Hasil untuk "Medical emergencies. Critical care. Intensive care. First aid"

E. Ely, R. Margolin, J. Francis et al.

Pierre Bardoult, Elodie Cadic, Olivier Brichory et al.

Abstract Background The purpose of this study was to identify the main greenhouse gas (GHG) emitting activities or products among the medical devices (MD) and medicines used in a polyvalent Intensive Care Unit (ICU). Methods A pragmatic eco-audit was conducted in a 21-beds polyvalent ICU, in Saint-Brieuc, Bretagne, France. It consisted of estimating GHG emissions of products or activities, considering process-based life cycle analysis (LCA), economic input–output analysis (EIO) and hybrid-LCA. Results were expressed as Carbon Dioxide Equivalent (CO2 e) emissions per patient-day considering each medication and MD (including personal protective equipment). Results With remaining uncertainty, GHG emissions were estimated at 61.1 kgCO2 e per patient-day. Two hundred and two individual MD were used per patient-day, equivalent to 5.1 kgCO2 e per patient-day (process-based LCA). Gloves accounted for the main part of kgCO2 e emissions (representing 1.8 kgCO2 e per patient-day). Then, syringes (1.1 kgCO2 e per patient-day), perfusion tubings (1.0 per patient-day) and gauze pads (0.4 kgCO2 e per patient-day) were the most important sources of MD related GHG emissions. Forty-seven individual medicines were used per patient-day. Most consumed medications were sterile water for injection, propofol, and sodium chlorure. The GHG emissions of medications were estimated with EIO-LCA at 21.5 kgCO2 e per patient-day, mostly due to injectable medicines (15.3 kgCO2 e per patient-day). Conclusion Upcoming studies focusing on actions on these particular hot spots would be of interest in order to significantly decrease GHG emissions but also to increase resilience of critical care.

Ankit Dutta, Nabarup Ghosh, Ankush Chatterjee

Large Language models have demonstrated excellent domain-specific question-answering capabilities when finetuned with a particular dataset of that specific domain. However, fine-tuning the models requires a significant amount of training time and a considerable amount of hardware. In this work, we propose CARE (Customer Assistance and Response Engine), a lightweight model made by fine-tuning Phi3.5-mini on very minimal hardware and data, designed to handle queries primarily across three domains: telecommunications support, medical support, and banking support. For telecommunications and banking, the chatbot addresses issues and problems faced by customers regularly in the above-mentioned domains. In the medical domain, CARE provides preliminary support by offering basic diagnoses and medical suggestions that a user might take before consulting a healthcare professional. Since CARE is built on Phi3.5-mini, it can be used even on mobile devices, increasing its usability. Our research also shows that CARE performs relatively well on various medical benchmarks, indicating that it can be used to make basic medical suggestions.

Junyi Fan, Li Sun, Negin Ashrafi et al.

Nursing documentation in intensive care units (ICUs) provides essential clinical intelligence but often suffers from inconsistent terminology, informal styles, and lack of standardization, challenges that are particularly critical in heart failure care. This study applies Direct Preference Optimization (DPO) to adapt Mistral-7B, a locally deployable language model, using 8,838 heart failure nursing notes from the MIMIC-III database and 21,210 preference pairs derived from expert-verified GPT outputs, model generations, and original notes. Evaluation across BLEU, ROUGE, BERTScore, Perplexity, and expert qualitative assessments demonstrates that DPO markedly enhances documentation quality. Specifically, BLEU increased by 84% (0.173 to 0.318), BERTScore improved by 7.6% (0.828 to 0.891), and expert ratings rose across accuracy (+14.4 points), completeness (+14.5 points), logical consistency (+14.1 points), readability (+11.1 points), and structural clarity (+6.0 points). These results indicate that DPO can align lightweight clinical language models with expert standards, supporting privacy-preserving, AI-assisted documentation within electronic health record systems to reduce administrative burden and improve ICU patient safety.

