medRxiv最新文献

{"title":"Plasma and CSF neurofilament light chain distinguish neurodegenerative from primary psychiatric conditions in a clinical setting","authors":"D. Eratne, M. Kang, C. Lewis, C. Dang, C. Malpas, M. Keem, J. Grewal, Vladimir Marinov, A. Coe, Cath Kaylor-Hughes, Thomas Borchard, Chhoa Keng-Hong, Alexandra Waxmann, Burcu Saglam, Tomáš Kalinčík, Richard Kanaan, W. Kelso, Andrew Evans, S. Farrand, S. Loi, M. Walterfang, C. Stehmann, Qiao-Xin Li, Steven Collins, C. L. Masters, A. Santillo, Henrik Zetterberg, K. Blennow, S. Berkovic, Dennis Velakoulis, Terence J. O’Brien, Patrick Kwan, O. Hansson, Christopher Fowler, Jane Gunn","doi":"10.1101/2024.08.11.24311847","DOIUrl":"https://doi.org/10.1101/2024.08.11.24311847","url":null,"abstract":"INTRODUCTION: Many patients with neurodegenerative disorders (ND) face diagnostic delay and misdiagnosis. We investigated blood and cerebrospinal fluid (CSF) neurofilament light chain (NfL) to distinguish ND from primary psychiatric disorders (ND), a common challenge in clinical settings. METHODS: Plasma and CSF NfL levels were measured and compared between groups, adjusting for age, sex, weight. RESULTS: 337 participants included: 136 ND, 77 PPD, 124 Controls. Plasma NfL was 2.5 fold elevated in ND compared to PPD and had strong diagnostic performance (area under the curve, AUC 0.86, 81%/85% specificity/sensitivity) that was comparable to CSF NfL (2 fold elevated, AUC 0.89, 95%/71% specificity/sensitivity). Diagnostic performance was especially strong in younger people (40-<60years). Additional findings were cut-offs optimised for sensitivity and specificity, and issues important for future clinical translation CONCLUSIONS: This study adds important evidence for a simple blood-based biomarker to assist as a screening test for neurodegeneration and distinction from PPD, in clinical settings.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"11 41","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantifying the impact of a broadly protective sarbecovirus vaccine in a future SARS-X pandemic","authors":"Charles Whittaker, Gregory Barnsley, D. Mesa, D. Laydon, Chee Wah Tan, Feng Zhu, Rob Johnson, P. Doohan, G. Nedjati-Gilani, P. Winskill, Alexandra B. Hogan, A. Deol, Christinah Mukandavire, Katharina Hauck, David Chien, Boon Lye, Lin-Fa Wang, Oliver J Watson, Azra C Ghani","doi":"10.1101/2024.08.12.24311730","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311730","url":null,"abstract":"COVID-19 has underscored the need for more timely access to vaccines during future pandemics. This has motivated development of broad-spectrum vaccines providing protection against viral families, which could be stockpiled ahead of an outbreak and deployed rapidly following detection. We use mathematical modelling to evaluate the utility of a broadly protective sarbecovirus vaccine (BPSV) during a hypothetical SARS-X outbreak, including ring-vaccination, spatial targeting and mass vaccination of high-risk populations. Our results show BPSV ring- or spatially-targeted vaccination strategies are unlikely to contain a SARS-CoV-2-like virus but could contain or slow the spread of a SARS-CoV-1-like virus. Vaccination of high-risk populations with the BPSV ahead of a virus-specific vaccine (VSV) becoming available could substantially reduce mortality. For a 250-day VSV development timeline, BPSV availability reduced infection-related deaths in our model by 54% on average, though exact impact depended on the non-pharmaceutical intervention (NPI) scenario considered. We further show that BPSV availability enables shorter and less stringent NPIs to be imposed whilst limiting disease burden to that observed in the VSV-only scenario, though results are sensitive to vaccine properties (e.g. efficacy), health system capabilities (e.g. vaccination rollout speed) and the assumed timeline to VSV availability. Our modelling suggests that availability of a BPSV for those aged 60+ years could have averted 40-65% of COVID-19 deaths during the pandemic's first year, though exact impact depends on the size of the maintained stockpile. Our work highlights significant potential impact of a BPSV, but that achieving this impact depends on investment into health systems enabling rapid and equitable access during future SARS-X pandemics.