McGinnis, Michele T

Background/Objectives: Novel vision-language models (VLMs) can integrate patient textual data with image data to support medical diagnosis. Recent studies reported conflicting results regarding the performance of multimodal VLMs compared to other models and physician performance. This systematic review aims to assess the diagnostic performance of multimodal VLMs integrating both patient textual and image data across diverse real-world hospital settings. Methods: We performed comprehensive searches of eight resources, including Embase, MEDLINE, and SCOPUS, on 17 December 2025. Eligible studies reporting diagnostic performance of VLMs integrating both image and patient history textual data from real-world adult patients compared to that of other models and physicians were included. The review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The Prediction model study Risk Of Bias Assessment Tool + AI (PROBAST + AI) was used to assess the quality and risk of bias. The study protocol was registered in the PROSPERO database (CRD420251244054). This review received no external funding. Results: We screened 11,026 records, of which 18 studies met the inclusion criteria. Six studies comparing multimodal and unimodal models demonstrated the consistent superiority of the multimodal models. Four studies evaluating VLM accuracy as standalone agents compared with physician performance reported conflicting evidence. One study assessing VLMs as a clinical copilot demonstrated higher accuracy from the group of physicians using VLM assistance. A meta-analysis could not be performed due to the heterogeneity across study populations and outcomes. The majority of the studies were assessed as having a high risk of bias due to dataset quality. Primary limitations identified across studies include small sample size, a lack of external validation, and the need for prospective clinical deployment studies. No study provided documented considerations regarding model safety or data security. Conclusions: This systematic review suggests that multimodal VLMs consistently outperform unimodal models with access to only image or text. While model performance as standalone agents compared to humans remains inconclusive, a copilot model has demonstrated high diagnostic accuracy. Given substantial methodological concerns across studies, cautious interpretation is required, No firm clinical recommendation can be made regarding the use of standalone VLMs. Further research employing high-quality datasets is needed to ensure the reliability and clinical applicability of future VLMs.

Abstract Background and Objectives Implementation of screening, brief intervention, and referral to treatment (SBIRT) for substance use in primary care remains challenging. Care coordination (CC) may strengthen SBIRT by supporting brief interventions, improving referral follow‐through, and enhancing continuity. CC models and outcomes vary, however, and no prior scoping review has synthesized these. Methods A comprehensive search of 10 resources, including Embase, MEDLINE, and PsycINFO, was conducted on 4/15/2025. Eligible studies described outpatient SBIRT models adding CC for any age group. Articles were excluded if they lacked CC, presented no original data, or were not in English. Relevant implementation and patient‐level outcomes were extracted to describe CC models and their impact. Results Of 737 abstracts and 50 full texts reviewed, 15 met the inclusion criteria. Studies spanned primary care, specialty care, and age ranges. CC models included embedded behavioral health providers, centralized linkage managers, and remote coordinators using phone or digital platforms. CC was associated with higher screening completion, brief intervention delivery, referral initiation, and treatment engagement. Successful integration incorporated structured communication, follow‐up protocols, and accessible technology. Barriers included limited billing infrastructure, weak electronic record integration, and unclear roles. Discussion and Conclusions CC appears to enhance SBIRT implementation and treatment engagement across outpatient settings. Embedded approaches offer relational benefits, while remote and digital strategies show promise for scalability. Comparative studies are needed to assess effectiveness, cost, and contextual fit. Scientific Significance Provides the first review of CC models within SBIRT, highlighting their implementation impact and guiding future optimization.

Background: Substance use disorders (SUDs) are highly prevalent, affecting over 48.5 million Americans. Available treatments for SUD remain insufficient, and many patients do not respond to existing interventions despite adequate adherence to treatments. While novel therapies for SUD are urgently needed, the use of psychedelic drugs for the treatment of SUDs has shown promise. Objectives: This overview of systematic reviews summarizes existing evidence on hallucinogens—serotonergic psychedelics and ketamine—for the treatment of SUD. Methods: A comprehensive search of the literature was conducted to identify relevant evidence for using serotonergic and non-serotonergic psychedelics for the treatment of SUDs. After initial screening (n = 468 studies), 62 studies were retrieved and assessed for eligibility, and a total of 16 systematic reviews were included. Conclusions: Although preliminary, evidence suggests that the use of serotonergic and non-serotonergic psychedelics for the treatment of SUD may provide advantages over traditional therapeutics, and these compounds may eventually become part of the next generation of treatments for SUD under specific circumstances. Research with these drugs has faced significant challenges, though, and caution when interpreting results is warranted, given high risk of bias and several other methodological limitations from the studies to date. Furthermore, risks associated with these drugs are not negligible. For now, the use of psychedelic drugs for the treatment of SUDs remains experimental, and existing evidence is insufficient to support its use in clinical practice.

