Pub Date : 2025-01-10DOI: 10.1038/s41591-024-03343-6
Andrés Villaveces, Yu Chen, Sydney Tucker, Alexandra Blenkinsop, Lucie Cluver, Lorraine Sherr, Jan L. Losby, Linden Graves, Rita Noonan, Francis Annor, Victor Kojey-Merle, Douhan Wang, Greta Massetti, Laura Rawlings, Charles A. Nelson, H. Juliette T. Unwin, Seth Flaxman, Susan Hillis, Oliver Ratmann
Deaths of parents and grandparent caregivers threaten child well-being owing to losses of care, financial support, safety and family stability, but are relatively unrecognized as a public health crisis. Here we used cause-specific vital statistics death registrations in a modeling approach to estimate the full magnitude of orphanhood incidence and prevalence among US children aged 0–17 years between 2000 and 2021 by cause, child age, race and ethnicity, sex of deceased parent and state, and also accounted for grandparent caregiver loss using population survey data. In 2021, we estimate that 2.91 million children (4.2% of children) had in their lifetime experienced prevalent orphanhood and caregiver death combined, with incidence increasing by 49.5% and prevalence by 7.9% since 2000. Populations disproportionately affected by orphanhood included 5.2% of all adolescents; 6.4% and 4.7%, respectively, of non-Hispanic American Indian or Alaska Native, and non-Hispanic Black children; and children in southern and eastern states. In 2021, drug overdose was the leading cause of orphanhood among non-Hispanic white children, but not among minoritized subgroups. Effective policies and programs to support nearly three million bereaved children are needed to reduce the acute and long-term negative effects of orphanhood.
{"title":"Orphanhood and caregiver death among children in the United States by all-cause mortality, 2000–2021","authors":"Andrés Villaveces, Yu Chen, Sydney Tucker, Alexandra Blenkinsop, Lucie Cluver, Lorraine Sherr, Jan L. Losby, Linden Graves, Rita Noonan, Francis Annor, Victor Kojey-Merle, Douhan Wang, Greta Massetti, Laura Rawlings, Charles A. Nelson, H. Juliette T. Unwin, Seth Flaxman, Susan Hillis, Oliver Ratmann","doi":"10.1038/s41591-024-03343-6","DOIUrl":"https://doi.org/10.1038/s41591-024-03343-6","url":null,"abstract":"<p>Deaths of parents and grandparent caregivers threaten child well-being owing to losses of care, financial support, safety and family stability, but are relatively unrecognized as a public health crisis. Here we used cause-specific vital statistics death registrations in a modeling approach to estimate the full magnitude of orphanhood incidence and prevalence among US children aged 0–17 years between 2000 and 2021 by cause, child age, race and ethnicity, sex of deceased parent and state, and also accounted for grandparent caregiver loss using population survey data. In 2021, we estimate that 2.91 million children (4.2% of children) had in their lifetime experienced prevalent orphanhood and caregiver death combined, with incidence increasing by 49.5% and prevalence by 7.9% since 2000. Populations disproportionately affected by orphanhood included 5.2% of all adolescents; 6.4% and 4.7%, respectively, of non-Hispanic American Indian or Alaska Native, and non-Hispanic Black children; and children in southern and eastern states. In 2021, drug overdose was the leading cause of orphanhood among non-Hispanic white children, but not among minoritized subgroups. Effective policies and programs to support nearly three million bereaved children are needed to reduce the acute and long-term negative effects of orphanhood.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"14 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142961387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-09DOI: 10.1038/s41591-024-03431-7
Zhanwei Du, Abhishek Pandey, Seyed M. Moghadas, Yuan Bai, Lin Wang, Laura Matrajt, Burton H. Singer, Alison P. Galvani
Respiratory syncytial virus (RSV) causes a substantial health burden among infants and older adults. Prefusion F protein-based vaccines have shown high efficacy against RSV disease in clinical trials, offering promise for mitigating this burden through maternal and older adult immunization. Employing an individual-based model, we evaluated the impact of RSV vaccination on hospitalizations and deaths in 13 high-income countries, assuming that the vaccine does not prevent infection or transmission. Using country-specific vaccine uptake rates for seasonal influenza, we found that vaccination of older adults would prevent hospitalizations by a median of 35–64% across the countries studied here. Vaccination of pregnant women could avert infant hospitalizations by 5–50%. Reductions in RSV-related mortality mirrored those estimated for hospitalizations. While substantial hospitalization costs could be averted, the impact of vaccination depends critically on uptake rates. Enhancing uptake and accessibility is crucial for maximizing the real-world impact of vaccination on reducing RSV burden among vulnerable populations.
