Malaz Boustani

Background: Postoperative delirium is common among older adults after major surgery, especially those with cognitive impairment, and is associated with prolonged hospitalization, functional decline, mortality, and subsequent cognitive deterioration. Nonpharmacologic delirium-prevention strategies are recommended but are difficult to deliver reliably in high-acuity environments because they require repeated staff-intensive interactions, consistent patient engagement, and integration into complex intensive care unit (ICU) workflows. Objective: This study aims to evaluate the feasibility, usability, safety, acceptance, and adherence of ReCognitionVR 2.0, an immersive virtual reality (VR)–based cognitive stimulation intervention designed for older surgical ICU patients with mild-to-moderate cognitive impairment. Methods: This is a single-center, 3-arm, parallel-group randomized pilot trial. Ninety adults aged 65 years or older undergoing major surgery with expected postoperative ICU admission and Montreal Cognitive Assessment scores of 15 to 24 will be randomized in a 1:1:1 ratio to ReCognitionVR 2.0 cognitive stimulation, passive VR control exposure, or standard-of-care nurse-led reality orientation. VR sessions will be delivered once daily for up to 20 minutes on postoperative days 1 through 3 when participants are calm and alert. The primary outcomes are feasibility metrics, including recruitment, outcome completion, safety and tolerability, usability, acceptance, and adherence. Secondary outcomes estimate preliminary changes in attention and executive function using the Months Backward Test and NIH Toolbox Flanker Inhibitory Control and Attention Test. Postoperative delirium incidence is an exploratory clinical outcome. Results: This study was funded in September 2025. Participant recruitment and data collection are currently ongoing. Preliminary data quality assessments and exploratory reviews of the collected data are underway. Formal statistical analyses of the trial outcomes will be conducted after participant enrollment is completed and the study database has been finalized and locked. As of July 29, 2026, we had enrolled 18 participants. Early operational experience informed the refinement of exploratory VR performance metrics and the documentation of protocol deviations. Following an oxygen-connection near miss, we added a preintervention oxygen/readiness checklist as an operational safety safeguard. Enrollment of 90 participants is projected to be completed in 2027 if the current recruitment pace is sustained. Results are expected to be published in winter 2027. Conclusions: This protocol will generate data on feasibility, safety, usability, adherence, digital performance, cognitive assessment, and delirium measurement, all needed to optimize ReCognitionVR 2.0 and to design a future multicenter efficacy trial of VR-enabled delirium prevention in critical care.

This article reflects on the life, illness, and legacy of geriatrician and scientist Dr. James “Jim” Rudolph by using his medical and personal journey to examine the limitations of population‐based medicine. Population evidence establishes necessary baselines, but frequently fails to capture the complexity of the individualized reality of patients and their family members as they navigate a serious illness. Agile Science and Agile Analytic frameworks can help bridge this gap by leveraging population data as a starting point while utilizing small scale adaptive experiments to tailor and adjust care to the specific goals, values, and needs of an individual patient. Modern healthcare must evolve beyond rigid protocols to build adaptive networks capable of humanizing clinical care and reducing the systemic burdens placed on patients and caregivers, and Agile Science and Analytics are a path to this evolution.

UNSTRUCTURED Digital therapeutics are digital technology-based interventions designed to treat, manage, or prevent medical or behavioral conditions. Early products delivered fixed educational, behavioral, or monitoring content. Newer products increasingly resemble adaptive software ecosystems that incorporate large language models, multimodal real-time data, memory, retrieval, tool use, and multi-agent coordination. Such systems change in two distinct ways. The first is planned: a team updates the model, data, configuration, or permissions. Regulatory mechanisms such as predetermined change control plans govern these updates. The second is not planned. An agentic system can act differently at runtime with nothing newly deployed, because retrieval surfaces different context, memory accumulates, or agents select different tools. These changes cannot be enumerated in advance, and neither clinical stage models nor software governance frameworks specify how to develop, evaluate, or retire products that exhibit unplanned behavior. We propose an Agile Stage Model for minimally viable development and governance of digital therapeutics, including those built on agentic artificial intelligence. The model draws on the NIH Stage Model, software-as-a-medical device principles, Agile Science and its derivatives, the artificial intelligence agent literature, intelligent delegation, and recent work on agentic artificial intelligence for executing clinical trials. It extends the Agile Analytic Engine, an architecture that connects population evidence to a local treatment-to-target decision contract specifying target, baseline, action, reassessment interval, and response rule, and then to a governed chain of minimally viable experiments. The model comprises six stages: minimum viable specification (Stage 0), minimum viable product and internal validation (Stage 1), first-in-human sensor-action testing (Stage 2), local real-world minimally viable experiment (Stage 3), population effectiveness and cross-site learning (Stage 4), and lifecycle stewardship and deimplementation (Stage 5). The stages are learning gates rather than a fixed sequence, and teams may advance, repeat, revise, pause, or stop. At each stage the team asks a single question: what is the smallest valid experiment that can inform the next decision? For agentic systems, every stage adds explicit requirements for bounded authority, human oversight, permissioning, verification, and monitoring that separates degradation in performance from change in system behavior. One rule governs the model: standardize only what protects patient safety and the integrity of evidence and let everything else adapt to context. Oversight should scale with the severity, irreversibility, and reach of potential harm rather than being uniform across products. The model is conceptual and requires prospective evaluation.

