Use of a large language model with instruction-tuning for reliable clinical frailty scoring.

BACKGROUND: Frailty is an important predictor of health outcomes, characterized by increased vulnerability due to physiological decline. The Clinical Frailty Scale (CFS) is commonly used for frailty assessment but may be influenced by rater bias. Use of artificial intelligence (AI), particularly Large Language Models (LLMs) offers a promising method for efficient and reliable frailty scoring. METHODS: The study utilized seven standardized patient scenarios to evaluate the consistency and reliability of CFS scoring by OpenAI's GPT-3.5-turbo model. Two methods were tested: a basic prompt and an instruction-tuned prompt incorporating CFS definition, a directive for accurate responses, and temperature control. The outputs were compared using the Mann-Whitney U test and Fleiss' Kappa for inter-rater reliability. The outputs were compared with historic human scores of the same scenarios. RESULTS: The LLM's median scores were similar to human raters, with differences of no more than one point. Significant differences in score distributions were observed between the basic and instruction-tuned prompts in five out of seven scenarios. The instruction-tuned prompt showed high inter-rater reliability (Fleiss' Kappa of 0.887) and produced consistent responses in all scenarios. Difficulty in scoring was noted in scenarios with less explicit information on activities of daily living (ADLs). CONCLUSIONS: This study demonstrates the potential of LLMs in consistently scoring clinical frailty with high reliability. It demonstrates that prompt engineering via instruction-tuning can be a simple but effective approach for optimizing LLMs in healthcare applications. The LLM may overestimate frailty scores when less information about ADLs is provided, possibly as it is less subject to implicit assumptions and extrapolation than humans. Future research could explore the integration of LLMs in clinical research and frailty-related outcome prediction.

Comparison of the Quality of Discharge Letters Written by Large Language Models and Junior Clinicians: Single-Blinded Study.

BACKGROUND: Discharge letters are a critical component in the continuity of care between specialists and primary care providers. However, these letters are time-consuming to write, underprioritized in comparison to direct clinical care, and are often tasked to junior doctors. Prior studies assessing the quality of discharge summaries written for inpatient hospital admissions show inadequacies in many domains. Large language models such as GPT have the ability to summarize large volumes of unstructured free text such as electronic medical records and have the potential to automate such tasks, providing time savings and consistency in quality. OBJECTIVE: The aim of this study was to assess the performance of GPT-4 in generating discharge letters written from urology specialist outpatient clinics to primary care providers and to compare their quality against letters written by junior clinicians. METHODS: Fictional electronic records were written by physicians simulating 5 common urology outpatient cases with long-term follow-up. Records comprised simulated consultation notes, referral letters and replies, and relevant discharge summaries from inpatient admissions. GPT-4 was tasked to write discharge letters for these cases with a specified target audience of primary care providers who would be continuing the patient's care. Prompts were written for safety, content, and style. Concurrently, junior clinicians were provided with the same case records and instructional prompts. GPT-4 output was assessed for instances of hallucination. A blinded panel of primary care physicians then evaluated the letters using a standardized questionnaire tool. RESULTS: GPT-4 outperformed human counterparts in information provision (mean 4.32, SD 0.95 vs 3.70, SD 1.27; P=.03) and had no instances of hallucination. There were no statistically significant differences in the mean clarity (4.16, SD 0.95 vs 3.68, SD 1.24; P=.12), collegiality (4.36, SD 1.00 vs 3.84, SD 1.22; P=.05), conciseness (3.60, SD 1.12 vs 3.64, SD 1.27; P=.71), follow-up recommendations (4.16, SD 1.03 vs 3.72, SD 1.13; P=.08), and overall satisfaction (3.96, SD 1.14 vs 3.62, SD 1.34; P=.36) between the letters generated by GPT-4 and humans, respectively. CONCLUSIONS: Discharge letters written by GPT-4 had equivalent quality to those written by junior clinicians, without any hallucinations. This study provides a proof of concept that large language models can be useful and safe tools in clinical documentation.

ChatGPT on guidelines: Providing contextual knowledge to GPT allows it to provide advice on appropriate colonoscopy intervals.