{"title":"Impact of RSVpreF vaccination on reducing the burden of respiratory syncytial virus in infants and older adults","authors":"Zhanwei Du, Abhishek Pandey, Seyed M. Moghadas, Yuan Bai, Lin Wang, Laura Matrajt, Burton H. Singer, Alison P. Galvani","doi":"10.1038/s41591-024-03431-7","DOIUrl":"https://doi.org/10.1038/s41591-024-03431-7","url":null,"abstract":"<p>Respiratory syncytial virus (RSV) causes a substantial health burden among infants and older adults. Prefusion F protein-based vaccines have shown high efficacy against RSV disease in clinical trials, offering promise for mitigating this burden through maternal and older adult immunization. Employing an individual-based model, we evaluated the impact of RSV vaccination on hospitalizations and deaths in 13 high-income countries, assuming that the vaccine does not prevent infection or transmission. Using country-specific vaccine uptake rates for seasonal influenza, we found that vaccination of older adults would prevent hospitalizations by a median of 35–64% across the countries studied here. Vaccination of pregnant women could avert infant hospitalizations by 5–50%. Reductions in RSV-related mortality mirrored those estimated for hospitalizations. While substantial hospitalization costs could be averted, the impact of vaccination depends critically on uptake rates. Enhancing uptake and accessibility is crucial for maximizing the real-world impact of vaccination on reducing RSV burden among vulnerable populations.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"44 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142937534","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Premarital genomic screening can help curb the burden of rare recessive disorders in populations with high rates of consanguineous marriages. Arab populations of the Middle East have a unique genetic blend characterized by founder mutations due to historical bottlenecks, and diverse allelic series shaped by a long history of migrations and intermixing. The risk of homozygous pathogenic variants, which can lead to recessive genetic disorders, is elevated because of the high prevalence of consanguineous marriages, particularly unions between first cousins, culturally rooted in this region1. As a result, the cumulative incidence of rare recessive disorders is expected to be higher in Arab populations compared to others, often leading to chronic, severe and life-limiting conditions, such as hemoglobinopathies, spinal muscular atrophy, congenital malformations and metabolic disorders2. This places an immense emotional and financial strain on affected families and healthcare systems across the region.
Premarital genomic screening presents a promising solution to mitigate this growing burden. By identifying carrier couples before marriage (particularly those who may unknowingly harbor the same deleterious variants), this approach provides critical information that enables informed reproductive decisions and could substantially reduce the incidence of genetic disorders and their associated burden across the region. Couples identified as carriers can seek genetic counseling to understand the implications of their results and to explore a range of options, including in vitro fertilization and pre-implantation genetic diagnosis or alternative family planning strategies. Regions that have successfully implemented screening programs have seen a marked decline in the incidence of recessive genetic conditions, further demonstrating their efficacy3,4,5. Growing evidence from the region suggests that carrier screening offers a cost-effective preventive strategy relative to postnatal treatment or other management alternatives6.
{"title":"Premarital genomic screening in Arab populations of the Middle East","authors":"Ikram Chekroun, Fatma Rabea, Ruchi Jain, Alawi Alsheikh-Ali, Ahmad Abou Tayoun","doi":"10.1038/s41591-024-03442-4","DOIUrl":"https://doi.org/10.1038/s41591-024-03442-4","url":null,"abstract":"<p>Premarital genomic screening can help curb the burden of rare recessive disorders in populations with high rates of consanguineous marriages. Arab populations of the Middle East have a unique genetic blend characterized by founder mutations due to historical bottlenecks, and diverse allelic series shaped by a long history of migrations and intermixing. The risk of homozygous pathogenic variants, which can lead to recessive genetic disorders, is elevated because of the high prevalence of consanguineous marriages, particularly unions between first cousins, culturally rooted in this region<sup>1</sup>. As a result, the cumulative incidence of rare recessive disorders is expected to be higher in Arab populations compared to others, often leading to chronic, severe and life-limiting conditions, such as hemoglobinopathies, spinal muscular atrophy, congenital malformations and metabolic disorders<sup>2</sup>. This places an immense emotional and financial strain on affected families and healthcare systems across the region.</p><p>Premarital genomic screening presents a promising solution to mitigate this growing burden. By identifying carrier couples before marriage (particularly those who may unknowingly harbor the same deleterious variants), this approach provides critical information that enables informed reproductive decisions and could substantially reduce the incidence of genetic disorders and their associated burden across the region. Couples identified as carriers can seek genetic counseling to understand the implications of their results and to explore a range of options, including in vitro fertilization and pre-implantation genetic diagnosis or alternative family planning strategies. Regions that have successfully implemented screening programs have seen a marked decline in the incidence of recessive genetic conditions, further demonstrating their efficacy<sup>3,4,5</sup>. Growing evidence from the region suggests that carrier screening offers a cost-effective preventive strategy relative to postnatal treatment or other management alternatives<sup>6</sup>.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"23 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142937532","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03429-1
Bartley P. Griffith, Alison Grazioli, Avneesh K. Singh, Andy Tully, Javier Galindo, Kapil K. Saharia, Aakash Shah, Erik R. Strauss, Patrick N. Odonkor, Brittney Williams, Henry J. Silverman, Allen Burke, Cinthia B. Drachenberg, Chris L. Wells, Timm Dickfeld, Susie N. Hong, Albert J. Hicks, Manjula Ananthram, Anuj Gupta, Robert H. Christenson, Lo Tamburro, Tianshu Zhang, Alena Hershfeld, Billeta Lewis, Erika D. Feller, Kasinath Kuravi, Lori Sorrells, Erwan Morgand, Fariza Mezine, Valentin Goutaudier, Martine Rothblatt, Christine L. Lau, Bradley Taylor, Steve Perrin, Alexandre Loupy, David Ayares, Muhammad M. Mohiuddin
Following our previous experience with cardiac xenotransplantation of a genetically modified porcine heart into a live human, we sought to achieve improved results by selecting a healthier recipient and through more sensitive donor screening for potential zoonotic pathogens. Here we transplanted a 10-gene-edited pig heart into a 58-year-old man with progressive, debilitating inotrope-dependent heart failure due to ischemic cardiomyopathy who was not a candidate for standard advanced heart failure therapies. He was maintained on a costimulation (anti-CD40L, Tegoprubart) blockade-based immunomodulatory regimen. The xenograft initially functioned well, with excellent systolic and diastolic function during the first several weeks posttransplantation. Subsequently, the xenograft developed rapidly progressing diastolic heart failure, biventricular wall thickening and, ultimately, near-complete loss of systolic function necessitating initiation of extracorporeal membranous oxygenation on day 31. Given these setbacks, the patient chose to transition to comfort care after 40 days. As with our first patient, histology did not reveal substantial immune cell infiltration but suggested capillary endothelial injury with interstitial edema and early fibrosis. No evidence of porcine cytomegalovirus replication in the xenograft was observed. Strategies to overcome the obstacle of antibody-mediated rejection are needed to advance the field of xenotransplantation.