Jiaying Liu, Yan Zhang

Mental health care-seeking among marginalized young adults has received limited attention in CSCW research. Through in-depth interviews and visual elicitation methods with 18 diverse U.S. participants, our study reveals how marginalized identities shape mental health care-seeking journeys, often characterized by low aspirations and passive care-seeking influenced by lived experiences of marginalization. However, we found the transformative function of "care encounters" - serendipitous interactions with mental health resources that occur when individuals are not actively seeking support. These encounters serve as critical turning points, catalyzing shifts in aspiration and enabling more proactive care-seeking behaviors. Our analysis identifies both the infrastructural conditions that enable transformative care encounters and the aspiration breakdowns that impede care-seeking processes. This work makes conceptual contributions by supplementing traditional motivation-based care-seeking models with a reconceptualization of "care encounters" that accounts for the infrastructural and serendipitous nature of mental health access. We advance understanding of how marginalized identity uniquely influences care-seeking behaviors while providing actionable design implications for embedding technology-mediated "care encounters" into socio-technical interventions that can better support mental health care access for vulnerable populations.

Shyama Sastha Krishnamoorthy Srinivasan, Mohan Kumar, Pushpendra Singh

Personal Health Informatics (PHI), which leverages digital tools and information systems to support health assessment and self-care, promises more proactive, user-centered care, yet adoption and meaningful utilization barriers in India remain underexplored. Through a sequential mixed-methods study in urban India (Initial survey (n=87) and semi-structured interviews (n=22), follow-up survey = 116, and co-design workshops (n=6)), we surface practices, perceptions, and behaviors to identify ways PHI can be better utilized for proactive care in the Indian context. We find that PHI is valued for monitoring and enabling collective care; however, its adoption is constrained by low health and technology literacy, usability and integration issues, fragmented and costly technology ecosystems, and mistrust of digital health platforms. From triangulated evidence, we derive concrete design requirements, including user-controlled sharing, accessible analytics, and verifiable health information, and present a culturally grounded design vision for an integrated platform for collective care through design and evaluation of a figma prototype. The prototype evaluation provides further directions for design and development to better orient PHI for proactive care through the PHI-Proact operational map, which involves agency, elicitation, and engagement. Finally, using PHI-Proact, we conclude with concrete recommendations for designing and responsibly deploying PHI systems for proactive collective care in emerging contexts, which differ socially, culturally, infrastructurally, and technologically from WEIRD contexts.

R. Baron, A. Binder, R. Biniek et al.

In 2010, under the guidance of the DGAI (German Society of Anaesthesiology and Intensive Care Medicine) and DIVI (German Interdisciplinary Association for Intensive Care and Emergency Medicine), twelve German medical societies published the “Evidence- and Consensus-based Guidelines on the Management of Analgesia, Sedation and Delirium in Intensive Care”. Since then, several new studies and publications have considerably increased the body of evidence, including the new recommendations from the American College of Critical Care Medicine (ACCM) in conjunction with Society of Critical Care Medicine (SCCM) and American Society of Health-System Pharmacists (ASHP) from 2013. For this update, a major restructuring and extension of the guidelines were needed in order to cover new aspects of treatment, such as sleep and anxiety management. The literature was systematically searched and evaluated using the criteria of the Oxford Center of Evidence Based Medicine. The body of evidence used to formulate these recommendations was reviewed and approved by representatives of 17 national societies. Three grades of recommendation were used as follows: Grade “A” (strong recommendation), Grade “B” (recommendation) and Grade “0” (open recommendation). The result is a comprehensive, interdisciplinary, evidence and consensus-based set of level 3 guidelines. This publication was designed for all ICU professionals, and takes into account all critically ill patient populations. It represents a guide to symptom-oriented prevention, diagnosis, and treatment of delirium, anxiety, stress, and protocol-based analgesia, sedation, and sleep-management in intensive care medicine.