BACKGROUND AND AIM: Colonoscopy is commonly used in screening and surveillance for colorectal cancer. Multiple different guidelines provide recommendations on the interval between colonoscopies. This can be challenging for non-specialist healthcare providers to navigate. Large language models like ChatGPT are a potential tool for parsing patient histories and providing advice. However, the standard GPT model is not designed for medical use and can hallucinate. One way to overcome these challenges is to provide contextual information with medical guidelines to help the model respond accurately to queries. Our study compares the standard GPT4 against a contextualized model provided with relevant screening guidelines. We evaluated whether the models could provide correct advice for screening and surveillance intervals for colonoscopy. METHODS: Relevant guidelines pertaining to colorectal cancer screening and surveillance were formulated into a knowledge base for GPT. We tested 62 example case scenarios (three times each) on standard GPT4 and on a contextualized model with the knowledge base. RESULTS: The contextualized GPT4 model outperformed the standard GPT4 in all domains. No high-risk features were missed, and only two cases had hallucination of additional high-risk features. A correct interval to colonoscopy was provided in the majority of cases. Guidelines were appropriately cited in almost all cases. CONCLUSIONS: A contextualized GPT4 model could identify high-risk features and quote appropriate guidelines without significant hallucination. It gave a correct interval to the next colonoscopy in the majority of cases. This provides proof of concept that ChatGPT with appropriate refinement can serve as an accurate physician assistant.

Clinical outcomes in end stage renal disease on dialysis and severe coronary artery disease: A real-world study.

BACKGROUND: The optimal management of patients with end-stage renal disease (ESRD) on dialysis with severe coronary artery disease (CAD) has not been determined. METHODS: Between 2013 and 2017, all patients with ESRD on dialysis who had left main (LM) disease, triple vessel disease (TVD) and/or severe CAD for consideration of coronary artery bypass graft (CABG) were included. Patients were divided into 3 groups based on final treatment modality: CABG, percutaneous coronary intervention (PCI), optimal medical therapy (OMT). Outcome measures include in-hospital, 180-day, 1-year and overall mortality and major adverse cardiac events (MACE). RESULTS: In total, 418 patients were included (CABG 11.0%, PCI 65.6%, OMT 23.4%). Overall, 1-year mortality and MACE rates were 27.5% and 55.0% respectively. Patients who underwent CABG were significantly younger, more likely to have LM disease and have no prior heart failure. In this non-randomized setting, treatment modality did not impact on 1-year mortality, although the CABG group had significantly lower 1-year MACE rates (CABG 32.6%, PCI 57.3%, OMT 59.2%; CABG vs. OMT p < 0.01, CABG vs. PCI p < 0.001). Independent predictors of overall mortality include STEMI presentation (HR 2.31, 95% CI 1.38-3.86), prior heart failure (HR 1.84, 95% CI 1.22-2.75), LM disease (HR 1.71, 95% CI 1.26-2.31), NSTE-ACS presentation (HR 1.40, 95% CI 1.03-1.91) and increased age (HR 1.02, 95% CI 1.01-1.04). CONCLUSION: Treatment decisions for patients with severe CAD with ESRD on dialysis are complex. Understanding independent predictors of mortality and MACE in specific treatment subgroups may provide valuable insights into the selection of optimal treatment options.

Retrospective analysis of endoscopic retrograde cholangiopancreatography after pancreaticoduodenectomy / 中华消化内镜杂志

Objective:To evaluate the efficacy and safety of endoscopic retrograde cholangiopancreatography (ERCP) after pancreaticoduodenectomy and endoscopic selection strategies.Methods:Clinical data of 34 patients treated with ERCP after pancreaticoduodenectomy at the Endoscopic Center of the First Affiliated Hospital of Air Force Medical University from January 2013 to December 2021 were retrospectively analyzed. The success rates of endoscopic insertion, diagnosis, treatment and ERCP, and the incidence of adverse events were analyzed.Results:Fifty ERCP treatments were performed in 34 patients. The success rates of endoscopic insertion, diagnosis, treatment, and ERCP after pancreaticoduodenectomy were 92.0% (46/50), 93.5% (43/46), 88.4% (38/43) and 76.0% (38/50), respectively. The success rates of ERCP assisted with colonoscope and balloon-assisted enterosocpe were 76.0% (19/25) and 75.0% (18/24), respectively. There were 3 adverse events, including 1 case of anastomotic mucosa tear during surgery, 1 case of cardiopulmonary arrest and 1 case of postoperative cholangitis.Conclusion:ERCP is effective and safe after pancreaticoduodenectomy in general. ERCP assisted with colonoscope and balloon-assisted colonoscope shows similar success rate after pancreaticoduodenectomy.