{"title":"Transplantation of a genetically modified porcine heart into a live human","authors":"Bartley P. Griffith, Alison Grazioli, Avneesh K. Singh, Andy Tully, Javier Galindo, Kapil K. Saharia, Aakash Shah, Erik R. Strauss, Patrick N. Odonkor, Brittney Williams, Henry J. Silverman, Allen Burke, Cinthia B. Drachenberg, Chris L. Wells, Timm Dickfeld, Susie N. Hong, Albert J. Hicks, Manjula Ananthram, Anuj Gupta, Robert H. Christenson, Lo Tamburro, Tianshu Zhang, Alena Hershfeld, Billeta Lewis, Erika D. Feller, Kasinath Kuravi, Lori Sorrells, Erwan Morgand, Fariza Mezine, Valentin Goutaudier, Martine Rothblatt, Christine L. Lau, Bradley Taylor, Steve Perrin, Alexandre Loupy, David Ayares, Muhammad M. Mohiuddin","doi":"10.1038/s41591-024-03429-1","DOIUrl":"https://doi.org/10.1038/s41591-024-03429-1","url":null,"abstract":"<p>Following our previous experience with cardiac xenotransplantation of a genetically modified porcine heart into a live human, we sought to achieve improved results by selecting a healthier recipient and through more sensitive donor screening for potential zoonotic pathogens. Here we transplanted a 10-gene-edited pig heart into a 58-year-old man with progressive, debilitating inotrope-dependent heart failure due to ischemic cardiomyopathy who was not a candidate for standard advanced heart failure therapies. He was maintained on a costimulation (anti-CD40L, Tegoprubart) blockade-based immunomodulatory regimen. The xenograft initially functioned well, with excellent systolic and diastolic function during the first several weeks posttransplantation. Subsequently, the xenograft developed rapidly progressing diastolic heart failure, biventricular wall thickening and, ultimately, near-complete loss of systolic function necessitating initiation of extracorporeal membranous oxygenation on day 31. Given these setbacks, the patient chose to transition to comfort care after 40 days. As with our first patient, histology did not reveal substantial immune cell infiltration but suggested capillary endothelial injury with interstitial edema and early fibrosis. No evidence of porcine cytomegalovirus replication in the xenograft was observed. Strategies to overcome the obstacle of antibody-mediated rejection are needed to advance the field of xenotransplantation.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"5092 3 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03445-1
Daniel Alexander Alber, Zihao Yang, Anton Alyakin, Eunice Yang, Sumedha Rai, Aly A. Valliani, Jeff Zhang, Gabriel R. Rosenbaum, Ashley K. Amend-Thomas, David B. Kurland, Caroline M. Kremer, Alexander Eremiev, Bruck Negash, Daniel D. Wiggan, Michelle A. Nakatsuka, Karl L. Sangwon, Sean N. Neifert, Hammad A. Khan, Akshay Vinod Save, Adhith Palla, Eric A. Grin, Monika Hedman, Mustafa Nasir-Moin, Xujin Chris Liu, Lavender Yao Jiang, Michal A. Mankowski, Dorry L. Segev, Yindalon Aphinyanaphongs, Howard A. Riina, John G. Golfinos, Daniel A. Orringer, Douglas Kondziolka, Eric Karl Oermann
The adoption of large language models (LLMs) in healthcare demands a careful analysis of their potential to spread false medical knowledge. Because LLMs ingest massive volumes of data from the open Internet during training, they are potentially exposed to unverified medical knowledge that may include deliberately planted misinformation. Here, we perform a threat assessment that simulates a data-poisoning attack against The Pile, a popular dataset used for LLM development. We find that replacement of just 0.001% of training tokens with medical misinformation results in harmful models more likely to propagate medical errors. Furthermore, we discover that corrupted models match the performance of their corruption-free counterparts on open-source benchmarks routinely used to evaluate medical LLMs. Using biomedical knowledge graphs to screen medical LLM outputs, we propose a harm mitigation strategy that captures 91.9% of harmful content (F1 = 85.7%). Our algorithm provides a unique method to validate stochastically generated LLM outputs against hard-coded relationships in knowledge graphs. In view of current calls for improved data provenance and transparent LLM development, we hope to raise awareness of emergent risks from LLMs trained indiscriminately on web-scraped data, particularly in healthcare where misinformation can potentially compromise patient safety.