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"11 30","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919661","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Lack of Epistatic Interaction of SNCA with APOE in Synucleinopathies","authors":"P. Saini, E. Yu, M. Estiar, L. Krohn, Kheireddin Mufti, Uladzislau Rudakou, J. Ruskey, F. Asayesh, S. Laurent, D. Spiegelman, J. Trempe, T. Quinnell, Nicholas Oscroft, Isabelle Arnulf, J. Montplaisir, J. Gagnon, A. Desautels, Y. Dauvilliers, Gian Luigi Gigli, M. Valente, Francesco Janes, A. Bernardini, K. Šonka, D. Kemlink, Wolfgang Oertel, Karri Kaivola, International Lbd Genomics Consortium, A. Janzen, G. Plazzi, E. Antelmi, F. Biscarini, M. Figorilli, M. Puligheddu, B. Mollenhauer, C. Trenkwalder, F. Sixel-Döring, V. C. Cock, C. Monaca, Donald G. Grosset, A. Heidbreder, Luigi Ferini-Strambi, F. Dijkstra, M. Viaene, B. Abril, B. Boeve, R. Postuma, Guy A. Rouleau, Victoria Anselmi, Abubaker Ibrahim, A. Stefani, Birgit Högl, Michele T.M. Hu, Sonja W Scholz, Z. Gan-Or, Montreal Neurological, Institute","doi":"10.1101/2024.08.12.24311821","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311821","url":null,"abstract":"Two recent studies suggested that the APOE {varepsilon}4 haplotype was associated with increased -synuclein pathology in cell and mouse models. Genetic variants in the SNCA region have strong association with Parkinson's disease (PD), Dementia with Lewy Bodies (DLB), and idiopathic REM Sleep Behavior Disorder (iRBD), while APOE is a genetic risk determinant for only DLB. To determine if genetic-level interactions between SNCA and APOE exists that can explain the protein-level association, we investigated the genotypic interaction of APOE and SNCA in cohorts of PD, DLB, and iRBD. We analyzed genome-wide association study (GWAS) data from 5,229 PD patients and 5,480 controls, 2,610 DLB patients and 1,920 controls, and 1,055 iRBD patients and 3,667 controls. We used logistic regression interaction models across all 3 cohorts independently between the 1) top GWAS signals of SNCA SNPs and APOE haplotypes, 2) SNP x SNP and 3-way SNP interaction across the entire coding region plus 200kb flanking each gene. No significant interactions were found to be associated with any of the synucleinopathies after correction for multiple testing. Our results do not support a role for genetic interactions between APOE and SNCA across PD, DLB, and iRBD. Since the tested genetic variants affect the expression and function of these proteins, it is likely that any interactions between them does not affect the risk of PD, DLB and iRBD.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"7 14","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The effects of food-based versus supplement-based very low-energy diets on gut microbiome composition and health outcomes in women with high body mass index (The MicroFit Study): a randomised controlled trial","authors":"Mohammadreza Mohebbi, Mojtaba Lotfaliany, Martin O’Hely, Jessica Batti, M. Kotowicz, Lucy Saunders, Richard Page, Sally Beatti, Wolfgang Marx, F. Jacka, Postdoctoral Mojtaba Lotfaliany, Research, Postdoctoral Martin O’Hely, Research Fellow c. Mark, Professor d. Michael Kotowicz, Berk, Research Sally Beattie, Affiliate, Dr Amelia J McGuinness","doi":"10.1101/2024.08.11.24311823","DOIUrl":"https://doi.org/10.1101/2024.08.11.24311823","url":null,"abstract":"Objective: To compare the effects of consuming food-based versus supplement-based very low-energy diet (VLED) programs on gut microbiome composition in women with a high body mass index (BMI). Design: An investigator-initiated, single-blind, two-arm, parallel-group randomised controlled-feeding trial with computer-generated 1:1 randomisation. From May 2021 to February 2022, women aged 30-65 years with BMI 30-45 kg/m2 were recruited from southwest Victoria, Australia, and randomised to a three-week food-based or supplement-based VLED program. The primary outcome was between-group differential change in faecal microbiome alpha diversity (Shannon index) from baseline to week three, assessed using shotgun metagenomics. Outcome assessors, study investigators, and analysing statisticians were blinded to group allocation until analysis completion. Allocation concealment was managed by an independent researcher using a computer software system. Modified intention-to-treat (mITT) analyses using linear mixed-effects regression models estimated mean between-group differential changes, reported as beta-coefficient point estimates ({beta}) and 95% confidence intervals (95%CI), adjusted for multiple comparisons. Results: Forty-seven participants were randomised (food-based: n=23, supplement-based: n=24). Of the 45 participants analysed, there was a between-group differential change in the Shannon index (mITT {beta}: 0.37, 95%CI: 0.15 to 0.60) from baseline to week three, with a greater increase in the food-based group (mean change: 0.26, 95%CI: 0.09 to 0.44; n=23) versus supplement-based group (mean change: -0.10, 95%CI: -0.25 to 0.05; n=22). There were 27 non-serious adverse events (food-based: 8, supplement-based: 19), all non-serious. Conclusion: A food-based VLED, with more whole food components and fewer highly processed industrial ingredients, increases gut microbiome diversity more than a supplement-based VLED.