Citations

Abstract Cancer outcomes vary widely between individual patients, each accumulating an irregular record of treatments, diagnoses, measurements, and complications. Current prognostic models reduce this complexity into a single snapshot, focus on narrow clinical settings, and rarely generalize across hospitals. Here we introduce Chronicle , an explainable transformer that learns from entire patient trajectories to predict diverse clinical outcomes throughout the disease course while capturing both short- and long-term temporal dependencies. Trained on 53.7 million longitudinal data points from 51,711 patients spanning 67 cancer types, Chronicle operates natively on irregular data without imputation and jointly predicts eight endpoints within a flexible framework adaptable to additional outcomes. Chronicle outperformed cross-sectional models for overall survival prediction (C-index 0.84 vs 0.76-0.79), stratified patients more accurately than established prognostic systems, including TNM stage, and predicted seven adverse event and transfusion endpoints (AUC 0.80-0.92). Applied without retraining to 69,341 patients in Germany, Switzerland, and the United States, Chronicle generalized across healthcare systems and improved further with local fine-tuning. Integrated explainability traced each risk update to patient-specific clinical factors, revealing distinct temporal persistence of prognostic information, with relevance half-lives ranging from weeks for therapies to nearly one year for baseline characteristics. These findings demonstrate that learning from hospital-wide patient trajectories enables interpretable and continuously updated predictions, providing a scalable framework to support individualized treatment decisions.

Abstract Purpose: Homozygous deletion of the 9p21 locus, containing CDKN2A/B and MTAP [CDKN2A/MTAP deletion (CMdel)], underlies emerging therapeutic strategies, including PRMT5/MAT2A inhibition. The clinicogenomic context and treatment outcome associations of CMdel in non–small cell lung cancer (NSCLC) remain incompletely defined. Experimental Design: We analyzed Tempus Lens records with paired DNA (xT) and RNA (xR) sequencing. After quality filtering (purity ≥40%), 16,947 patients were included in the primary dataset, and 10,996 were included in the validation dataset and stratified by deletion status [CMdel vs. CM wild-type (CMwt)]. IFN status was defined by IFN-kappa (IFNK) expression [log2(TPM + 1)] using the 25th percentile of CMwt as the threshold. Immune composition was inferred by quanTIseq. First-line regimens were grouped as immune checkpoint inhibitor (ICI) monotherapy, ICI chemotherapy, chemotherapy, or tyrosine kinase inhibitors (TKI). Best response and time to next treatment (TTNT) with risk-set adjustment and 18-month truncation were assessed. Results: CMdel occurred in 11.7% (n = 1,991) and 15% (n = 1,676), respectively, and was enriched in mucinous histology, stage IV disease, and never smokers. CMdel was associated with EGFR, TERT, and SMARCA4 alterations, whereas CMwt was associated with RB1, SETD2, and ARID2. More than 50% of CMdel tumors harbored targetable coalterations, enriched in EGFR/ALK/RET and relatively depleted in KRAS. CMdel tumors exhibited lower PD-L1, mutation burden, and immune infiltration. TTNT did not differ by CM status for ICI monotherapy or TKI but was shorter in CMdel for ICI chemotherapy and chemotherapy. Discordant and broad deletions were less common and had weaker TTNT association. IFNK-low expression marked immune-excluded subgroups with worse treatment outcomes. Conclusions: CDKN2A/MTAP loss defines a prevalent, genomically distinct NSCLC subset enriched in actionable drivers. CMdel associates with worse outcomes to ICI/chemotherapy-containing regimens. These findings highlight convergent biology at 9p21 and support biomarker-guided development of PRMT5/MAT2A inhibitors in MTAP-deleted NSCLC.