{"title":"Medical large language models are vulnerable to data-poisoning attacks","authors":"Daniel Alexander Alber, Zihao Yang, Anton Alyakin, Eunice Yang, Sumedha Rai, Aly A. Valliani, Jeff Zhang, Gabriel R. Rosenbaum, Ashley K. Amend-Thomas, David B. Kurland, Caroline M. Kremer, Alexander Eremiev, Bruck Negash, Daniel D. Wiggan, Michelle A. Nakatsuka, Karl L. Sangwon, Sean N. Neifert, Hammad A. Khan, Akshay Vinod Save, Adhith Palla, Eric A. Grin, Monika Hedman, Mustafa Nasir-Moin, Xujin Chris Liu, Lavender Yao Jiang, Michal A. Mankowski, Dorry L. Segev, Yindalon Aphinyanaphongs, Howard A. Riina, John G. Golfinos, Daniel A. Orringer, Douglas Kondziolka, Eric Karl Oermann","doi":"10.1038/s41591-024-03445-1","DOIUrl":"https://doi.org/10.1038/s41591-024-03445-1","url":null,"abstract":"<p>The adoption of large language models (LLMs) in healthcare demands a careful analysis of their potential to spread false medical knowledge. Because LLMs ingest massive volumes of data from the open Internet during training, they are potentially exposed to unverified medical knowledge that may include deliberately planted misinformation. Here, we perform a threat assessment that simulates a data-poisoning attack against The Pile, a popular dataset used for LLM development. We find that replacement of just 0.001% of training tokens with medical misinformation results in harmful models more likely to propagate medical errors. Furthermore, we discover that corrupted models match the performance of their corruption-free counterparts on open-source benchmarks routinely used to evaluate medical LLMs. Using biomedical knowledge graphs to screen medical LLM outputs, we propose a harm mitigation strategy that captures 91.9% of harmful content (F1 = 85.7%). Our algorithm provides a unique method to validate stochastically generated LLM outputs against hard-coded relationships in knowledge graphs. In view of current calls for improved data provenance and transparent LLM development, we hope to raise awareness of emergent risks from LLMs trained indiscriminately on web-scraped data, particularly in healthcare where misinformation can potentially compromise patient safety.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"28 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03425-5
Jack Gallifant, Majid Afshar, Saleem Ameen, Yindalon Aphinyanaphongs, Shan Chen, Giovanni Cacciamani, Dina Demner-Fushman, Dmitriy Dligach, Roxana Daneshjou, Chrystinne Fernandes, Lasse Hyldig Hansen, Adam Landman, Lisa Lehmann, Liam G. McCoy, Timothy Miller, Amy Moreno, Nikolaj Munch, David Restrepo, Guergana Savova, Renato Umeton, Judy Wawira Gichoya, Gary S. Collins, Karel G. M. Moons, Leo A. Celi, Danielle S. Bitterman
Large language models (LLMs) are rapidly being adopted in healthcare, necessitating standardized reporting guidelines. We present transparent reporting of a multivariable model for individual prognosis or diagnosis (TRIPOD)-LLM, an extension of the TRIPOD + artificial intelligence statement, addressing the unique challenges of LLMs in biomedical applications. TRIPOD-LLM provides a comprehensive checklist of 19 main items and 50 subitems, covering key aspects from title to discussion. The guidelines introduce a modular format accommodating various LLM research designs and tasks, with 14 main items and 32 subitems applicable across all categories. Developed through an expedited Delphi process and expert consensus, TRIPOD-LLM emphasizes transparency, human oversight and task-specific performance reporting. We also introduce an interactive website ( https://tripod-llm.vercel.app/ ) facilitating easy guideline completion and PDF generation for submission. As a living document, TRIPOD-LLM will evolve with the field, aiming to enhance the quality, reproducibility and clinical applicability of LLM research in healthcare through comprehensive reporting. TRIPOD-LLM (transparent reporting of a multivariable model for individual prognosis or diagnosis–large language model) is a checklist of items considered essential for good reporting of studies that are developing or evaluating an LLM for use in healthcare settings. It is a ‘living guideline’ that emphasizes transparency, human oversight and task-specific performance reporting.