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"10 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919668","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Maternal obesogenic environment and its association with childhood obesity in Peru: A 9-year analysis of a national survey","authors":"J. Loayza-Castro, Víctor Juan, Vera-Ponce, Luisa Erika, Milagros Vásquez-Romero, Jhonatan Roberto, Astucuri Hidalgo, Nataly Mayely Sanchez-Tamay, F. E. Zuzunaga-Montoya, Juan Vera-Ponce, Milagros Vásquez Romero","doi":"10.1101/2024.08.11.24311822","DOIUrl":"https://doi.org/10.1101/2024.08.11.24311822","url":null,"abstract":"Introduction: Childhood obesity is a global public health concern with increasing prevalence in Peru. The obesogenic environment, including maternal and family environmental factors, plays a crucial role in the development of childhood obesity. Objective: To analyze the association between the maternal obesogenic environment and obesity in children under five years of age in Peru. Methods: An analytical cross-sectional study used data from the Demographic and Family Health Survey (DHS) from 2014 to 2022. To assess the obesogenic environment, variables such as maternal obesity, television use, smoking, and maternal anemia were analyzed. Childhood obesity was a body mass index Z-score > +3 standard deviations. A Poisson regression model was used to calculate crude and adjusted prevalence ratios. Results: The prevalence of childhood obesity was 1.99%. Obese mothers were found to be 1.52 times more likely to have obese children (aPR=1.52, 95% CI 1.40-1.65; p<0.001). No significant associations were found between frequent television use, maternal smoking, and anemia with childhood obesity after adjusting for multiple factors. Conclusion: This study highlights the importance of the maternal obesogenic environment, especially maternal obesity, in developing childhood obesity in Peru. Comprehensive interventions that address multiple aspects of the family obesogenic environment, including the prevention and management of maternal obesity, promotion of healthy lifestyles, and strengthening of public policies that foster healthy environments, are recommended. Key Words: Tobacco Use Disorder; Obesity, Maternal; Pediatric Obesity; Public Health; Peru (MeSH)","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"34 7","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919212","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TIDAL: Tool to Implement Developmental Analysis of Longitudinal data","authors":"A. Kwong, A. Edmondson-Stait, Eileen Xu, Ellen J. Thompson, Richard M. A. Parker, Ahmed Elhakeem, L. Romaniuk, Rebecca M. Pearson, Kate Tilling, Thalia C. Eley, McIntosh Andrew M, Heather C. Whalley","doi":"10.1101/2024.08.12.24311854","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311854","url":null,"abstract":"Motivation: Growth curve modelling is one method used to model trajectories of traits and behaviours over time. However, accessing, analysing and interpreting trajectories requires statistical expertise, thereby creating potential barriers for users to implement and understand longitudinal traits. TIDAL is a user-friendly research tool designed to facilitate trajectory modelling by improving access, analysis and interpretation of trajectory and longitudinal data. Implementation: TIDAL is available in two formats: an R package and an online Shiny application. The R package can be used offline, negating the need to upload potentially sensitive data. General features: TIDAL includes all the main steps of trajectory analysis including: 1) data preparation, (converting data from wide to long format); 2) data exploration, via basic plots and descriptive information; 3) analysis of trajectories using mixed effects modelling, interpretation of results, visualisation of trajectories, and extraction of key features (scores at different ages; area under the curve); and 4) interactions to derive population specific trajectories, combined with all the above. TIDAL is built with a simple graphical interface to guide users through each step. R syntax accompanies each step. Availability: Both versions of TIDAL can be found here: [https://tidal-modelling.github.io/].","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"37 8","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919313","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Synthetic cannabinoids consumed via e-cigarettes in English schools","authors":"G. Cozier, ‡. M. Gardner, ‡. S. Craft, Martine Skumlien, Jack Spicer, Rachael An-drews, Alexander Power, Tom S F Haines, Richard Bowman, Amy E. Manley, Peter Sunderland, Oliver B. Sutcliffe, Stephen M. Husbands, Lindsey Hines, Gillian Taylor, Tom P Freeman, Jennifer Scott, Christopher R. Pudney","doi":"10.1101/2024.08.12.24311617","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311617","url":null,"abstract":"Synthetic