Qiao Pan, Zhaofang Mao

This paper addresses a realistic home health care and home care (HHC\&HC) problem which has become increasingly complex in the face of demographic aging and post-COVID-19 disruptions. The HHC\&HC sector, as the essential component of modern health care systems, faces unique challenges in efficiently scheduling and routing caregivers to meet the rising demand for home-based care services. Traditional approaches often fall short in addressing the dynamic nature of care requests, especially in accommodating new, same-day service requests without compromising scheduled visits. To tackle these issues, We define the problem as an HHC\&HC routing and rescheduling problem with rejection of new customers (HHC\&HCRRP-RNC), focusing on rescheduling for a single HHC\&HC caregiver in response to new customer requests within a single period. This problem is a variant of both the single-machine reschedule problem and the orienteering problem with mandatory visits (OPMV), where certain nodes must be visited while others are optional. A mixed integer linear programming (MILP) model is developed to cater to two groups of customers: pre-scheduled existing customers and same-day service new customers. The model emphasized maintaining minimal disruptions to the original schedule for existing customers as a constraint, highlighting the balance between adhering to scheduled visits and accommodating new customers. A hybrid memetic-Adaptive Neighborhood Search (ANS) optimization algorithm is proposed to tackle the model. This approach aims to minimize operational costs and opportunity costs while enhancing service quality and patient satisfaction. Through computational experiments, our proposed algorithm demonstrates notable performance, offering significant improvements in both efficiency and robustness within the problem domain.

Nur Yildirim, Susanna Zlotnikov, Aradhana Venkat et al.

Clinical practice guidelines, care pathways, and protocols are designed to support evidence-based practices for clinicians; however, their adoption remains a challenge. We set out to investigate why clinicians deviate from the ``Wake Up and Breathe'' protocol, an evidence-based guideline for liberating patients from mechanical ventilation in the intensive care unit (ICU). We conducted over 40 hours of direct observations of live clinical workflows, 17 interviews with frontline care providers, and 4 co-design workshops at three different medical intensive care units. Our findings indicate that unlike prior literature suggests, disagreement with the protocol is not a substantial barrier to adoption. Instead, the uncertainty surrounding the application of the protocol for individual patients leads clinicians to deprioritize adoption in favor of tasks where they have high certainty. Reflecting on these insights, we identify opportunities for technical systems to help clinicians in effectively executing the protocol and discuss future directions for HCI research to support the integration of protocols into clinical practice in complex, team-based healthcare settings.

Marceli Wac, Raul Santos-Rodriguez, Chris McWilliams et al.

Intensive Care Units are complex, data-rich environments where critically ill patients are treated using variety of clinical equipment. The data collected using this equipment can be used clinical staff to gain insight into the condition of the patients and provide adequate treatment, but it also provides ample opportunity for applications in machine learning and data science. While this data can frequently be used directly, complex problems may require additional annotations to provide context and meaning before it could be used to train the machine learning models. Annotating time-series datasets in clinical setting is a complex problem due to a large volume and complexity of the data, time-consuming nature of the process and the fact that clinicians' time is in both high demand and short supply. In this study, we present an evaluation of a bespoke tool designed to annotate large, clinical time-series datasets with staff from intensive care units. The software incorporates two modes for annotation: by annotating individual admissions and by generating rulesets which are applied to the entire dataset. Our study was split into two stages focusing on individual and semi-automated annotation and included 28 annotators across both stages who utilised 50 clinical parameters to guide their annotations. We experienced significant challenges in recruitment and engagement of the participants in the annotation activities and developed interventions which improved the participation over the course of the study. During the semi-automated annotation, we observed preferences for different parameter types (measured vs. observed), as well as relative agreement of participants across shared admissions to the decision-tree model trained using their rulesets.

Mohit Tomar, Abhisek Tiwari, Tulika Saha et al.