{"title":"The TRIPOD-LLM reporting guideline for studies using large language models","authors":"Jack Gallifant, Majid Afshar, Saleem Ameen, Yindalon Aphinyanaphongs, Shan Chen, Giovanni Cacciamani, Dina Demner-Fushman, Dmitriy Dligach, Roxana Daneshjou, Chrystinne Fernandes, Lasse Hyldig Hansen, Adam Landman, Lisa Lehmann, Liam G. McCoy, Timothy Miller, Amy Moreno, Nikolaj Munch, David Restrepo, Guergana Savova, Renato Umeton, Judy Wawira Gichoya, Gary S. Collins, Karel G. M. Moons, Leo A. Celi, Danielle S. Bitterman","doi":"10.1038/s41591-024-03425-5","DOIUrl":"10.1038/s41591-024-03425-5","url":null,"abstract":"Large language models (LLMs) are rapidly being adopted in healthcare, necessitating standardized reporting guidelines. We present transparent reporting of a multivariable model for individual prognosis or diagnosis (TRIPOD)-LLM, an extension of the TRIPOD + artificial intelligence statement, addressing the unique challenges of LLMs in biomedical applications. TRIPOD-LLM provides a comprehensive checklist of 19 main items and 50 subitems, covering key aspects from title to discussion. The guidelines introduce a modular format accommodating various LLM research designs and tasks, with 14 main items and 32 subitems applicable across all categories. Developed through an expedited Delphi process and expert consensus, TRIPOD-LLM emphasizes transparency, human oversight and task-specific performance reporting. We also introduce an interactive website ( https://tripod-llm.vercel.app/ ) facilitating easy guideline completion and PDF generation for submission. As a living document, TRIPOD-LLM will evolve with the field, aiming to enhance the quality, reproducibility and clinical applicability of LLM research in healthcare through comprehensive reporting. TRIPOD-LLM (transparent reporting of a multivariable model for individual prognosis or diagnosis–large language model) is a checklist of items considered essential for good reporting of studies that are developing or evaluating an LLM for use in healthcare settings. It is a ‘living guideline’ that emphasizes transparency, human oversight and task-specific performance reporting.","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"31 1","pages":"60-69"},"PeriodicalIF":58.7,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s41591-024-03425-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936182","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03423-7
Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Mohamed Amin, Le Hou, Kevin Clark, Stephen R. Pfohl, Heather Cole-Lewis, Darlene Neal, Qazi Mamunur Rashid, Mike Schaekermann, Amy Wang, Dev Dash, Jonathan H. Chen, Nigam H. Shah, Sami Lachgar, Philip Andrew Mansfield, Sushant Prakash, Bradley Green, Ewa Dominowska, Blaise Agüera y Arcas, Nenad Tomašev, Yun Liu, Renee Wong, Christopher Semturs, S. Sara Mahdavi, Joelle K. Barral, Dale R. Webster, Greg S. Corrado, Yossi Matias, Shekoofeh Azizi, Alan Karthikesalingam, Vivek Natarajan
Large language models (LLMs) have shown promise in medical question answering, with Med-PaLM being the first to exceed a ‘passing’ score in United States Medical Licensing Examination style questions. However, challenges remain in long-form medical question answering and handling real-world workflows. Here, we present Med-PaLM 2, which bridges these gaps with a combination of base LLM improvements, medical domain fine-tuning and new strategies for improving reasoning and grounding through ensemble refinement and chain of retrieval. Med-PaLM 2 scores up to 86.5% on the MedQA dataset, improving upon Med-PaLM by over 19%, and demonstrates dramatic performance increases across MedMCQA, PubMedQA and MMLU clinical topics datasets. Our detailed human evaluations framework shows that physicians prefer Med-PaLM 2 answers to those from other physicians on eight of nine clinical axes. Med-PaLM 2 also demonstrates significant improvements over its predecessor across all evaluation metrics, particularly on new adversarial datasets designed to probe LLM limitations (P < 0.001). In a pilot study using real-world medical questions, specialists preferred Med-PaLM 2 answers to generalist physician answers 65% of the time. While specialist answers were still preferred overall, both specialists and generalists rated Med-PaLM 2 to be as safe as physician answers, demonstrating its growing potential in real-world medical applications.