Background: Dementia is associated with increased mortality among patients with colorectal cancer (CRC), and they receive less invasive evaluations, chemotherapy, and surgery. The objectives of this study were to (1) compare demographics and treatment patterns among individuals with CRC with and without dementia, and (2) evaluate the relationship between dementia, surgery receipt or refusal, and survival. Methods: The Indiana Cancer Registry was reviewed (2000–2020) for adult patients with clinical Stage I–III CRC. The diagnosis of dementia was determined by the presence by ICD‐9/10 codes. Overall survival was estimated using the Kaplan–Meier method and compared between patients with and without dementia using the log‐rank test. To assess the extent to which the association between dementia and survival was mediated by surgery utilization, a regression‐based causal mediation analysis of surgery receipt and refusal was performed, using covariates chosen a priori considered possible confounders of diagnosis, treatment, and survival. Results: A total of 69,318 patients met the inclusion criteria; 1516 (2.1%) had dementia and 67,802 (97.8%) did not. The dementia cohort had a higher percentage of female patients (55% vs. 49%, p < 0.001) and was older, with a median age at diagnosis of 79 (vs. 69, p < 0.001). Patients without dementia had higher survival probabilities across all time points (12, 36, 60, 120, and 180 months), and the survival difference was statistically significant ( p < 0.001). Receipt of surgery was a statistically significant mediator of survival in patients with dementia and CRC, with a proportion mediated of 9.6% (95% CI 6.2%–13.2%). Conclusion: Individuals with dementia and CRC receive lower rates of surgery with higher rates of refusal. Differences in surgical utilization are a significant mediator of survival differences, emphasizing the importance of patient‐centered goals‐of‐care discussions with patients and family members to assess the benefits and burdens of surgical treatment when planning.

BACKGROUND: Many injury survivors experience persistent biopsychosocial morbidity after hospital discharge. The trauma medical home (TMH), a multidisciplinary coordinated care model, demonstrated modest benefit in a multicenter randomized trial, but its effectiveness across rural and socioeconomically deprived populations remains unclear. METHODS: This post hoc retrospective analysis of the TMH parent randomized trial included patients aged older than or equal to 50 years with an Injury Severity Score ≥9 across four level I trauma centers. For this study, patients were categorized by rurality and Area Deprivation Index (ADI). Outcomes included the SF-36 Physical and Mental Component Scores (PCS, MCS), depression [Patient Health Questionnaire-9 (PHQ-9)], anxiety [Generalized Anxiety Disorder (GAD-7)], physical function [Short Physical Performance Battery (SPPB)], and health care utilization. Interaction effects between TMH and rural/urban status or ADI were evaluated at baseline, 6 months, and 12 months RESULTS: Four hundred thirty patients were included [216 TMH and 214 in usual care (UC)]. On the basis of urbanicity, the effectiveness of TMH on quality of life, psychological, physical, and health care utilization was similar at 6 and 12 months. ADI stratification showed greater differences. In high-ADI urban areas, baseline MCS scores were similar between TMH (47.5 ± 13.0) and UC (48.6 ± 12.8). By 6 months, TMH patients improved to 54.9 ± 8.8 compared with 49.0 ± 13.1 in UC, with a predicted difference of 7.7 (95% CI: 1.32, 12.83, P =0.016), demonstrating a clinically and statistically significant benefit of TMH CONCLUSIONS: The TMH was associated with greater mental health improvement among patients with high area deprivation, particularly in urban settings, suggesting a potential role in reducing disparities through coordinated care. LEVEL OF EVIDENCE: Prognostic and Epidemiological; Level III