cannabinoids (SCs), colloquially spice or K2, are the most common drug to be found in prisons in the UK, where they are associated with nearly half of non-natural deaths. In the community, SCs are associated with poly-drug users who are also likely to be homeless. People who use SCs report debilitating side effects and withdrawal symptoms, coupled with dependence. Until now, SC use was believed to be largely restricted to prison and homeless populations. However, media reporting in the UK has increasingly identified cases of children collapsing in schools, which are claimed to be as-sociated with vaping and putatively the vaping of a drug, variously reported as tetrahydrocannabinol (THC) 'synthetic cannabis' or 'spice'. We therefore conducted the first study to identify and quantity SCs in e-cigarettes routinely collected from schools. We sampled 27 schools from geographically distinct regions of England, representing a very broad range of social metrics (free school meals, persistent absenteeism, and SEN). The material was sampled by self-submission by individual schools of e-cigarettes seized during normal school operation and transferred to us for analysis via local po-lice forces. We found a remarkably consistent picture where SCs were detected in 17.5 % of all e-cigarettes sampled, and in 21 of 27 (78 %) of all sampled schools. Moreover, the percentage of SC e-cigarettes positively correlated with a metric of social deprivation, the fraction of pupils eligible for free school meals. The SC positive e-cigarettes were almost entire-ly found in e-cigarette liquid bottles and refillable e-cigarette devices, with very few identified in single use e-cigarette products. Within the positive samples we found an average SC concentration of 1.03 mg/mL with a maximum of 3.6 mg mL-1. In contrast to the high prevalence of SCs, few samples contained THC (1.6 %). We suggest that pupils are being sold SC e-cigarettes as 'cannabis' and may be unaware they are consuming (and sometimes supplying) considerably more harmful drugs. Our findings are immediately crucial to policy policing and healthcare in the UK as well as to educational bodies and schools.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"38 9","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mapping Epigenetic Gene Variant Dynamics: Comparative Analysis of Frequency, Functional Impact and Trait Associations in African and European Populations","authors":"M. Sinkala, Gaone Retshabile, P. Mpangase, Salia Bamba, Modibo K Goita, Vicky Nembaware, S. Elsheikh, Jeannine Heckmann, K. Esoh, M. Matshaba, Clement A. Adebamowo, S. Adebamowo, Ofon Elvis, Amih, A. Wonkam, Michele Ramsay, Nicola Mulder","doi":"10.1101/2024.08.11.24311816","DOIUrl":"https://doi.org/10.1101/2024.08.11.24311816","url":null,"abstract":"Epigenetic modifications influence gene expression levels, impact organismal traits, and play a role in the development of diseases. Therefore, variants in genes involved in epigenetic processes are likely to be important in disease susceptibility, and the frequency of variants may vary between populations with African and European ancestries. Here, we analyse an integrated dataset to define the frequencies, associated traits, and functional impact of epigenetic gene variants among individuals of African and European ancestry represented in the UK Biobank. We find that the frequencies of 88.4% of epigenetic gene variants significantly differ between these groups. Furthermore, we find that the variants are associated with many traits and diseases, and some of these associations may be population-specific owing to allele frequency differences. Additionally, we observe that variants associated with traits are significantly enriched for quantitative trait loci that affect DNA methylation, chromatin accessibility, and gene expression. We find that methylation quantitative trait loci account for 71.2% of the variants influencing gene expression. Moreover, variants linked to biomarker traits exhibit high correlation. We therefore conclude that epigenetic gene variants associated with traits tend to differ in their allele frequencies among African and European populations and are enriched for QTLs.