In recent times, there has been an increasing awareness about imminent environmental challenges, resulting in people showing a stronger dedication to taking care of the environment and nurturing green life. The current $19.6 billion indoor gardening industry, reflective of this growing sentiment, not only signifies a monetary value but also speaks of a profound human desire to reconnect with the natural world. However, several recent surveys cast a revealing light on the fate of plants within our care, with more than half succumbing primarily due to the silent menace of improper care. Thus, the need for accessible expertise capable of assisting and guiding individuals through the intricacies of plant care has become paramount more than ever. In this work, we make the very first attempt at building a plant care assistant, which aims to assist people with plant(-ing) concerns through conversations. We propose a plant care conversational dataset named Plantational, which contains around 1K dialogues between users and plant care experts. Our end-to-end proposed approach is two-fold : (i) We first benchmark the dataset with the help of various large language models (LLMs) and visual language model (VLM) by studying the impact of instruction tuning (zero-shot and few-shot prompting) and fine-tuning techniques on this task; (ii) finally, we build EcoSage, a multi-modal plant care assisting dialogue generation framework, incorporating an adapter-based modality infusion using a gated mechanism. We performed an extensive examination (both automated and manual evaluation) of the performance exhibited by various LLMs and VLM in the generation of the domain-specific dialogue responses to underscore the respective strengths and weaknesses of these diverse models.

A. Farcas, A. Joiner, Jordan S Rudman et al.

Abstract Background Emergency medical services (EMS) often serve as the first medical contact for ill or injured patients, representing a critical access point to the health care delivery continuum. While a growing body of literature suggests inequities in care within hospitals and emergency departments, limited research has comprehensively explored disparities related to patient demographic characteristics in prehospital care. Objective We aimed to summarize the existing literature on disparities in prehospital care delivery for patients identifying as members of an underrepresented race, ethnicity, sex, gender, or sexual orientation group. Methods We conducted a scoping review of peer-reviewed and non-peer-reviewed (gray) literature. We searched PubMed, CINAHL, Web of Science, Proquest Dissertations, Scopus, Google, and professional websites for studies set in the U.S. between 1960 and 2021. Each abstract and full-text article was screened by two reviewers. Studies written in English that addressed the underrepresented groups of interest and investigated EMS-related encounters were included. Studies were excluded if a disparity was noted incidentally but was not a stated objective or discussed. Data extraction was conducted using a standardized electronic form. Results were summarized qualitatively using an inductive approach. Results One hundred forty-five full-text articles from the peer-reviewed literature and two articles from the gray literature met inclusion criteria: 25 studies investigated sex/gender, 61 studies investigated race/ethnicity, and 58 studies investigated both. One study investigated sexual orientation. The most common health conditions evaluated were out-of-hospital cardiac arrest (n = 50), acute coronary syndrome (n = 36), and stroke (n = 31). The phases of EMS care investigated included access (n = 55), pre-arrival care (n = 46), diagnosis/treatment (n = 42), and response/transport (n = 40), with several studies covering multiple phases. Disparities were identified related to all phases of EMS care for underrepresented groups, including symptom recognition, pain management, and stroke identification. The gray literature identified public perceptions of EMS clinicians’ cultural competency and the ability to appropriately care for transgender patients in the prehospital setting. Conclusions Existing research highlights health disparities in EMS care delivery throughout multiple health outcomes and phases of EMS care. Future research is needed to identify structured mechanisms to eliminate disparities, address clinician bias, and provide high-quality equitable care for all patient populations.