{"title":"Toward expert-level medical question answering with large language models","authors":"Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Mohamed Amin, Le Hou, Kevin Clark, Stephen R. Pfohl, Heather Cole-Lewis, Darlene Neal, Qazi Mamunur Rashid, Mike Schaekermann, Amy Wang, Dev Dash, Jonathan H. Chen, Nigam H. Shah, Sami Lachgar, Philip Andrew Mansfield, Sushant Prakash, Bradley Green, Ewa Dominowska, Blaise Agüera y Arcas, Nenad Tomašev, Yun Liu, Renee Wong, Christopher Semturs, S. Sara Mahdavi, Joelle K. Barral, Dale R. Webster, Greg S. Corrado, Yossi Matias, Shekoofeh Azizi, Alan Karthikesalingam, Vivek Natarajan","doi":"10.1038/s41591-024-03423-7","DOIUrl":"https://doi.org/10.1038/s41591-024-03423-7","url":null,"abstract":"<p>Large language models (LLMs) have shown promise in medical question answering, with Med-PaLM being the first to exceed a ‘passing’ score in United States Medical Licensing Examination style questions. However, challenges remain in long-form medical question answering and handling real-world workflows. Here, we present Med-PaLM 2, which bridges these gaps with a combination of base LLM improvements, medical domain fine-tuning and new strategies for improving reasoning and grounding through ensemble refinement and chain of retrieval. Med-PaLM 2 scores up to 86.5% on the MedQA dataset, improving upon Med-PaLM by over 19%, and demonstrates dramatic performance increases across MedMCQA, PubMedQA and MMLU clinical topics datasets. Our detailed human evaluations framework shows that physicians prefer Med-PaLM 2 answers to those from other physicians on eight of nine clinical axes. Med-PaLM 2 also demonstrates significant improvements over its predecessor across all evaluation metrics, particularly on new adversarial datasets designed to probe LLM limitations (<i>P</i> < 0.001). In a pilot study using real-world medical questions, specialists preferred Med-PaLM 2 answers to generalist physician answers 65% of the time. While specialist answers were still preferred overall, both specialists and generalists rated Med-PaLM 2 to be as safe as physician answers, demonstrating its growing potential in real-world medical applications.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"34 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936184","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03422-8
Alissa Muller, Jack Sullivan, Wibke Schwarzer, Mantian Wang, Cindy Park-Windhol, Pascal W. Hasler, Lucas Janeschitz-Kriegl, Mert Duman, Beryll Klingler, Jane Matsell, Simon Manuel Hostettler, Patricia Galliker, Yanyan Hou, Pierre Balmer, Tamás Virág, Luis Alberto Barrera, Lauren Young, Quan Xu, Dániel Péter Magda, Ferenc Kilin, Arogya Khadka, Pierre-Henri Moreau, Lyne Fellmann, Thierry Azoulay, Mathieu Quinodoz, Duygu Karademir, Juna Leppert, Alex Fratzl, Georg Kosche, Ruchi Sharma, Jair Montford, Marco Cattaneo, Mikaël Croyal, Therese Cronin, Simone Picelli, Alice Grison, Cameron S. Cowan, Ákos Kusnyerik, Philipp Anders, Magdalena Renner, Zoltán Zsolt Nagy, Arnold Szabó, Kapil Bharti, Carlo Rivolta, Hendrik P. N. Scholl, David Bryson, Giuseppe Ciaramella, Botond Roska, Bence György
Stargardt disease is a currently untreatable, inherited neurodegenerative disease that leads to macular degeneration and blindness due to loss-of-function mutations in the ABCA4 gene. We have designed a dual adeno-associated viral vector encoding a split-intein adenine base editor to correct the most common mutation in ABCA4 (c.5882G>A, p.Gly1961Glu). We optimized ABCA4 base editing in human models, including retinal organoids, induced pluripotent stem cell-derived retinal pigment epithelial (RPE) cells, as well as adult human retinal explants and RPE/choroid explants in vitro. The resulting gene therapy vectors achieved high levels of gene correction in mutation-carrying mice and in female nonhuman primates, with average editing of 75% of cones and 87% of RPE cells in vivo, which has the potential to translate to a clinical benefit. No off-target editing was detectable in human retinal explants and RPE/choroid explants. The high editing rates in primates show promise for efficient gene editing in other ocular diseases that are targetable by base editing.
Stargardt病是一种目前无法治疗的遗传性神经退行性疾病,由于ABCA4基因的功能丧失突变导致黄斑变性和失明。我们设计了一种双腺相关病毒载体,编码分裂间质腺嘌呤碱基编辑器,以纠正ABCA4中最常见的突变(c.5882G> a, p.Gly1961Glu)。我们在人体模型中优化了ABCA4碱基编辑,包括视网膜类器官、诱导多能干细胞衍生的视网膜色素上皮(RPE)细胞,以及体外成人视网膜外植体和RPE/choroid外植体。由此产生的基因治疗载体在携带突变的小鼠和雌性非人灵长类动物中实现了高水平的基因校正,在体内平均编辑了75%的视锥细胞和87%的RPE细胞,这有可能转化为临床益处。在人视网膜外植体和RPE/脉络膜外植体中未检测到脱靶编辑。灵长类动物的高编辑率显示了对其他可通过碱基编辑靶向的眼部疾病进行有效基因编辑的希望。
