Pub Date : 2026-01-06DOI: 10.1038/s44220-025-00560-x
The MULTI Consortium, Aleix Boquet-Pujadas, Filippos Anagnostakis, Zhijian Yang, Ye Ella Tian, Michael R. Duggan, Guray Erus, Dhivya Srinivasan, Cassandra M. Joynes, Wenjia Bai, Praveen J. Patel, Keenan A. Walker, Andrew Zalesky, Christos Davatzikos, Junhao Wen
Disease heterogeneity and commonality pose critical challenges to precision medicine, as traditional approaches frequently focus on single disease entities and overlook shared mechanisms across conditions. Here, inspired by pan-cancer and multi-organ research, we introduce the concept of ‘pan-disease’ to investigate the heterogeneity and shared etiology in brain, eye and heart diseases. Leveraging individual-level data from 129,340 participants and summary-level data, curated from the MULTI consortium, we applied a weakly supervised deep learning model (Surreal-GAN) to multi-organ imaging, genetic and proteomic data, identifying 11 artificial intelligence (AI)-derived biomarkers, called multi-organ AI endophenotypes, for the brain (Brain 1–6), eye (Eye 1–3) and heart (Heart 1–2). We found Brain 3 to be a risk factor for Alzheimer’s disease progression and mortality, whereas Brain 5 was protective against Alzheimer’s disease progression. In data from an anti-amyloid Alzheimer’s disease drug (solanezumab), heterogeneity in cognitive decline trajectories was observed across treatment groups. At week 240, patients with lower Brain 1–3 expression had slower cognitive decline, whereas patients with higher expression had faster cognitive decline. A multilayer causal pathway pinpointed Brain 1 as a mediational endophenotype linking the FLRT2 protein to migraine, exemplifying new therapeutic targets and pathways. In addition, genes associated with Eye 1 and Eye 3 were enriched in cancer drug-related gene sets with causal links to specific cancer types and proteins. Finally, Heart 1 and Heart 2 had the highest mortality risk and unique medication history profiles, with Heart 1 showing favorable responses to antihypertensive medications and Heart 2 to digoxin treatment. The 11 multi-organ AI endophenotypes provide new AI dimensional representations for precision medicine and highlight the potential of AI-driven patient stratification for disease risk monitoring, clinical trials and drug discovery. Disease heterogeneity complicates precision medicine, which focuses on single conditions and ignores shared mechanisms. Here the authors introduce ‘pan-disease’ analysis using a deep learning model on multi-organ data, identifying 11 AI-derived biomarkers that reveal new therapeutic targets and pathways, enhancing patient stratification for disease risk monitoring and drug discovery.
{"title":"Multi-organ AI endophenotypes chart the heterogeneity of brain, eye and heart pan-disease","authors":"The MULTI Consortium, Aleix Boquet-Pujadas, Filippos Anagnostakis, Zhijian Yang, Ye Ella Tian, Michael R. Duggan, Guray Erus, Dhivya Srinivasan, Cassandra M. Joynes, Wenjia Bai, Praveen J. Patel, Keenan A. Walker, Andrew Zalesky, Christos Davatzikos, Junhao Wen","doi":"10.1038/s44220-025-00560-x","DOIUrl":"10.1038/s44220-025-00560-x","url":null,"abstract":"Disease heterogeneity and commonality pose critical challenges to precision medicine, as traditional approaches frequently focus on single disease entities and overlook shared mechanisms across conditions. Here, inspired by pan-cancer and multi-organ research, we introduce the concept of ‘pan-disease’ to investigate the heterogeneity and shared etiology in brain, eye and heart diseases. Leveraging individual-level data from 129,340 participants and summary-level data, curated from the MULTI consortium, we applied a weakly supervised deep learning model (Surreal-GAN) to multi-organ imaging, genetic and proteomic data, identifying 11 artificial intelligence (AI)-derived biomarkers, called multi-organ AI endophenotypes, for the brain (Brain 1–6), eye (Eye 1–3) and heart (Heart 1–2). We found Brain 3 to be a risk factor for Alzheimer’s disease progression and mortality, whereas Brain 5 was protective against Alzheimer’s disease progression. In data from an anti-amyloid Alzheimer’s disease drug (solanezumab), heterogeneity in cognitive decline trajectories was observed across treatment groups. At week 240, patients with lower Brain 1–3 expression had slower cognitive decline, whereas patients with higher expression had faster cognitive decline. A multilayer causal pathway pinpointed Brain 1 as a mediational endophenotype linking the FLRT2 protein to migraine, exemplifying new therapeutic targets and pathways. In addition, genes associated with Eye 1 and Eye 3 were enriched in cancer drug-related gene sets with causal links to specific cancer types and proteins. Finally, Heart 1 and Heart 2 had the highest mortality risk and unique medication history profiles, with Heart 1 showing favorable responses to antihypertensive medications and Heart 2 to digoxin treatment. The 11 multi-organ AI endophenotypes provide new AI dimensional representations for precision medicine and highlight the potential of AI-driven patient stratification for disease risk monitoring, clinical trials and drug discovery. Disease heterogeneity complicates precision medicine, which focuses on single conditions and ignores shared mechanisms. Here the authors introduce ‘pan-disease’ analysis using a deep learning model on multi-organ data, identifying 11 AI-derived biomarkers that reveal new therapeutic targets and pathways, enhancing patient stratification for disease risk monitoring and drug discovery.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 2","pages":"203-230"},"PeriodicalIF":8.7,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s44220-025-00560-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146148322","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-05DOI: 10.1038/s44220-025-00557-6
Brianna L. Gonzalez, Patrick Amoateng, Nana Kwadwo Obiri, Turhan Canli
Collaborations between neuroscientists and traditional medical practitioners can strengthen the scientific foundations of traditional medicine and enrich neuroscience with culturally grounded insights. Such partnerships, built on mutual learning, can promote more equitable and context-sensitive mental health research.