Comparison of existing methods of low-density lipoprotein cholesterol estimation in patients with type 2 diabetes mellitus.

Introduction: Elevated low-density lipoprotein cholesterol (LDL-C) is an important risk factor for atherosclerotic cardiovascular disease (ASCVD). Direct LDL-C measurement is not widely performed. LDL-C is routinely calculated using the Friedewald equation (FLDL), which is inaccurate at high triglyceride (TG) or low LDL-C levels. We aimed to compare this routine method with other estimation methods in patients with type 2 diabetes mellitus (T2DM), who typically have elevated TG levels and ASCVD risk. Method: We performed a retrospective cohort study on T2DM patients from a multi-institutional diabetes registry in Singapore from 2013 to 2020. LDL-C values estimated by the equations: FLDL, Martin/Hopkins (MLDL) and Sampson (SLDL) were compared using measures of agreement and correlation. Subgroup analysis comparing estimated LDL-C with directly measured LDL-C (DLDL) was conducted in patients from a single institution. Estimated LDL-C was considered discordant if LDL-C was <1.8mmol/L for the index equation and ≥1.8mmol/L for the comparator. Results: A total of 154,877 patients were included in the final analysis, and 11,475 patients in the subgroup analysis. All 3 equations demonstrated strong overall correlation and goodness-of-fit. Discordance was 4.21% for FLDL-SLDL and 6.55% for FLDL-MLDL. In the subgroup analysis, discordance was 21.57% for DLDL-FLDL, 17.31% for DLDL-SLDL and 14.44% for DLDL-MLDL. All discordance rates increased at TG levels >4.5mmol/L. Conclusion: We demonstrated strong correlations between newer methods of LDL-C estimation, FLDL, and DLDL. At higher TG concentrations, no equation performed well. The Martin/Hopkins equation had the least discordance with DLDL, and may minimise misclassification compared with the FLDL and SLDL.

Risk factors for moderate to severe pancreatitis after endoscopic retrograde cholangiopancreatography / 中华消化内镜杂志

Objective:To investigate the risk factors for moderate to severe pancreatitis (PEP) after endoscopic retrograde cholangiopancreatography (ERCP) .Methods:Data of 6 731 patients diagnosed as having biliary and pancreatic diseases with initial papilla who underwent endoscopic retrograde cholangiopancreatography (ERCP) from June 2010 to June 2020 in the First Affiliated Hospital of Air Force Medical University were retrospectively analyzed. Parameters related to intubation and postoperative complications were prospectively collected. The main end point was moderate to severe PEP. Logistic regression analysis was used to explore the risk factors for moderate to severe PEP.Results:The incidence of overall PEP and moderate to severe PEP in 6 731 ERCP patients with initial papilla were 5.3% ( n=359) and 1.0% ( n=68) respectively. Univariate analysis showed that female, indications of ERCP, cannulation method, cannulation time, cannulation attempts, times of inadvertent pancreatic duct cannulation and cannulation with or without trainee involvement were all associated with moderate to severe PEP ( P<0.10). Multivariate analysis showed that female ( OR=2.32, 95% CI:1.28-4.21, P=0.006), non-common bile duct stones indication ( OR=2.04, 95% CI:1.16-3.59, P=0.014), cannulation time ≥5 min ( OR=2.23, 95% CI: 1.20-4.13, P=0.011), inadvertent pancreatic duct cannulation time ≥1 ( OR=1.88, 95% CI: 1.03-3.44, P=0.040) and non-trainee involvement cannulation ( OR=1.81, 95% CI: 1.02-3.22, P=0.043) were independent risk factors for moderate to severe PEP. Conclusion:The independent risk factors for moderate to severe PEP include female, non-common bile duct stones indication, non-trainee involvement cannulation and difficult cannulation. Great importance should be attached to these factors above during the whole perioperative period of ERCP.

Mercury wet depositions study at suburban, agriculture and traffic sampling sites.

