Ethical and regulatory challenges of large language models in medicine.

With the rapid growth of interest in and use of large language models (LLMs) across various industries, we are facing some crucial and profound ethical concerns, especially in the medical field. The unique technical architecture and purported emergent abilities of LLMs differentiate them substantially from other artificial intelligence (AI) models and natural language processing techniques used, necessitating a nuanced understanding of LLM ethics. In this Viewpoint, we highlight ethical concerns stemming from the perspectives of users, developers, and regulators, notably focusing on data privacy and rights of use, data provenance, intellectual property contamination, and broad applications and plasticity of LLMs. A comprehensive framework and mitigating strategies will be imperative for the responsible integration of LLMs into medical practice, ensuring alignment with ethical principles and safeguarding against potential societal risks.

Deep learning for precision medicine: Guiding laser therapy in ischemic retinal diseases.

In this issue of Cell Reports Medicine, Zhao and colleagues1 report a multi-tasking artificial intelligence system that can assist the whole process of fundus fluorescein angiography (FFA) imaging and reduce the reliance on retinal specialists in FFA examination.

Impact of Age of Packed RBC Transfusion on Oxygenation in Patients Receiving Extracorporeal Membrane Oxygenation.

OBJECTIVE: To determine the effect of age of packed RBCs on tissue oxygenation in children receiving extracorporeal membrane oxygenation support. DESIGN: A retrospective study was done between March 2013 and August 2015. The following biomarkers were examined 6 hours before and 6 hours after the conclusion of a transfusion: serum and circuit hematocrits, serum and circuit venous saturations, serum lactate levels (mg/dL), and cerebral saturation via near-infrared spectroscopy. Biomarkers were examined with respect to time relative to transfusion using four discrete categories (< 3, -3 to 0, 0-3, and > 3 hr). The association between age of blood transfusion and change in biomarkers was accessed analyzing time relative to transfusion as described above. In addition, the age of blood transfusion was analyzed similarly, using four discrete categories (0-7, 7-14, 14-21, and > 21 d). SETTING: Twenty-four bed mixed pediatric medical and cardiac ICU at a tertiary care center. SUBJECTS: Zero- to 18-year-old patients of required extracorporeal membrane oxygenation support. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Circuit venous saturation demonstrated an increase of 2.5% (p < 0.001) in first 3 hours posttransfusion. This was followed by a 1.4% decrease after the initial 3 hours posttransfusion. Serum venous saturation showed no statistically significant change with relation to transfusions. Neither lactate levels nor near-infrared spectroscopy demonstrated any observed statistical change with relation to transfusion. With regards to the relationship between the age of RBC transfusion and tissue oxygenation biomarkers, none of the biomarkers exhibited a consistent interaction. CONCLUSIONS: Our study demonstrates that the age of packed RBC transfusion does not affect the degree tissue oxygenation in children receiving extracorporeal membrane oxygenation support, as measured by mixed venous oxygen saturation, lactate, and near-infrared spectroscopy. In addition, packed RBC transfusion, in general, did not produce any meaningful change in these markers of tissue oxygenation.

Sepsis at a Safety Net Hospital: Risk Factors Associated With 30-Day Readmission.

BACKGROUND: Sepsis is a leading cause of hospitalization, and subsequent readmissions are frequent and costly. There is an expanding body of literature describing risk factors for readmissions in patients with sepsis. However, there are little data studying medically underserved patients who typically receive their care at a safety net hospital. METHODS: In a retrospective cohort study, we evaluated 1355 sepsis survivors at risk of hospital readmission in fiscal year 2013 at a safety net hospital. We described patient characteristics during their initial and readmission hospitalizations and analyzed risk factors associated with 30-day readmission. RESULTS: The 30-day readmission rate among sepsis survivors was 22.6%. Comorbid conditions associated with readmissions included end-stage renal disease (odds ratio [OR], 1.26; 95% confidence interval [CI], 1.17-1.36), malignancy (OR, 1.14; 95% CI, 1.08-1.21), and cirrhosis (OR, 1.11; 95% CI, 1.02-1.20). Bacteremia during the initial hospitalization (OR, 1.07; 95% CI, 1.01-1.15) and being discharged with a vascular catheter (OR, 1.10; 95% CI, 1.01-1.20) were associated with readmission. Less severe sepsis during the initial hospitalization was associated with a reduced risk of 30-day readmission (OR, 0.91; 95% CI, 0.87-0.94). CONCLUSIONS: At a safety net hospital, patients who survived their initial sepsis hospitalization had a 30-day readmission rate to our institution of 22.6% that is comparable to rates described in prior studies. Readmission was commonly due to infection. Factors associated with readmission included multiple comorbid medical conditions, bacteremia, and being discharged with a vascular catheter. Further studies in this population are needed to determine potential modifiability of these risk factors in an attempt to reduce sepsis readmissions.