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"11 14","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919457","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Avoided and avoidable deaths with the use of COVID-19 convalescent plasma in Italy during the first two years of pandemic","authors":"M. Franchini, A. Casadevall, Q. Dragotakes, D. Focosi","doi":"10.1101/2024.08.12.24311864","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311864","url":null,"abstract":"Italy was the first western country to be hit by the COVID-19 pandemic and suffered nearly 200,000 deaths so far during the four years of the pandemic. In March 2020, Italy first deployed COVID-19 convalescent plasma (CCP) to treat hospitalized patients. Despite this initial effort, the proportion of COVID-19 patients treated with CCP during the first two years of the pandemic (2020-2021) was very low (approximately 2% of individuals hospitalized for COVID-19). In this study, we estimated the number of actual inpatient lives saved by CCP treatment in Italy using national mortality data, and CCP mortality reduction data from meta-analyses of randomized controlled trials and real-world data. We also estimated the potential number of lives saved if CCP had been deployed to 100% of hospitalized patients or used in 15% to 75% of outpatients. According to these models, CCP usage in 2020-2021 saved between 385-1304 lives , but this number would have increased to 17,751-60,079 if 100% of inpatients had been transfused with CCP. Similarly, broader (15-75%) usage in outpatients could have prevented 21,187-190,689 hospitalizations (desaturating hospitals) and 6,144-81,926 deaths. These data have important implications for convalescent plasma use in future infectious disease emergencies.","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"35 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generative Large Language Models in Electronic Health Records for Patient Care Since 2023: A Systematic Review","authors":"Xinsong Du, Zhengyang Zhou, Yifei Wang, Ya-Wen Chuang, Richard Yang, Wenyu Zhang, Xinyi Wang, Rui Zhang, Pengyu Hong, David W. Bates, Li Zhou","doi":"10.1101/2024.08.11.24311828","DOIUrl":"https://doi.org/10.1101/2024.08.11.24311828","url":null,"abstract":"Background: Generative Large language models (LLMs) represent a significant advancement in natural language processing, achieving state-of-the-art performance across various tasks. However, their application in clinical settings using real electronic health records (EHRs) is still rare and presents numerous challenges. Objective: This study aims to systematically review the use of generative LLMs in patient care-related topics involving EHRs, summarize the challenges faced, and suggest future directions. Methods: A Boolean search for peer-reviewed articles was conducted in May 2024 using PubMed and Web of Science to include research articles published since 2023, which was one month after the release of ChatGPT. The search results were deduplicated. Multiple reviewers, including biomedical informaticians, computer scientists, and a physician, screened the publications for eligibility and extracted bibliometric and clinically relevant information. Only papers utilizing generative LLMs to analyze real EHR data were included. We summarized the use of prompt engineering, fine-tuning, multimodal EHR data, and evaluation matrices. Additionally, we identified current challenges in applying LLMs in clinical settings as reported by the included papers and proposed future directions. Results: The initial search identified 6,328 unique studies, with 76 studies included after eligibility screening. Of these, 67 studies (88.2%) employed zero-shot prompting, five of them reported 100% accuracy on five specific clinical tasks. Nine studies used advanced prompting strategies; four tested these strategies experimentally, finding that prompt engineering improved performance, with one study noting a non-linear relationship between the number of examples in a prompt and performance improvement. Eight studies explored fine-tuning generative LLMs, all reported performance improvements on specific tasks, but three of them noted potential performance degradation after fine-tuning on certain tasks. Only two studies utilized multimodal data, which improved LLM-based decision-making and enabled accurate rare disease diagnosis and prognosis. The studies employed 55 different evaluation metrics for 22 purposes, such as correctness, completeness, and conciseness. Two studies investigated LLM bias, with one detecting no bias and the other finding that male patients received more appropriate clinical decision-making suggestions. Six studies identified hallucinations, such as fabricating patient names in structured thyroid ultrasound reports. Additional challenges included but not limited to the impersonal tone of LLM consultations, which made patients uncomfortable, and the difficulty patients had in understanding LLM responses. Conclusion: Our review indicates that few studies have employed advanced computational techniques to enhance LLM performance. The diverse evaluation metrics used highlight the need for standardization. LLMs currently cannot replace physicians due to cha","PeriodicalId":18505,"journal":{"name":"medRxiv","volume":"11 42","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141919450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}