This study aimed to measure and discuss the relationship of ambient air precipitations with respect to mercury wet depositions at suburban, agriculture and traffic three characteristic sampling sites during the year of 2019. In addition, the mercury volume weighted mean concentrations (VWM) at three characteristic sampling sites were also calculated. Finally, the ambient mercury wet depositions data obtained in this study to various world sampling sites were also compared and discussed in this study. The results indicated that the average mercury wet depositions for suburban, agriculture and traffic areas were 0.62, 0.55 and 2.32 ng/m2 min, respectively. And the average mercury VWM values were 0.9, 0.72 and 1.85 ng/m2 min for suburban, agriculture and traffic sites, respectively. In addition, the highest VWM and wet depositions for mercury both occurred in March at traffic and suburban areas. And the mercury wet depositions displayed a declined trend when the month was moved from March to July at both traffic and suburban sampling sites. In addition, the relationship between wet depositions and precipitations was low to moderate correlated in traffic area, while the relationship between wet depositions and precipitations was insignificant at both suburban and agriculture areas. Moreover, the average highest mercury wet deposition occurred in Nepal when compared to the other world sites. In addition, the average value of mercury wet depositions in Nepal was about 17.23 times to that of data obtained in this study during the period of 2007-2019. Finally, the average highest VWM (ng/L) occurred in the China. In addition, the average value mercury VWM in China was about 14.82 times to that of data obtained in this study during the period of 2007-2019.

Particulate and particulate-bound mercury concentrations and size distributions as related to seasonal variations during peak demand/non-peak demand periods.

The particulate size distributions of aerosol pollutants (particulates and Hg(p)) at a mixed site were measured and their seasonal variations identified. Atmospheric particulates and the Hg(p) mass median diameter (m.m.d.) were obtained. Hg(p) concentrations increased by approximately 20% during the peak demand period for all particle sizes (18, 10, 2.5, 1 and 0.3 µm). The mean percentage concentration of Hg(p) was highest in summer and followed the order summer > spring > winter > autumn for all particle sizes. Hg(p) concentration exhibited increased from 2004 to 2019.

Particulates and particulates-bound mercury (Hg(p)) sizes (PM18, PM10, PM2.5, PM1, PM<1) distributions study by using MOUDI sampler at a complex sampling site.

The objectives of this study were to measure ambient air particles concentrations of different particulates sizes ranges (PM18, PM10, PM2.5, PM1, PM<1) at a complex (traffic, residential and commercial) site. Besides, particulates-bound mercury (Hg(p)) concentrations for various particulates sizes (PM18, PM10, PM2.5, PM1, PM<1) at mixed site were also studied. Finally, ambient air particulates and Hg(p) size distributions were also described at this complex sampling site. The results showed that the average PM18, PM10, PM2.5, PM1, PM<1 concentrations were 48.83, 41.78, 35.41, 19.89, and 11.86 µg/m3, respectively. And the average ambient air particulates-bound mercury (Hg(p)) which attached on PM18, PM10, PM2.5, PM1, PM<1 particles concentrations were 0.0838, 0.0867, 0.0790, 0.0546, and 0.0373 ng/m3, respectively, in the summer season. In addition, the average ambient air Hg(p) which attached on PM18, PM10, PM2.5, PM1, PM<1 particles concentrations were 0.0175, 0.0144, 0.0120, 0.0092, and 0.0057 ng/m3, respectively, in the autumn season. Finally, the average ambient air Hg(p) which attached on PM18, PM10, PM2.5, PM1, PM<1 particles concentrations were 0.0070, 0.0053, 0.0038, 0.0026, and 0.0014 ng/m3, respectively, in the winter season. And July has the average highest PM18 and PM10 concentrations. As for PM2.5, PM1 and PM<1 particulates, the average highest particulates concentrations all occurred in November. In addition, the highest average Hg(p) in PM18, PM10, PM2.5, PM1, and PM<1 concentrations all occurred in July. Moreover, the average particles and particulates-bound mercury m.m.d. values were ranged from 1.0 to 1.8 and 0.7 to 2.0 µm from July to December of 2018, respectively, at this mixed sampling site. As for monthly ambient air particles sizes distributions, the results further showed that the main peaks for July, September, and December all occurred in the sizes of 10-18 µm. The main peaks for October and November all occurred in the sizes of 2.5-10 µm. As for monthly Hg(p) sizes distributions, the results further showed that the main peaks for July occurred in the size of 0.3-1 µm. The main peak for September occurred in the size of 1-2.5 µm. The main peaks for October to December all occurred in the size of 10-18 µm. The above finding further concluded that the particulates-bound mercury (Hg(p)) was tended to be associated with the large particles sizes mode during the winter season. Finally, this study further shows that the Taichung Thermal Power Plant was responsible for the main emission source of Hg(p) especially in summer season of Central Taiwan.