Comparison of Cytopathologist-Performed Ultrasound-Guided Fine-Needle Aspiration With Cytopathologist-Performed Palpation-Guided Fine-Needle Aspiration: A Single Institutional Experience.

CONTEXT.­: Although fine-needle aspiration (FNA) practice by pathologists is now well established, it has been primarily performed by manual palpation. In recent years, pathologists have begun to venture into ultrasound-guided FNAs (UGFNAs). Reports on experiences with this relatively new technique for pathologists have shown promising results. However to date, there have been few studies in the literature comparing pathologist-performed UGFNA with the more traditional pathologist-performed palpation-guided FNA (PGFNA). OBJECTIVE.­: To compare UGFNA to PGFNA by cytopathologists at an academic medical center. DESIGN.­: A retrospective study of FNAs performed by cytopathologists within the University of California, Los Angeles (UCLA) pathology departmental FNA clinic was performed. Data collected included performance technique (UGFNA versus PGFNA), lesion site and size, adequacy status (nondiagnostic rate), and number of passes per procedure. Corresponding surgical pathology/flow cytometric/cytogenetic result follow-up was compared to FNA results. Findings between UGFNA and PGFNA cases were compared. RESULTS.­: Of 1029 FNA cases during the study period, there were 449 UGFNA cases (43.6%) and 580 PGFNA cases (56.4%). Nondiagnostic rates with UGFNA and PGFNA were 6.7% (30 of 449 cases) and 20.7% (120 of 580 cases), respectively. Nondiagnostic rate was also significantly lower with UGFNA than with PGFNA for lesions within the thyroid (6.0% versus 33.3%), head and neck (6.6% versus 21.2%), and salivary gland (6.2% versus 17.1%), and across all nodule sizes. A total of 495 of 1029 FNA cases (48.1%) had follow-up. Discordance rate was significantly lower with UGFNA than with PGFNA (5.4% versus 12.8%). CONCLUSIONS.­: This study shows improved performance characteristics of cytopathologist-performed UGFNA versus PGFNA.

Blood donation mobile applications: are donors ready?

BACKGROUND: The rapid rise of mobile communication technologies has the potential to dramatically change and improve blood donor recruitment and retention efforts. STUDY DESIGN AND METHODS: E-mail invitations were sent to blood donors in a large metropolitan area to participate in a Web-based survey designed to gauge their readiness and interest level for a blood donation mobile application ("app"). RESULTS: A total of 982 ethnically diverse respondents of various age groups and prior donation experiences were surveyed. Among the respondents, 87.3% had ready access to smart phones. E-mail was chosen by 62.1% as the currently preferred method when contacted by the blood center, followed by texting (10.1%). App features desired by most respondents were the abilities to request appointments 24/7 (76.8%) and to receive appointment confirmations quickly (81.3%). Many were concerned about receiving too many alerts or messages (64.1%) or insufficient protection for personal information (53.5%). Overall, 67.7% of respondents indicated that they were likely to use a blood donation mobile app. Likelihood was not significantly different by sex or ethnicity, and the impact of education level was limited. Donors who currently made donation appointments via telephone or a website were equally likely to use such an app. However, donors older than 45 years were less likely than younger donors (p = 0.001), and donors with more than five lifetime donations were more likely than less frequent donors to use such an app (p = 0.02). CONCLUSION: In a metropolitan area, donors are very receptive to using a mobile app to manage their donations.