Abstract Immune checkpoint inhibitors (ICIs) can produce durable tumor control, yet many patients experience primary or acquired resistance, mixed lesion-level responses, or immune-related adverse events. Tissue and blood biomarkers provide biologically informative but spatially incomplete views of a heterogeneous and time-varying tumor–host system. Radiologic imaging complements these assays by repeatedly measuring whole-body tumor burden, organ-specific disease patterns, lesion-level heterogeneity, and treatment-related change. This critical narrative review evaluates imaging-centered artificial intelligence (AI) for precision cancer immunotherapy, including baseline and serial CT, MRI, and PET; radiomics and learned image representations; lesion-level and whole-body heterogeneity; toxicity assessment; and integration with pathology, genomics, circulating tumor DNA, laboratory measurements, and electronic health records. Representative cohorts range from early radiogenomic studies of approximately 100–200 patients to externally validated and longitudinal models involving hundreds or thousands of patients. Rather than treating all reported performance estimates as equally actionable, we appraise the clinical question, cohort design, endpoint, validation strategy, calibration, and intended clinical action. Current evidence supports noninvasive immune phenotyping, risk stratification, longitudinal monitoring, and multimodal biomarker integration. However, most models predict outcomes among patients already receiving ICIs and therefore remain prognostic under treatment rather than predictive of differential treatment benefit. Radiotherapy (RT) combined with ICI serves as a focused imaging use case because lesion identity, dose distribution, fractionation, sequence, and thoracic and normal-tissue exposure must be linked to defined actions. Existing patient-level AI studies estimate efficacy or toxicity after treatment selection; none has prospectively shown that an AI-selected RT strategy improves patient outcomes. Translation will require patient-level leakage control, calibration, geographic and temporal validation, robustness to missing modalities, and prospective evaluation of prespecified model-guided decisions.

Introduction Community-acquired pneumonia (CAP) is common and remains the leading infectious cause of death worldwide. Antibiotics remain the mainstay of treatment, but organisms are often not identified, necessitating empirical treatment. Current treatment guidelines recommend combination therapy with beta-lactams and macrolides or monotherapy with fluoroquinolones for non-severe CAP. However, rising antibiotic resistance and safety concerns have led to increased doxycycline use, despite limited evidence supporting its efficacy. Our goal is to conduct a pragmatic, open-label, randomised clinical trial comparing the effectiveness of azithromycin with that of doxycycline in combination with beta-lactam therapy in adult patients hospitalised for CAP. Methods and analysis Patients presenting to the emergency department or directly referred to the hospital with a clinical presentation concerning for CAP and need for hospitalisation are eligible. We plan to enrol 1120 adults hospitalised at six hospitals across four states, including rural, suburban and urban settings and academic and community hospitals. Randomisation will be 1:1 using an electronic health record (EHR)-embedded screening and randomisation tool triggered by the treating clinician’s orders for azithromycin or doxycycline. Outcomes will be ascertained pragmatically via EHR-based data retrieval and will not require additional study visits. Our primary outcome is days alive and out of the hospital through day 28. Secondary outcomes include oxygen-free days; a composite of a need for advanced respiratory support or in-hospital mortality with analyses of each component separately; an ordinal measure of clinical deterioration defined by a prespecified severity scale; and 60-day and 180-day mortality. Safety outcomes are the incidence of Clostridioides difficile infection and QT prolongation. Analyses will follow an intention-to-treat framework, with per-protocol analyses for sensitivity. The primary endpoint will be compared between groups using a stratified van Elteren test, with supportive analyses using proportional odds regression and analysis of covariance adjusted for key covariates. Ethics and dissemination The Mayo Clinic Institutional Review Board reviewed and approved this protocol with a waiver of written consent, requiring verbal consent to be obtained by the clinician initiating the antibiotics (IRB# 24–0 06 503). Findings will be disseminated through peer-reviewed publications and international conference presentations. Trial registration number NCT07164131 .

Background Mild cognitive impairment (MCI) represents a critical window for both pharmacological therapies and non-pharmacological interventions to delay cognitive decline. Objective To evaluate whether combining pharmacotherapy with multicomponent behavioral interventions enhances outcomes relative to behavioral interventions alone. Methods The sample included people with MCI (pwMCI) of any etiology who completed the HABIT Healthy Action to Benefit Independence & Thinking ® program. Baseline pharmacotherapy (acetylcholinesterase inhibitor and/or glutamate regulators) use was dichotomized (yes/no). Multivariate linear mixed-effects models examined changes from baseline in care partner-reported everyday functional skills, quality of life, self-efficacy, anxiety, depression, and HABIT compliance at post-intervention, 6 months, and 12 months, adjusting for baseline cognitive impairment severity and relevant demographic covariates. Results Among 1074 pwMCI, 413 (38.5%) were receiving pharmacotherapy at baseline. Pharmacotherapy use was more prevalent among White than non-White participants (39.4% versus 23.3%, p = 0.01). At baseline, pharmacotherapy users had more severe cognitive impairment and lower utilization of rehabilitation strategies (all p < 0.05). Both groups demonstrated significant improvements across all outcomes at post-intervention. During follow-up (mean score change compared to baseline), pharmacotherapy users showed greater increases in HABIT compliance at 6 months (3.2 versus 2.0) and 12 months (2.1 versus 1.4). In contrast, by 12 months, pharmacotherapy users experienced greater worsening of everyday functional skills related to memory (−2.1 versus −1.3) and executive function (−5.8 versus-3.7). Conclusions pwMCI undergoing pharmacotherapy entered the HABIT program with more advanced cognitive impairment. The pharmacotherapy was not associated with greater intervention-related benefit or attenuation of long-term cognitive decline.