Sources and monthly and seasonal concentration variation study of atmospheric particulates and particles-bound PAEs.

The main goal of this study is to measure the ambient air of total suspended particulates at a mixed (traffic, residential and commercial) sampling site. And the concentrations of phthalate acid ester (PAE) which attached on total suspended particles were also analyzed. In addition, the possible sources of PAEs were also analyzed by the method of back trajectories. Finally, appropriate statistical methods were also used to test monthly and seasonal mean pollutants' (particulates, PAEs) concentration differences at this sampling site. The results indicated that the monthly concentration variation trends were similar for DEHP and total PAEs with the results as followed: November > October > July > December > September. In addition, back trajectories results also indicated that the main pollutant parcels were came from the east side of Taiwan in July. And the pollutant parcels were came from the north side of Taiwan during the month of September, October, November and December in this study. Moreover, the results also showed that the DEHP, DNOP, total PAEs' concentrations with TSP and meteorological conditions were not significantly different. But the relationship among DEHP, DNOP and total PAEs was significantly different; particularly, the relationship between DEHP and total PAEs was highly correlated in this study (R2 = 0.994). Finally, the statistical analysis of total PAEs T test statistic for mean monthly concentrations results suggested that the sample population means were not differed significantly. In other words, there were not any mean monthly concentration differences for PAEs at this sampling site. The only exception was occurred in the month of September versus November. The results showed that there is a statistically significant PAEs' concentration difference between the input groups (September vs. November).

Effect of comprehensive nursing care on first aid and mood of patients with drug poisoning / 中国生化药物杂志

Objective To study the effect of comprehensive nursing care on patients with drug poisoning. Methods 102 cases of drugs in our hospital from September 2014 to September 2016 the poisoning patients were analyzed, using randomly divided into study group and control group, 51 cases in each group, the control group of nursing mode of patients on the basis of comprehensive nursing intervention group were treated by using SAS model, and SDS measurement of adverse psychological state of patients in the two groups, the success probability, death rate, SAS score, SDS score and total satisfaction were compared between the two groups. Results The rescue failure rate of the study group (7.8%) was significantly lower than that of the reference group (23.6%), and there was a difference between the two groups. The SAS score and SDS score in the study group than the control group (78.4%) , with statistical significance (P<0.05), the satisfaction of study group (98%) was significantly higher than that of control group, with statistical significance (P<0.05). Conclusion Drug poisoning patients using evidence-based nursing can improve the emergency into Power, ease the patient's bad mood, patient satisfaction is relatively high

Effect of comprehensive nursing care on first aid and mood of patients with drug poisoning / 中国生化药物杂志

Objective To study the effect of comprehensive nursing care on patients with drug poisoning. Methods 102 cases of drugs in our hospital from September 2014 to September 2016 the poisoning patients were analyzed, using randomly divided into study group and control group, 51 cases in each group, the control group of nursing mode of patients on the basis of comprehensive nursing intervention group were treated by using SAS model, and SDS measurement of adverse psychological state of patients in the two groups, the success probability, death rate, SAS score, SDS score and total satisfaction were compared between the two groups. Results The rescue failure rate of the study group (7.8%) was significantly lower than that of the reference group (23.6%), and there was a difference between the two groups. The SAS score and SDS score in the study group than the control group (78.4%) , with statistical significance (P<0.05), the satisfaction of study group (98%) was significantly higher than that of control group, with statistical significance (P<0.05). Conclusion Drug poisoning patients using evidence-based nursing can improve the emergency into Power, ease the patient's bad mood, patient satisfaction is relatively high

Accuracy of endoscopic ultrasonography for evaluating minimal submucosal invasion of early gastro-intestinal tumor / 中华消化内镜杂志