Perception vs Reality: Methicillin-Resistant Staphylococcus aureus Carriage Among Healthcare Workers at a Veterans Affairs Medical Center.

In a prevalence study of 209 healthcare workers, 18 (8.6%) and 13 (6.2%) carried methicillin-resistant Staphylococcus aureus in their nares or on their hands, respectively. However, 100 (62%) of 162 workers completing an associated survey believed themselves to be colonized, revealing a knowledge deficit about methicillin-resistant Staphylococcus aureus epidemiology. Infect. Control Hosp. Epidemiol. 2015;37(1):110-112.

Coccidioides thyroiditis in an HIV-infected patient.

We report a case of Coccidioides thyroiditis in an HIV-infected patient with a history of recent Coccidioides pneumonia but with negative Coccidioides serology determined by enzyme immunoassay at presentation. Diagnosis of Coccidioides thyroiditis was made based on histopathologic examination and culture of thyroid abscess material obtained by fine-needle aspiration biopsy.

A negative nares screen in combination with absence of clinical risk factors can be used to identify patients with very low likelihood of methicillin-resistant Staphylococcus aureus infection in a Veterans Affairs hospital.

BACKGROUND: Screening for nares carriage of methicillin-resistant Staphylococcus aureus (MRSA) could be used to identify patients at reduced risk for MRSA infection. However, recent studies suggest that negative results of nares surveillance testing are not reliable to predict a low risk for MRSA infection. METHODS: To derive a clinical prediction rule, logistic regression was performed to identify predictors of MRSA infection in patients with negative nares surveillance results. To validate the prediction rule and modifications of the rule, the sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated in a prospective cohort of inpatients receiving empiric vancomycin. RESULTS: A prediction rule that included 8 clinical risk factors and nares surveillance results was sensitive with NPV of 99.9%. In the validation cohort, 43 of 451 (10%) patients receiving empiric vancomycin had MRSA infection. In comparison with nares surveillance results (sensitivity, 72%; NPV, 96%), the prediction rule was more reliable for prediction of low risk for MRSA infection (sensitivity, 93%; NPV, 98%). A modified rule that included additional indications for empiric MRSA coverage based on practice guidelines reliably predicted MRSA infection (sensitivity, 100%) while excluding those at low risk (NPV, 100%). CONCLUSION: Negative nares screening results in combination with absence of certain clinical risk factors or syndromes could provide a useful strategy to predict patients with low likelihood of MRSA infection.

Acquisition of spores on gloved hands after contact with the skin of patients with Clostridium difficile infection and with environmental surfaces in their rooms.

In a prospective study of 30 patients with Clostridium difficile infection, we found that acquisition of spores on gloved hands was as likely after contact with commonly touched environmental surfaces (ie, bed rail, bedside table, telephone, call button) as after contact with commonly examined skin sites (ie, chest, abdomen, arm, hand).

Epidemiology of clostridium difficile and vancomycin-resistant Enterococcus colonization in patients on a spinal cord injury unit.

BACKGROUND/OBJECTIVE: Patients with spinal cord injury (SCI) are at risk of acquiring colonization with Clostridium difficile and vancomycin-resistant Enterococcus (VRE) due to prolonged hospitalization and frequent antimicrobial use. We examined the frequency of stool, skin, and environmental contamination with C. difficile and VRE in hospitalized patients with SCl. METHODS: We performed a cross-sectional study of 22 hospitalized patients with SCI with no symptoms of C. difficile infection. Stool samples, skin, and environmental sites were cultured for C. difficile and VRE, and polymerase chain reaction ribotyping was performed for C. difficile isolates. Fisher's exact test was used to compare the proportions of skin and environmental contamination among stool carriers and non-carriers. Univariate analysis was used to assess factors associated with asymptomatic carriage of C. difficile. RESULTS: Of 22 asymptomatic patients, 11 (50%) were asymptomatic carriers of toxigenic C. difficile and 12 (55%) were carriers of VRE. In comparison with non-carriers, asymptomatic carriers of toxigenic C. difficile had higher rates of skin (45 versus 9%) (P = 0.07) and environmental contamination (55 versus 9%) (P = 0.03) and longer length of stay (median, 57 versus 6 days; P = 0.04). A majority of skin and environmental C. difficile isolates from individuals were identical to isolates from stool. In comparison with non-carriers, patients with VRE stool colonization had non-significant trends toward more frequent skin (27 versus 9%) and environmental (18 versus 9%) contamination. CONCLUSION: Asymptomatic stool carriage of toxigenic C. difficile and VRE was common on an acute-care SCI unit. Asymptomatic carriers of toxigenic C. difficile had frequent skin and environmental contamination, suggesting the potential to contribute to transmission.