E. Finkelsztein, Daniel S. Jones, Kevin C. Ma et al.

BackgroundThe Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) Task Force recently introduced a new clinical score termed quick Sequential (Sepsis-related) Organ Failure Assessment (qSOFA) for identification of patients at risk of sepsis outside the intensive care unit (ICU). We attempted to compare the discriminatory capacity of the qSOFA versus the Systemic Inflammatory Response Syndrome (SIRS) score for predicting mortality, ICU-free days, and organ dysfunction-free days in patients with suspicion of infection outside the ICU.MethodsThe Weill Cornell Medicine Registry and Biobank of Critically Ill Patients is an ongoing cohort of critically ill patients, for whom biological samples and clinical information (including vital signs before and during ICU hospitalization) are prospectively collected. Using such information, qSOFA and SIRS scores outside the ICU (specifically, within 8 hours before ICU admission) were calculated. This study population was therefore comprised of patients in the emergency department or the hospital wards who had suspected infection, were subsequently admitted to the medical ICU and were included in the Registry and Biobank.ResultsOne hundred fifty-two patients (67% from the emergency department) were included in this study. Sixty-seven percent had positive cultures and 19% died in the hospital. Discrimination of in-hospital mortality using qSOFA [area under the receiver operating characteristic curve (AUC), 0.74; 95% confidence intervals (CI), 0.66–0.81] was significantly greater compared with SIRS criteria (AUC, 0.59; 95% CI, 0.51–0.67; p = 0.03). The qSOFA performed better than SIRS regarding discrimination for ICU-free days (p = 0.04), but not for ventilator-free days (p = 0.19), any organ dysfunction-free days (p = 0.13), or renal dysfunction-free days (p = 0.17).ConclusionsIn patients with suspected infection who eventually required admission to the ICU, qSOFA calculated before their ICU admission had greater accuracy than SIRS for predicting mortality and ICU-free days. However, it may be less clear whether qSOFA is also better than SIRS criteria for predicting ventilator free-days and organ dysfunction-free days. These findings may help clinicians gain further insight into the usefulness of qSOFA.

Barbora Bircak-Kuchtova, Ha-Yeun Chung, Jonathan Wickel et al.

Abstract Sepsis is the most common cause of admission to intensive care units worldwide. Sepsis patients frequently suffer from sepsis-associated encephalopathy (SAE) reflecting acute brain dysfunction. SAE may result in increased mortality, extended length of hospital stay, and long-term cognitive dysfunction. The diagnosis of SAE is based on clinical assessments, but a valid biomarker to identify and confirm SAE and to assess SAE severity is missing. Several blood-based biomarkers indicating neuronal injury have been evaluated in sepsis and their potential role as early diagnosis and prognostic markers has been studied. Among those, the neuroaxonal injury marker neurofilament light chain (NfL) was identified to potentially serve as a prognostic biomarker for SAE and to predict long-term cognitive impairment. In this review, we summarize the current knowledge of biomarkers, especially NfL, in SAE and discuss a possible future clinical application considering existing limitations.

David Bar-Or, Constance McGraw, Muhammad Darwish et al.

Objectives Falling from height may lead to significant injuries and time hospitalized; however, there are few studies comparing the specific mechanism of fall. The purpose of this study was to compare injuries from falls after attempting to cross the USA-Mexico border fence (intentional) with injuries from domestic falls (unintentional) of comparable height.Methods This retrospective cohort study included all patients admitted after a fall from a height of 15–30 ft to a level II trauma center between April 2014 and November 2019. Patient characteristics were compared by falls from the border fence with those who fell domestically. Fisher’s exact test, χ2 test and Wilcoxon Mann-Whitney U test were used as appropriate. A significance level of α<0.05 was used.Results Of the 124 patients included, 64 (52%) were falls from the border fence while 60 (48%) were domestic falls. Patients sustaining injuries from border falls were on average younger than patients who had domestic falls (32.6 (10) vs 40.0 (16), p=0.002), more likely males (58% vs 41%, p<0.001), fell from a significantly higher distance (20 (20–25) vs 16.5 (15–25), p<0.001), and had a significantly lower median injury severity score (ISS) (5 (4–10) vs 9 (5–16.5), p=0.001). Additionally, compared with domestic falls, border falls had fewer injuries to the head (3% vs 25%, p=0.004) and chest (5% vs 27%, p=0.007), yet more extremity injuries (73% vs 42%, p=0.003), and less had an intensive care unit (ICU) stay (30% vs 63%, p=0.002). No significant differences in mortality were found.Conclusion Patients sustaining injuries from border crossing falls were slightly younger, and although fell from higher, had a lower ISS, more extremity injuries, and fewer were admitted to the ICU compared with patients sustaining falls domestically. There was no difference in mortality between groups.Level of evidence Level III, retrospective study.