Objective To explore the diagnostic value of endoscopic ultrasonography(EUS)for the minimal submucosal invasion of early gastrointestinal tumor. Methods A total of 242 patients with early gastrointestinal tumor,who underwent endoscopic submucosal dissection,were retrospectively analyzed. The accuracy of EUS diagnosis was calculated based on postoperative histopathological findings as the golden standard,and influencing factors were also analyzed. Results Overall diagnostic accuracy of EUS for sub-mucosal invasion of early gastrointestinal tumors was 72. 3%(175/ 242),with an overstaging rate of 21. 5%(52/ 242)and an understaging rate of 6. 2%(15/ 242).Tumor size(P = 0. 018)and location(P = 0. 005) had significant effects on the diagnostic accuracy of the minimal submucosal invasion of early gastrointestinal tumor by EUS. The overstaging rate in the lesion length of diameter>3 cm was higher than those of 3 cm or less[27. 0%(33/ 122)VS 15. 8%(19/ 120),P = 0. 807],the overstaging rates of early colonrectal and gastric cancer were also significantly higher than the understaging rate[ Colonrectum:12. 2%(9/ 74)VS 2. 7%(2/ 74),P= 0. 028;Stomach:26. 9%(28/ 104)VS 2. 9%(3/ 104),P = 0. 000]. Conclusion Endoscopic ultrasonography is of diagnostic value for the invasion depth of early cancer in gastrointestinal tract. However,precaution should be taken in large lesions and the tendency of overstaging in gastrointestinal tract.

Complex and novel determinants of empathy change in medical students / 한국의학교육

PURPOSE: Physician empathy is a core attribute in medical professionals, giving better patient outcomes. Medical school is an opportune time for building empathetic foundations. This study explores empathy change and focuses on contributory factors. METHODS: We conducted a cross-sectional study involving 881 students (63%) from Years 1 to 5 in a Singaporean medical school using the Jefferson Scale of Physician Empathy-Student version (JSPE-S) and a questionnaire investigating the relationship between reported and novel personal-social empathy determinants. RESULTS: Empathy declined significantly between preclinical and clinical years. Female and medical specialty interest respondents had higher scores than their counterparts. Despite strong internal consistency, factor analysis suggested that the JSPE model is not a perfect fit. Year 1 students had highest Perspective Taking scores and Year 2 students had highest Compassionate Care scores. High workload and inappropriate learning environments were the most relevant stressors. Time spent with family, arts, and community service correlated with higher empathy scores, whilst time spent with significant others and individual leisure correlated with lower scores. Thematic analysis revealed that the most common self-reported determinants were exposure to activity (community service) or socialisation, personal and family-related event as well as environment (high work-load). CONCLUSION: While the empathy construct in multicultural Singapore is congruent with a Western model, important differences remain. A more subtle understanding of the heterogeneity of the medical student experience is important. A greater breadth of determinants of empathy, such as engagement in arts-related activities should be considered.

Tigecycline-induced Drug Fever and Leukemoid Reaction: A Case Report.

In this study, we describe a patient in whom tigecycline-induced drug fever and leukemoid reaction (LR) after 3 weeks of therapy for pneumonia.A 62-year-old man developed aspiration pneumonia on February 1, 2015. He had received multiple antibiotics at another hospital, but did not respond well. Disease rapidly progressed, and he was referred to our department on February 14. We adjusted the antibiotic therapy to tigecycline + vancomycin, and added voriconazole to empiric antifungal therapy. Pneumonia largely improved, and we discontinued vancomycin and voriconazole on February 28. With tigecycline monotherapy, his clinical status remained stable.On March 7, he developed high fever and LR (white blood cell count: 38.25 × 10(9)/L). Erythrocyte sedimentation rate and C-reactive protein were elevated, and CD8+ T cells had been abnormally activated. After a careful physical examination and laboratory investigation, we confirmed that primary infection did not progress and no other cause was evident. So we figured fever and LR might be induced by tigecycline. After discontinuing tigecycline and adding low-dose steroid, fever and LR totally resolved in 3 days, which further confirmed our diagnosis.According to this case and literature review, drug-induced hypersensitivity should be considered in the differential diagnosis of fever and LR when the therapeutic duration of tetracycline approximates 3 weeks. Monitoring T-cell subsets may facilitate early diagnosis. When necessary, we should discontinue the suspected drug to confirm diagnosis.