{"title":"Building evidence-based knowledge in traditional medicine provides an opportunity for neuroscientists and traditional medical practitioners","authors":"Brianna L. Gonzalez, Patrick Amoateng, Nana Kwadwo Obiri, Turhan Canli","doi":"10.1038/s44220-025-00557-6","DOIUrl":"10.1038/s44220-025-00557-6","url":null,"abstract":"Collaborations between neuroscientists and traditional medical practitioners can strengthen the scientific foundations of traditional medicine and enrich neuroscience with culturally grounded insights. Such partnerships, built on mutual learning, can promote more equitable and context-sensitive mental health research.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"3-5"},"PeriodicalIF":8.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931247","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}
Pub Date : 2026-01-05DOI: 10.1038/s44220-025-00565-6
Briana S. Last, Gabriela Kattan Khazanov
Spurred by billions of dollars in public and private investments, artificial intelligence (AI) technologies are being rapidly developed and deployed to automate, supplement and even replace the role of skilled behavioral health providers. Most discussions of AI in behavioral healthcare have focused on the safety and efficacy of these technologies and have largely neglected more fundamental questions about who decides whether and how AI should be used in behavioral healthcare. We argue that, despite substantial public investments in AI and the significant impacts these technologies will have on the lives of behavioral health service users, the public and providers, the private sector—not these key stakeholders—has played an outsized role in shaping the future of AI in behavioral healthcare. We offer recommendations to democratize the development and deployment of AI technologies in behavioral healthcare by prioritizing the needs and interests of behavioral health service users, the public and providers. In this Perspective, Last and Khazanov call for democratizing AI in behavioral healthcare, urging that service users, providers and the public—not private interests—shape its development and deployment.
{"title":"Empowering service users, the public, and providers to determine the future of artificial intelligence in behavioral healthcare","authors":"Briana S. Last, Gabriela Kattan Khazanov","doi":"10.1038/s44220-025-00565-6","DOIUrl":"10.1038/s44220-025-00565-6","url":null,"abstract":"Spurred by billions of dollars in public and private investments, artificial intelligence (AI) technologies are being rapidly developed and deployed to automate, supplement and even replace the role of skilled behavioral health providers. Most discussions of AI in behavioral healthcare have focused on the safety and efficacy of these technologies and have largely neglected more fundamental questions about who decides whether and how AI should be used in behavioral healthcare. We argue that, despite substantial public investments in AI and the significant impacts these technologies will have on the lives of behavioral health service users, the public and providers, the private sector—not these key stakeholders—has played an outsized role in shaping the future of AI in behavioral healthcare. We offer recommendations to democratize the development and deployment of AI technologies in behavioral healthcare by prioritizing the needs and interests of behavioral health service users, the public and providers. In this Perspective, Last and Khazanov call for democratizing AI in behavioral healthcare, urging that service users, providers and the public—not private interests—shape its development and deployment.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"52-59"},"PeriodicalIF":8.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931252","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}
Pub Date : 2026-01-05DOI: 10.1038/s44220-025-00561-w
Adam Benzekri, Marco Thimm-Kaiser, Francis Kwadwo Amankwah, Vincent Guilamo-Ramos
A behavioral healthcare workforce — concordant in race, ethnicity, lived experience, language, and geography with the populations it serves — is urgently needed to end the US behavioral health crisis.
{"title":"The need for a representative workforce to address the US behavioral health crisis","authors":"Adam Benzekri, Marco Thimm-Kaiser, Francis Kwadwo Amankwah, Vincent Guilamo-Ramos","doi":"10.1038/s44220-025-00561-w","DOIUrl":"10.1038/s44220-025-00561-w","url":null,"abstract":"A behavioral healthcare workforce — concordant in race, ethnicity, lived experience, language, and geography with the populations it serves — is urgently needed to end the US behavioral health crisis.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"6-8"},"PeriodicalIF":8.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931213","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}
Pub Date : 2026-01-05DOI: 10.1038/s44220-025-00564-7
Stephen V. Faraone, Jeffrey H. Newcorn
Stimulant medications are the first-line treatment for ADHD, with non-stimulants often used if stimulants are ineffective. Here, by reinterpreting randomized controlled trials, addressing heterogeneity of treatment effects, and considering societal impact, we argue for equal consideration of stimulant and non-stimulants as first-line treatment options.