Yan Zhang, Xiaomin Lu, Haoming Ji et al.

Objectives. To analyze the effects of deep hyperthermia combined with intraperitoneal chemotherapy on liver-kidney function, immune function, and long-term survival in patients with abdominal metastases. Methods. A total of 88 patients with abdominal metastases confirmed in the hospital were enrolled as the research objects between August 2018 and August 2021. They were randomly divided into control group (n = 44) and observation group (n = 44). The control group was treated with intraperitoneal chemotherapy, while observation group was additionally treated with deep hyperthermia. The general clinical data of patients were recorded. The short-term and long-term curative effects were evaluated. The occurrence of side effects in both groups was recorded. Before and after treatment, levels of alanine transaminase (ALT) and aspartate transaminase (AST) were detected by full-automatic biochemical analyzer. The level of blood urea nitrogen (BUN) was detected by the urease electrode method. The level of serum creatinine (Scr) was detected by the picric acid method. The levels of CD3+, CD4+, CD8+, and NK cells were detected by BD FACSCalibur flow cytometer. Results. There was no significant difference in clinical data between the two groups (P>0.05). In the observation group, ORR was significantly higher than that in the control group (54.55% vs 29.55%) (P<0.05), OS was significantly longer than that in the control group (P<0.05), and median survival time and mPFS were longer than those in the control group. After treatment, the levels of ALT, AST, BUN, and Scr were significantly increased in the control group (P<0.05), but there was no significant difference in peripheral blood CD3+, CD4+, and CD4+/CD8+ ratio or count of NK cells before and after treatment (P>0.05). Before and after treatment, there was no significant difference in the levels of ALT, AST, BUN, and Scr in the observation group (P>0.05). After treatment, peripheral blood CD3+, CD4+, and CD4+/CD8+ ratio and count of NK cells were all increased in the observation group, significantly higher than those in the control group (P<0.05). The incidence of chemotherapy side effects in the observation group was significantly lower than that in the control group (P<0.05). Conclusion. The short-term and long-term curative effects of deep hyperthermia combined with intraperitoneal chemotherapy are good on patients with intraperitoneal metastases, with less damage to liver-kidney function. It is beneficial to enhance immune function of patients, with mild side effects.

Sung-Min Cho, Christopher Wilcox, Steven Keller et al.

Abstract Background To assess the safety and feasibility of imaging of the brain with a point-of-care (POC) magnetic resonance imaging (MRI) system in patients on extracorporeal membrane oxygenation (ECMO). Early detection of acute brain injury (ABI) is critical in improving survival for patients with ECMO support. Methods Patients from a single tertiary academic ECMO center who underwent head CT (HCT), followed by POC brain MRI examinations within 24 h following HCT while on ECMO. Primary outcomes were safety and feasibility, defined as completion of MRI examination without serious adverse events (SAEs). Secondary outcome was the quality of MR images in assessing ABIs. Results We report 3 consecutive adult patients (median age 47 years; 67% male) with veno-arterial (n = 1) and veno-venous ECMO (n = 2) (VA- and VV-ECMO) support. All patients were imaged successfully without SAEs. Times to complete POC brain MRI examinations were 34, 40, and 43 min. Two patients had ECMO suction events, resolved with fluid and repositioning. Two patients were found to have an unsuspected acute stroke, well visualized with MRI. Conclusions Adult patients with VA- or VV-ECMO support can be safely imaged with low-field POC brain MRI in the intensive care unit, allowing for the assessment of presence and timing of ABI.

Konstantinos Dimitriadis, Jan Meis, Hermann Neugebauer et al.