Background/Objective: To externally validate our previously published machine learning model for identifying hospitalized patients with community-acquired pneumonia (CAP) at low risk of respiratory deterioration or death using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Methods: This is a retrospective cohort study of adult patients (≥18 years) who were admitted with CAP and acute hypoxemic respiratory failure using the publicly available MIMIC-IV critical care dataset (Beth Israel Deaconess Medical Center, Boston, MA). We conducted an external validation study of a previously developed gradient boosting machine (GBM) model without recalibration using data available within the first 6 h of hospital admission. Results: For the primary composite outcome (need for advanced respiratory support [high flow nasal cannula (HFNC), non-invasive mechanical ventilation (NIMV), invasive mechanical ventilation (IMV)] or in-hospital death), the gradient boosting model demonstrated comparable performance in the derivation and external validation cohorts. The area under the receiver operating characteristic curve (AUC) was 0.713 in the Mayo cohort (n = 4379) and 0.689 in the MIMIC-IV cohort. Accuracy was 0.612 (95% confidence interval [CI], 0.595–0.628) versus 0.606 (95% CI 0.600–0.611), specificity 0.574 versus 0.523, sensitivity 0.723 versus 0.754, NPV 0.860 versus 0.842, and PPV 0.364 versus 0.401, respectively. For secondary outcomes, model discrimination was comparable between cohorts. The AUC for in-hospital mortality was 0.727 in the Mayo Cohort versus 0.733 in MIMIC-IV; for IMV, 0.736 versus 0.724; and for NIMV, 0.732 versus 0.708. Conclusions: Our machine learning algorithm demonstrated good discrimination and a high negative predictive value for identifying low-risk hospitalized CAP patients for respiratory deterioration or death in the external MIMIC-IV dataset, supporting its potential utility for prognostic enrichment in pneumonia clinical trials.

Background: Rural communities face worsening surgical access because of hospital closures and workforce shortages. Although osteopathic physicians have historically practiced in rural settings, their contribution to the contemporary rural surgical workforce remains poorly defined. Study Design: Cross-sectional analysis of 74,404 US attending surgeons across seven specialties using the 2023 Centers for Medicare & Medicaid Services Doctors and Clinicians database. Practice addresses were linked to 2020 Rural-Urban Commuting Area codes, Area Deprivation Index quintiles, and census tract population density. Rural practice (mean Rural-Urban Commuting Area score ≥4) was modeled using multivariable logistic regression adjusted for specialty, graduation cohort, sex, Census division, and state urbanization, with state-clustered standard errors. Results: Rural practice was more common among DO than MD surgeons (18.3% vs 9.7%; risk difference, 8.6 percentage points; 95% CI, 7.6–9.6). Across all specialties, DO surgeons practiced in more socioeconomically deprived communities. After adjustment, DO surgeons had twice the odds of rural practice compared with MD surgeons (OR, 2.05; 95% CI, 1.85–2.28). Rural areas had 13.8 surgeons per 100,000 residents versus 23.6 per 100,000 in metropolitan areas. Surgeons graduating in 2010 or later had approximately half the odds of rural practice compared with those graduating in 1979 or earlier (adjusted OR, 0.48; 95% CI, 0.42–0.56). Conclusions: Osteopathic surgeons disproportionately practice in rural and socioeconomically disadvantaged communities, identifying an important workforce component for maintaining rural surgical access. The declining likelihood of rural practice among more recent training cohorts raises concerns regarding the future rural surgical workforce.