Introduction: The Trauma Medical Home (TMH) randomized clinical trial tested the efficacy of a collaborative care intervention on the biopsychosocial recovery of older injured adults compared to usual care. Study enrollment occurred between 11 October 2017 and 13 October 2021 and study operations were impacted by the COVID-19 pandemic. It is important to measure the potential effects of COVID-19 on research study participants to ensure accurate reporting of results. The purpose of this secondary analysis was to compare depression and anxiety symptoms and health-related quality of life in older adults at TMH study enrollment pre-COVID-19 and during COVID-19. We hypothesized that those enrolled during COVID-19 would have worse symptoms and quality of life. Methods: This study was a secondary analysis of data from a multicenter randomized clinical trial (R01AG052493-01A1) TMH, that examined the ability of a collaborative care model to enhance the recovery of older injured adults compared to usual care. It was hypothesized that the TMH intervention would improve biopsychosocial recovery for injured older adults compared to usual care. The enrollment periods pre-COVID-19 and during COVID-19 were compared. Chi-square tests and Wilcoxon Rank Sum tests, then multivariable linear and logistic regressions were completed with SF-36 component scores and depression (PHQ-9) and anxiety (GAD-7) scores as the dependent variables, respectively. Results: A total of 429 participants were analyzed. There was no significant difference in anxiety and depression symptoms between the pre-COVID-19 and during-COVID-19 enrollment periods. Perceived limitations due to physical health problems were worse for those enrolled during COVID-19 (adjusted β = −9.8, 95% CI −18.3 to −1.4), but other health-related quality of life measures were similar between the two groups. Conclusions: These findings support the conclusion that the negative psychological impacts of the pandemic are likely due to individual factors other than age or COVID-19. These results suggest a high degree of psychological resiliency in older adults under stress. Future research should focus on the potential effects of socioeconomic status, demographic background, and level of physical functioning on psychological outcomes.

BACKGROUND Postoperative delirium is common among older adults after major surgery, especially those with cognitive impairment, and is associated with prolonged hospitalization, functional decline, mortality, and subsequent cognitive deterioration. Nonpharmacologic delirium-prevention strategies are recommended but are difficult to deliver reliably in high-acuity environments because they require repeated staff-intensive interactions, consistent patient engagement, and integration into complex ICU workflows. OBJECTIVE This study aims to evaluate the feasibility, usability, safety, acceptance, and adherence of ReCognitionVR 2.0, an immersive virtual reality (VR)-based cognitive stimulation intervention designed for older surgical ICU patients with mild-to-moderate cognitive impairment. METHODS This is a single-center, three-arm, parallel-group randomized pilot trial. Ninety adults aged 65 years or older undergoing major surgery with expected postoperative ICU admission and Montreal Cognitive Assessment scores of 15 to 24 will be randomized in a 1:1:1 ratio to ReCognitionVR 2.0 cognitive stimulation, passive VR control exposure, or standard-of-care nurse-led reality orientation. VR sessions will be delivered once daily for up to 20 minutes on postoperative days 1 through 3 when participants are calm and alert. The primary outcomes are feasibility metrics, including recruitment, outcome completion, safety and tolerability, usability, acceptance, and adherence. Secondary outcomes estimate preliminary changes in attention and executive function using the Months Backward Test and NIH Toolbox Flanker Inhibitory Control and Attention Test. Postoperative delirium incidence is an exploratory clinical outcome. RESULTS The modified protocol was approved as version 2 on April 13, 2026. Enrollment began on April 24, 2026. As of June 12, 2026, 18 participants had been enrolled, and early operational experience had informed prespecified implementation refinements, including clarification of VR performance metrics, documentation of protocol deviations, and addition of a preintervention oxygen/readiness checklist. Enrollment of 90 participants is projected to be completed in 2027 if the current recruitment pace is sustained. CONCLUSIONS This protocol will generate feasibility, safety, usability, adherence, digital performance, cognitive assessment, and delirium measurement data needed to optimize ReCognitionVR 2.0 and to design a future multicenter efficacy trial of VR-enabled delirium prevention in critical care. CLINICALTRIAL ClinicalTrials.gov [NCT07641153]; ClinicalTrials.gov PRS: Record Summary NCT07641153.