Clostridium difficile infection in a Department of Veterans Affairs long-term care facility.

In a Veterans Affairs medical center, 39% of healthcare facility-onset, healthcare facility-associated Clostridium difficile infections had their onset in the affiliated long-term care facility (LTCF). Eighty-five percent of LTCF-onset patients had been transferred from the hospital within the past month. Delays in diagnosis and treatment were common for LTCF-onset patients.

Occurrence of skin and environmental contamination with methicillin-resistant Staphylococcus aureus before results of polymerase chain reaction at hospital admission become available.

BACKGROUND: Active surveillance to detect patients colonized with methicillin-resistant Staphylococcus aureus (MRSA) is increasingly practiced in healthcare settings. However, inpatients may already become sources of transmission before appropriate precautions are implemented. OBJECTIVE: To examine the frequency of MRSA contamination of commonly touched skin and environmental surfaces before patient carriage status became known. METHODS: We conducted a 6-week prospective study of patients who were identified by use of polymerase chain reaction (PCR) at hospital admission as having nasal MRSA colonization. Skin and environmental contamination was assessed within hours of completion of PCR screening. RESULTS: There were 116 patients identified by PCR screening as having nasal MRSA colonization during the period from mid-April to May 2008, of whom 83 (72%) were enrolled in our study. Overall, MRSA was detected on the skin of 38 (51%) of 74 patients and in the environment of 37 (45%) of 83 patients. Of 83 environmental culture samples, 63 (76%) were obtained within 7 hours after PCR results became available, and 73 (88%) were obtained before wards were notified of PCR results. Of the 83 MRSA-colonized patients, 15 (18%) had contaminated their environment 25 hours after admission, and 29 (35%) had contaminated their environment 33 hours after admission. Thirty-two (39%) of the 83 patients had roommates, 13 (41%) of whom contaminated their environment. The median interval from admission to PCR result was 20 hours, and the median interval from PCR result to notification was 23 hours. An increased quantity of MRSA cultured from a nasal sample was significantly associated with contamination. CONCLUSIONS: Before any contact precautions can be implemented, newly identified MRSA carriers frequently have contaminated their environment with MRSA and have contamination of commonly examined skin sites. In hospitals that perform active surveillance, strategies are needed to minimize delays in screening or to preemptively identify patients at high risk for disseminating MRSA.

Evaluation of an alcohol-based power sanitizing system for decontamination of hospital rooms of patients with methicillin-resistant Staphylococcus aureus carriage.

A power sanitizing system that delivers a sanitizing mist of alcohol, rendered nonflammable by a carbon dioxide carrier, and a low concentration of quaternary ammonium compound, was easy to use, provided thorough coverage of surfaces, and significantly reduced vancomycin-resistant Enterococcus and methicillin-resistant Staphylococcus aureus contamination on commonly touched hospital surfaces. However, it was not as consistently effective for elimination of pathogens as 10% bleach solution.

Skin and environmental contamination with methicillin-resistant Staphylococcus aureus among carriers identified clinically versus through active surveillance.

In a prospective study involving 115 patients with methicillin-resistant Staphylococcus aureus nares carriage, we found that skin and environmental contamination with methicillin-resistant S. aureus was as likely among individuals whose S. aureus carriage was identified only through active surveillance as it was among those individuals whose S. aureus carriage was identified clinically, which suggests that strategies to limit transmission must address colonized patients, as well as infected patients.