{"title":"High-efficiency base editing in the retina in primates and human tissues","authors":"Alissa Muller, Jack Sullivan, Wibke Schwarzer, Mantian Wang, Cindy Park-Windhol, Pascal W. Hasler, Lucas Janeschitz-Kriegl, Mert Duman, Beryll Klingler, Jane Matsell, Simon Manuel Hostettler, Patricia Galliker, Yanyan Hou, Pierre Balmer, Tamás Virág, Luis Alberto Barrera, Lauren Young, Quan Xu, Dániel Péter Magda, Ferenc Kilin, Arogya Khadka, Pierre-Henri Moreau, Lyne Fellmann, Thierry Azoulay, Mathieu Quinodoz, Duygu Karademir, Juna Leppert, Alex Fratzl, Georg Kosche, Ruchi Sharma, Jair Montford, Marco Cattaneo, Mikaël Croyal, Therese Cronin, Simone Picelli, Alice Grison, Cameron S. Cowan, Ákos Kusnyerik, Philipp Anders, Magdalena Renner, Zoltán Zsolt Nagy, Arnold Szabó, Kapil Bharti, Carlo Rivolta, Hendrik P. N. Scholl, David Bryson, Giuseppe Ciaramella, Botond Roska, Bence György","doi":"10.1038/s41591-024-03422-8","DOIUrl":"https://doi.org/10.1038/s41591-024-03422-8","url":null,"abstract":"<p>Stargardt disease is a currently untreatable, inherited neurodegenerative disease that leads to macular degeneration and blindness due to loss-of-function mutations in the <i>ABCA4</i> gene. We have designed a dual adeno-associated viral vector encoding a split-intein adenine base editor to correct the most common mutation in <i>ABCA4</i> (c.5882G>A, p.Gly1961Glu). We optimized <i>ABCA4</i> base editing in human models, including retinal organoids, induced pluripotent stem cell-derived retinal pigment epithelial (RPE) cells, as well as adult human retinal explants and RPE/choroid explants in vitro. The resulting gene therapy vectors achieved high levels of gene correction in mutation-carrying mice and in female nonhuman primates, with average editing of 75% of cones and 87% of RPE cells in vivo, which has the potential to translate to a clinical benefit. No off-target editing was detectable in human retinal explants and RPE/choroid explants. The high editing rates in primates show promise for efficient gene editing in other ocular diseases that are targetable by base editing.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"15 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-08DOI: 10.1038/s41591-024-03384-x
Emilie Combet, Laura Haag, Janice Richardson, Caroline E. Haig, Yvonne Cunningham, Heather L. Fraser, Naomi Brosnahan, Tracy Ibbotson, Jane Ormerod, Chris White, Emma McIntosh, Catherine A. O’Donnell, Naveed Sattar, Alex McConnachie, Michael E. J. Lean, David N. Blane
Long COVID (LC) is a complex multisymptom condition with no known disease-modifying treatments. This wait-list-controlled open-label trial tested whether a remotely delivered structured weight management program could improve respective LC symptoms in people living with overweight. Adults with LC (symptoms >12 weeks) and body mass index >27 kg m−2 (>25 kg m−2 for South Asians) were randomized (n = 234, 1:1) to control (n = 116, usual care) or the remotely delivered structured weight management (n = 118, total diet replacement (850 kcal per day) for 12 weeks, followed by food reintroduction and weight loss maintenance support) via minimization and randomization (80:20) to balance dominant LC symptom, sex, age, ethnicity and postcode-based index of multiple deprivation between groups. The control group received the intervention after 6 months. Participants selected the dominant LC symptom they would most like to improve (fatigue, breathlessness, pain, anxiety/depression or other) as the prespecified respective primary outcome. Individual symptoms were assessed using validated questionnaires and a visual analog scale for those without prespecified scales. At 6 months, the primary outcome improved in the intervention group (change −1.16 (s.d. 1.42), n = 97 analyzed) compared with the control group (change −0.83 (s.d. 1.14), n = 117 analyzed) with a treatment effect of −0.34 (95% confidence interval −0.67 to −0.01), with no excess of serious adverse events. International Standard Randomised Controlled Trial Number Registry registration: ISRCTN 12595520 . In a randomized trial, a remotely delivered weight management intervention for people with long COVID and excess weight improved respective long COVID symptoms and quality of life and resulted in substantial weight loss when compared with a control group, with no serious adverse effects reported.
长冠状病毒病(LC)是一种复杂的多症状疾病,目前尚无已知的改善疾病的治疗方法。这项等待名单控制的开放标签试验测试了远程交付的结构化体重管理程序是否可以改善超重人群的LC症状。患有LC(症状>;12周)和体重指数>;27 kg m - 2 (>;25 kg m - 2的南亚人)的成年人被随机(n = 234, 1:1)随机分配到对照组(n = 116,常规护理)或远程提供结构化体重管理(n = 118,总饮食替代(每天850千卡),持续12周,然后重新引入食物和减肥维持支持),通过最小化和随机化(80:20)来平衡主要的LC症状、性别、年龄、基于种族和邮政编码的群体多重剥夺指数。对照组于6个月后接受干预。参与者选择他们最想改善的主要LC症状(疲劳、呼吸困难、疼痛、焦虑/抑郁或其他)作为预先指定的各自的主要结局。使用有效的问卷和视觉模拟量表对没有预先指定量表的患者进行个体症状评估。6个月时,干预组的主要结局(变化- 1.16(标准差1.42),分析n = 97)较对照组(变化- 0.83(标准差1.14),分析n = 117)有所改善,治疗效果为- 0.34(95%可信区间- 0.67至- 0.01),未发生过多严重不良事件。国际标准随机对照试验号注册:ISRCTN12595520。
{"title":"Remotely delivered weight management for people with long COVID and overweight: the randomized wait-list-controlled ReDIRECT trial","authors":"Emilie Combet, Laura Haag, Janice Richardson, Caroline E. Haig, Yvonne Cunningham, Heather L. Fraser, Naomi Brosnahan, Tracy Ibbotson, Jane Ormerod, Chris White, Emma McIntosh, Catherine A. O’Donnell, Naveed Sattar, Alex McConnachie, Michael E. J. Lean, David N. Blane","doi":"10.1038/s41591-024-03384-x","DOIUrl":"10.1038/s41591-024-03384-x","url":null,"abstract":"Long COVID (LC) is a complex multisymptom condition with no known disease-modifying treatments. This wait-list-controlled open-label trial tested whether a remotely delivered structured weight management program could improve respective LC symptoms in people living with overweight. Adults with LC (symptoms >12 weeks) and body mass index >27 kg m−2 (>25 kg m−2 