INTRODUCTION: Patients with decompensated cirrhosis have a high symptom burden and poor outcomes. Collaborative care models that provide coordinated, personalized care could improve outcomes. In this pilot randomized trial, we tested a cirrhosis-centric collaborative care model: the Cirrhosis Medical Home (CMH). METHODS: A single-center pilot randomized trial enrolling 40 hospitalized adults with decompensated cirrhosis randomized 1:1 to CMH or usual care. CMH involved 6 months of postdischarge individualized care. Primary outcomes were feasibility-based (enrollment, retention, data completeness). Secondary outcomes at 3 and 6 months included quality of life (SF-36) and healthcare utilization. RESULTS: Of 205 patients screened, 40 were enrolled. Of the 20 patients randomized to CMH, 13 received postdischarge CMH follow-up. At 3 months, 19 died or had a transplant, 9 were lost to follow-up, and 12 completed the SF-36. At 6 months, an additional 4 died or were lost to follow-up, and 8 completed the SF-36. In intention-to-treat analysis at 3 months, CMH did not improve quality of life. In per protocol analysis at 3 months, CMH improved physical functioning (delta +15 vs −10, P = 0.015), energy/fatigue (+20 vs −5, P = 0.02), and physical component score (+5.5 vs −5.0, P = 0.009). High mortality and readmission rates were seen in both arms but without significant differences. DISCUSSION: Enrollment and retention in a randomized trial of the CMH for posthospitalization management of decompensated cirrhosis is challenging, and patient-reported outcome assessment is limited by high rates of mortality, transplant, and loss to follow-up. These data can be used to inform future design and testing of health services interventions for this population.

Large language models (LLMs) have shown great potential for healthcare applications. However, existing evaluation benchmarks provide minimal coverage of Alzheimer's Disease and Related Dementias (ADRD). To address this gap, we introduce ADRD-Bench, the first ADRD-specific benchmark dataset designed for rigorous evaluation of LLMs. ADRD-Bench has two components: 1) ADRD Unified QA, a synthesis of 1,352 questions consolidated from seven established medical benchmarks, providing a unified assessment of clinical knowledge; and 2) ADRD Caregiving QA, a novel set of 149 questions derived from the Aging Brain Care (ABC) program, a widely used, evidence-based brain health management program. Guided by a program with national expertise in comprehensive ADRD care, this new set was designed to mitigate the lack of practical caregiving context in existing benchmarks. We evaluated 33 state-of-the-art LLMs on the proposed ADRD-Bench. Results showed that the accuracy of open-weight general models ranged from 0.63 to 0.93 (mean: 0.78; std: 0.09). The accuracy of open-weight medical models ranged from 0.48 to 0.93 (mean: 0.82; std: 0.13). The accuracy of closed-source general models ranged from 0.83 to 0.91 (mean: 0.89; std: 0.03). While top-tier models achieved high accuracies (>0.9), case studies revealed that inconsistent reasoning quality and stability limit their reliability, highlighting a critical need for domain-specific improvement to enhance LLMs' knowledge and reasoning grounded in daily caregiving data. The entire dataset is available at https://github.com/IIRL-ND/ADRD-Bench.

We present a scalable, AI-powered system that identifies and extracts evidence-based behavioral nudges from unstructured biomedical literature. Nudges are subtle, non-coercive interventions that influence behavior without limiting choice, showing strong impact on health outcomes like medication adherence. However, identifying these interventions from PubMed's 8 million+ articles is a bottleneck. Our system uses a novel multi-stage pipeline: first, hybrid filtering (keywords, TF-IDF, cosine similarity, and a "nudge-term bonus") reduces the corpus to about 81,000 candidates. Second, we use OpenScholar (quantized LLaMA 3.1 8B) to classify papers and extract structured fields like nudge type and target behavior in a single pass, validated against a JSON schema. We evaluated four configurations on a labeled test set (N=197). The best setup (Title/Abstract/Intro) achieved a 67.0% F1 score and 72.0% recall, ideal for discovery. A high-precision variant using self-consistency (7 randomized passes) achieved 100% precision with 12% recall, demonstrating a tunable trade-off for high-trust use cases. This system is being integrated into Agile Nudge+, a real-world platform, to ground LLM-generated interventions in peer-reviewed evidence. This work demonstrates interpretable, domain-specific retrieval pipelines for evidence synthesis and personalized healthcare.