Abstract Background Neurologic manifestations are increasingly reported in patients with coronavirus disease 2019 (COVID-19). Yet, data on prevalence, predictors and relevance for outcome of neurological manifestations in patients requiring intensive care are scarce. We aimed to characterize prevalence, risk factors and impact on outcome of neurologic manifestations in critically ill COVID-19 patients. Methods In the prospective, multicenter, observational registry study PANDEMIC (Pooled Analysis of Neurologic DisordErs Manifesting in Intensive care of COVID-19), we enrolled COVID-19 patients with neurologic manifestations admitted to 19 German intensive care units (ICU) between April 2020 and September 2021. We performed descriptive and explorative statistical analyses. Multivariable models were used to investigate factors associated with disorder categories and their underlying diagnoses as well as to identify predictors of outcome. Results Of the 392 patients included in the analysis, 70.7% (277/392) were male and the mean age was 65.3 (SD ± 3.1) years. During the study period, a total of 2681 patients with COVID-19 were treated at the ICUs of 15 participating centers. New neurologic disorders were identified in 350 patients, reported by these centers, suggesting a prevalence of COVID-19-associated neurologic disorders of 12.7% among COVID-19 ICU patients. Encephalopathy (46.2%; 181/392), cerebrovascular (41.0%; 161/392) and neuromuscular disorders (20.4%; 80/392) were the most frequent categories identified. Out of 35 cerebrospinal fluid analyses with reverse transcriptase PCR for SARS-COV-2, only 3 were positive. In-hospital mortality was 36.0% (140/389), and functional outcome (mRS 3 to 5) of surviving patients was poor at hospital discharge in 70.9% (161/227). Intracerebral hemorrhage (OR 6.2, 95% CI 2.5–14.9, p < 0.001) and acute ischemic stroke (OR 3.9, 95% CI 1.9–8.2, p < 0.001) were the strongest predictors of poor outcome among the included patients. Conclusions Based on this well-characterized COVID-19 ICU cohort, that comprised 12.7% of all severe ill COVID-19 patients, neurologic manifestations increase mortality and morbidity. Since no reliable evidence of direct viral affection of the nervous system by COVID-19 could be found, these neurologic manifestations may for a great part be indirect para- or postinfectious sequelae of the infection or severe critical illness. Neurologic ICU complications should be actively searched for and treated.

Maurilio Gutzeit, Johannes Rauh, Maximilian Kähler et al.

Despite the ongoing strong interest in associations between quality of care and the volume of health care providers, a unified statistical framework for analyzing them is missing, and many studies suffer from poor statistical modelling choices. We propose a flexible, additive mixed model for studying volume-outcome associations in health care that takes into account individual patient characteristics as well as provider-specific effects through a multi-level approach. More specifically, we treat volume as a continuous variable, and its effect on the considered outcome is modelled as a smooth function. We take account of different case-mixes by including patient-specific risk factors and of clustering on the provider level through random intercepts. This strategy enables us to extract a smooth volume effect as well as volume-independent provider effects. These two quantities can be compared directly in terms of their magnitude, which gives insight into the sources of variability of quality of care. Based on a causal DAG, we derive conditions under which the volume-effect can be interpreted as a causal effect. The paper provides confidence sets for each of the estimated quantities relying on joint estimation of all effects and parameters. Our approach is illustrated through simulation studies and an application to German health care data. Keywords: health care quality measurement, volume-outcome analysis, minimum provider volume, additive regression models, random intercept

Varun A. Kelkar, Dimitrios S. Gotsis, Frank J. Brooks et al.

Modern generative models, such as generative adversarial networks (GANs), hold tremendous promise for several areas of medical imaging, such as unconditional medical image synthesis, image restoration, reconstruction and translation, and optimization of imaging systems. However, procedures for establishing stochastic image models (SIMs) using GANs remain generic and do not address specific issues relevant to medical imaging. In this work, canonical SIMs that simulate realistic vessels in angiography images are employed to evaluate procedures for establishing SIMs using GANs. The GAN-based SIM is compared to the canonical SIM based on its ability to reproduce those statistics that are meaningful to the particular medically realistic SIM considered. It is shown that evaluating GANs using classical metrics and medically relevant metrics may lead to different conclusions about the fidelity of the trained GANs. This work highlights the need for the development of objective metrics for evaluating GANs.