area under the curve 0.867 versus 0.798 for all hazard models) and the lowest mean event-interval negative log score (2.362). The hybrid model improved intermediate-landmark discrimination but not overall probabilistic accuracy. Conclusions: Mechanistic Multi-Task Logistic Regression provided competitive joint time-to-event prediction while avoiding baseline hazard-family selection. It represents a practical complementary alternative to parametric hazard modeling, particularly when hazard shape is uncertain and dynamic discrimination is important.

Predicting the course of Alzheimer's disease for individual patients remains a major challenge due to the heterogeneity of disease expression and the sparsity of longitudinal data. We introduce a variational Disease Progression Score (DPS) framework that maps multimodal biomarker dynamics (Cerebrospinal fluid, neuroimaging, and cognitive assessments) onto a continuous latent timeline with quantified uncertainty. The framework combines a neural encoder, which infers subject-specific progression parameters from demographic and clinical features, with a cascade of logistic functions structured according to the amyloid cascade hypothesis. Applied to the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort, the inferred timeline separated diagnostic groups it never observed (AUC 0.98 for cognitively normal vs Alzheimer's Disease), and the estimated cascade strengths and biomarker orderings were consistent with the established sequence of Alzheimer's pathology. The model produces individualised prognoses for previously unseen subjects from baseline data alone, with 95\% credible intervals achieving 89-98\% empirical coverage across biomarkers, and these predictions can be dynamically refined as new observations become available. The framework thus provides a biologically interpretable, uncertainty-aware index of disease severity, offering a probabilistic foundation for patient-level prognosis and precision monitoring in Alzheimer's disease.

Background. Accurate delineation of the gross tumour volume (GTV) - primary tumour (GTVp) and nodal disease (GTVn) - on FDG-PET/CT is a critical step of head and neck radiotherapy planning. Comparisons between lightweight custom networks and the auto-configured nnU-Net v2 are usually reported as end-to-end pipelines, conflating the contribution of the network with that of the inference-time post-processing applied on top of it. We separated the two. Methods. MiniUNet3D (custom 3D U-Net, 18.3 M parameters) and nnU-Net v2 (3d_fullres, 88.2 M parameters) were trained on the same 578 FDG-PET/CT cases (85/15 author-defined split of the HECKTOR 2025 Task 1 set, 8 centres) and evaluated on the same internal cohort. Three arms were compared pairwise: MiniUNet3D raw output at a fixed 0.5 threshold, MiniUNet3D with a locked adaptive post-processing pipeline, and nnU-Net v2. Comparisons used paired Wilcoxon tests with bootstrap confidence intervals, Bonferroni and Benjamini-Hochberg correction, and Cohen's d; catastrophic failure (Dice < 0.01) was compared with an exact McNemar test. Cases with an empty reference for a given target were excluded from that target's analysis (n = 98 GTVp, n = 93 GTVn). Results. With post-processing matched off, nnU-Net v2 was superior: median GTVp Dice 0.799 versus 0.592 (mean difference -0.244, 95 % CI -0.300 to -0.191; d = -0.88) and GTVn 0.774 versus 0.598 (d = -0.82). Post-processing raised MiniUNet3D to 0.800 (GTVp) and 0.738 (GTVn), recovering 79 % of that difference. Post-processed, MiniUNet3D matched nnU-Net v2 on GTVp Dice (p = 0.113) but remained inferior on nodal disease after Bonferroni correction (Dice p = 0.041; surface Dice p = 0.049). Catastrophic GTVp failures were 25/98 raw, 8/98 post-processed and 1/98 for nnU-Net v2 (McNemar p = 0.016). Inference took 34 s versus 78 s per case on the same GPU. Conclusions. Post-processing recovered most, but not all, of the difference between the two models, and it did not confer robustness: an eight-fold higher rate of empty contours on small primaries persisted, which is the more consequential difference for planning safety. Pipeline comparisons reported without a post-processing ablation risk attributing to a network what post-processing supplied.