for South Asians) were randomized (n = 234, 1:1) to control (n = 116, usual care) or the remotely delivered structured weight management (n = 118, total diet replacement (850 kcal per day) for 12 weeks, followed by food reintroduction and weight loss maintenance support) via minimization and randomization (80:20) to balance dominant LC symptom, sex, age, ethnicity and postcode-based index of multiple deprivation between groups. The control group received the intervention after 6 months. Participants selected the dominant LC symptom they would most like to improve (fatigue, breathlessness, pain, anxiety/depression or other) as the prespecified respective primary outcome. Individual symptoms were assessed using validated questionnaires and a visual analog scale for those without prespecified scales. At 6 months, the primary outcome improved in the intervention group (change −1.16 (s.d. 1.42), n = 97 analyzed) compared with the control group (change −0.83 (s.d. 1.14), n = 117 analyzed) with a treatment effect of −0.34 (95% confidence interval −0.67 to −0.01), with no excess of serious adverse events. International Standard Randomised Controlled Trial Number Registry registration: ISRCTN 12595520 . In a randomized trial, a remotely delivered weight management intervention for people with long COVID and excess weight improved respective long COVID symptoms and quality of life and resulted in substantial weight loss when compared with a control group, with no serious adverse effects reported.","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"31 1","pages":"258-266"},"PeriodicalIF":58.7,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s41591-024-03384-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Up to 50–70% of patients with liver cirrhosis develop hepatic encephalopathy (HE), which is closely related to gut microbiota dysbiosis, with an unclear mechanism. Here, by constructing gut–brain modules to assess bacterial neurotoxins from metagenomic datasets, we found that phenylalanine decarboxylase (PDC) genes, mainly from Ruminococcus gnavus, increased approximately tenfold in patients with cirrhosis and higher in patients with HE. Cirrhotic, not healthy, mice colonized with R. gnavus showed brain phenylethylamine (PEA) accumulation, along with memory impairment, symmetrical tremors and cortex-specific neuron loss, typically found in patients with HE. This accumulation of PEA was primarily driven by decreased monoamine oxidase-B activity in both the liver and serum due to cirrhosis. Targeting PDC or PEA reversed the neurological symptoms induced by R. gnavus. Furthermore, fecal microbiota transplantation from patients with HE to germ-free cirrhotic mice replicated these symptoms and further corroborated the efficacy of targeting PDC or PEA. Clinically, high baseline PEA levels were linked to a sevenfold increased risk of HE after intrahepatic portosystemic shunt procedures. Our findings expand the understanding of the gut–liver–brain axis and identify a promising therapeutic and predictive target for HE.
{"title":"The gut–brain axis underlying hepatic encephalopathy in liver cirrhosis","authors":"Xiaolong He, Mengyao Hu, Yi Xu, Fangbo Xia, Yang Tan, Yuqing Wang, Huiling Xiang, Hao Wu, Tengfei Ji, Qian Xu, Lei Wang, Zhenhe Huang, Meiling Sun, Yu Wan, Peng Cui, Shaocong Liang, Yuan Pan, Siyu Xiao, Yan He, Ruixin Song, Junqing Yan, Xin Quan, Yingge Wei, Changze Hong, Weizuo Liao, Fuli Li, Emad El-Omar, Jinjun Chen, Xiaolong Qi, Jie Gao, Hongwei Zhou","doi":"10.1038/s41591-024-03405-9","DOIUrl":"https://doi.org/10.1038/s41591-024-03405-9","url":null,"abstract":"<p>Up to 50–70% of patients with liver cirrhosis develop hepatic encephalopathy (HE), which is closely related to gut microbiota dysbiosis, with an unclear mechanism. Here, by constructing gut–brain modules to assess bacterial neurotoxins from metagenomic datasets, we found that phenylalanine decarboxylase (PDC) genes, mainly from <i>Ruminococcus gnavus</i>, increased approximately tenfold in patients with cirrhosis and higher in patients with HE. Cirrhotic, not healthy, mice colonized with <i>R. gnavus</i> showed brain phenylethylamine (PEA) accumulation, along with memory impairment, symmetrical tremors and cortex-specific neuron loss, typically found in patients with HE. This accumulation of PEA was primarily driven by decreased monoamine oxidase-B activity in both the liver and serum due to cirrhosis. Targeting PDC or PEA reversed the neurological symptoms induced by <i>R. gnavus</i>. Furthermore, fecal microbiota transplantation from patients with HE to germ-free cirrhotic mice replicated these symptoms and further corroborated the efficacy of targeting PDC or PEA. Clinically, high baseline PEA levels were linked to a sevenfold increased risk of HE after intrahepatic portosystemic shunt procedures. Our findings expand the understanding of the gut–liver–brain axis and identify a promising therapeutic and predictive target for HE.</p>","PeriodicalId":19037,"journal":{"name":"Nature Medicine","volume":"2 1","pages":""},"PeriodicalIF":82.9,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142936145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}