Background: Invasive and non-invasive neuromodulation in psychiatry represents a burgeoning field that leverages advanced neuromodulation techniques to address substance use disorders (SUDs). Aims: This narrative review synthesizes findings from multiple reviews to evaluate the efficacy of neuromodulation in treating SUDs. Methods: A comprehensive literature search was conducted between December 2024 and April 2025, focusing on systematic reviews and meta-analyses that examined various neuromodulation modalities, including repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), and deep brain stimulation (DBS). The selected reviews were analyzed to identify common themes, outcomes, and gaps in the current understanding of these treatments for SUDs. Results: 11 reviews met the final inclusion criteria; 5 focused on non-invasive neuromodulation (rTMS, tDCS) and 6 on invasive neuromodulation (DBS). Non-invasive neurostimulation was associated with modest improvements in craving and cognitive dysfunction in individuals with SUDs. Similarly, invasive neuromodulation (DBS), through high-frequency stimulation of the bilateral nucleus accumbens, appeared to reduce cravings and improve comorbid psychiatric symptoms in both preclinical and human studies. Importantly, small sample sizes, heterogeneity in targets and stimulation protocols, and short follow-up periods significantly limit the generalizability of current findings from both non-invasive and invasive neuromodulation studies. Conclusions: As novel and more effective therapies for the treatment of SUD are desperately needed, procedural interventional psychiatry holds promise. However, despite encouraging results, existing evidence is still preliminary, and larger, rigorously designed studies are warranted to further establish the safety and efficacy of neuromodulatory interventions for SUD treatment.

Abstract Background Geographic cohorting (GCh) is a popular model of care that localizes physician teams to a single hospital unit. Objectives We conducted a scoping review identifying the aims, implementation strategies, study methods, and measured outcomes of GCh. Methods We searched the medical literature analysis and retrieval system online, Embase, and Scopus databases. Eligible citations included English language reports of interventions including GCh in adult patients. Included studies were screened for their aim, GCh implementation strategy, study method, and outcomes measured. Results Of 1863 identified citations, 30 met inclusion criteria, representing 27 interventions. Implementation aims varied from specific goals measured by a single metric to multiple outcomes intended to capture wide‐ranging effects of GCh. A majority of studies ( n = 13, 48.1%) used a pre‐post cohort design. GCh was implemented in one of four ways: (1) as a stand‐alone intervention, (2) bundled with accountable care unit elements, (3) bundled with care components that did not include all ACU elements, (4) GCh bundled with ACU elements and additional components (“enhanced” ACU). The measured outcomes sorted to eight different categories: patient outcomes, patient safety, patient experience, work flow, workload, clinician experience, communication/team work, and cost. Conclusions The current literature on GCh describes implementation as both a stand‐alone intervention and bundled with other care elements. Current research has not delineated whether the degree to which GCh is implemented matters, nor what impact it has as part of a bundled care intervention. Future work would benefit from a prospective design that clarifies these questions, facilitating care models tailored to the needs of the practice ecosystem.

ABSTRACTOpioid use disorder is a persistent epidemic despite several FDA‐approved medications for its treatment. While the pathogenesis of opioid use disorder has been classically attributed to dopamine pathways in the brain, there is emerging evidence and interest surrounding the role of inflammation and inflammatory signaling in its development and treatment. Buprenorphine has become the most prescribed medication for opioid use disorder, largely due to its ease of access and tolerability. This review aimed to better characterize contemporary knowledge of how buprenorphine modulates the human immune system and inflammatory functions in this population. A comprehensive review was conducted using 11 key databases, including Embase, MEDLINE, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov. This review captured 8177 records, and 14 studies were ultimately selected for inclusion and discussion in this review. Notably, all 14 clinical studies evaluated buprenorphine's effect on the peripheral immune system, and the majority of the studies supported the notion that initiation and maintenance of buprenorphine restore immune suppression caused by opioid use disorder. In addition, we discuss how recent and ongoing work utilizing advanced imaging and cellular technologies is advancing the understanding of how buprenorphine affects the immune and inflammatory signaling in the brain.

Mortality rates from pulmonary embolism (PE) remain significant, highlighting the need for alternative treatment strategies beyond traditional anticoagulation. Percutaneous interventions, including mechanical thrombectomy and catheter-directed thrombolysis, are emerging as promising options. Given the complex pathophysiology and unique risk profiles of these patients, meticulous multidisciplinary planning is essential. Anesthesiologists play a central role in coordinating care and managing perioperative risks to improve outcomes. This article provides insights into best safety practices and shares experiences from a leading quaternary center. It offers guidance for anesthesia providers to proactively engage in comprehensive risk stratification, participate in multidisciplinary discussions, and support robust contingency planning for managing PE patients undergoing percutaneous interventions in non-operating room anesthesia settings.

Abstract Cardiac implantable electronic device infection in the context of corynebacterial bloodstream infection (BSI) remains poorly understood. From 2012 to 2023 at Mayo Clinic, 4 of 12 patients with corynebacterial BSI had cardiac implantable electronic device infection: 1 patient was diagnosed during a relapsing BSI episode. Undefined source, persistent BSI, and the presence of a prosthetic cardiac valve were common characteristics.