{"title":"Rethinking the role of non-stimulants in ADHD treatment","authors":"Stephen V. Faraone, Jeffrey H. Newcorn","doi":"10.1038/s44220-025-00564-7","DOIUrl":"10.1038/s44220-025-00564-7","url":null,"abstract":"Stimulant medications are the first-line treatment for ADHD, with non-stimulants often used if stimulants are ineffective. Here, by reinterpreting randomized controlled trials, addressing heterogeneity of treatment effects, and considering societal impact, we argue for equal consideration of stimulant and non-stimulants as first-line treatment options.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"9-12"},"PeriodicalIF":8.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931250","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}
Pub Date : 2026-01-05DOI: 10.1038/s44220-025-00555-8
Yunfei Luo, Iman Deznabi, Bhanu Teja Gullapalli, Mark Tuomenoksa, Madalina Brostean Fiterau, Eric L. Garland, Tauhidur Rahman
Fluctuations in pain, stress and craving are thought to contribute to opioid misuse. Developing accurate prediction models is vital for intervention and prevention efforts. In this work, we leverage physiological data and semantic analysis of electronic health records to tackle the challenge of detecting opioid misuse. Utilizing personalized hierarchical deep-learning models, we analyze trajectories of predicted pain, stress and craving states with 10,140 hours of heart-rate data collected by wearables from patients on long-term opioid therapy. From these trajectories, we extract entropy features from nonlinear dynamical analysis and develop a novel relevance-based temporal fusion model of opioid misuse risk. We incorporate clinical data into a large language model to enhance opioid misuse risk detection. We then fuse these modalities to achieve an accurate opioid misuse risk assessment with area under the precision-recall curve of 0.94 ± 0.05. This study marks a substantial advancement in personalized prediction of addictive behavior by elucidating the entropic nature of underlying affective state dynamics. This study addresses opioid misuse prediction by integrating physiological data and electronic health records. Utilizing personalized deep-learning models, it achieves a high accuracy in risk assessment through entropy feature extraction and relevance-based temporal fusion, demonstrating effective intervention potential.
{"title":"Personalized entropy-informed deep learning for identifying opioid misuse","authors":"Yunfei Luo, Iman Deznabi, Bhanu Teja Gullapalli, Mark Tuomenoksa, Madalina Brostean Fiterau, Eric L. Garland, Tauhidur Rahman","doi":"10.1038/s44220-025-00555-8","DOIUrl":"10.1038/s44220-025-00555-8","url":null,"abstract":"Fluctuations in pain, stress and craving are thought to contribute to opioid misuse. Developing accurate prediction models is vital for intervention and prevention efforts. In this work, we leverage physiological data and semantic analysis of electronic health records to tackle the challenge of detecting opioid misuse. Utilizing personalized hierarchical deep-learning models, we analyze trajectories of predicted pain, stress and craving states with 10,140 hours of heart-rate data collected by wearables from patients on long-term opioid therapy. From these trajectories, we extract entropy features from nonlinear dynamical analysis and develop a novel relevance-based temporal fusion model of opioid misuse risk. We incorporate clinical data into a large language model to enhance opioid misuse risk detection. We then fuse these modalities to achieve an accurate opioid misuse risk assessment with area under the precision-recall curve of 0.94 ± 0.05. This study marks a substantial advancement in personalized prediction of addictive behavior by elucidating the entropic nature of underlying affective state dynamics. This study addresses opioid misuse prediction by integrating physiological data and electronic health records. Utilizing personalized deep-learning models, it achieves a high accuracy in risk assessment through entropy feature extraction and relevance-based temporal fusion, demonstrating effective intervention potential.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"112-124"},"PeriodicalIF":8.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931211","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}
Pub Date : 2026-01-02DOI: 10.1038/s44220-025-00558-5
Isabella Goodwin, Kelly M. J. Diederen, Emily J. Hird, Veith Weilnhammer, Marta I. Garrido, Franziska Knolle
Predictive processing has revolutionized cognitive neuroscience, offering a comprehensive computational framework for understanding normative behavior and psychiatric illness. This narrative Review evaluates the role of predictive processing in understanding psychosis, revisiting the seminal work of Sterzer and colleagues. It consolidates recent experimental evidence on the alteration of priors and sensory likelihoods across different stages of psychosis in an attempt to reconcile top-down (that is, overly precise priors/noisy sensations) and bottom-up (that is, noisy priors/overly precise sensations) accounts. It evaluates predictive processing alterations across the continuum of psychosis, from non-clinical psychotic experiences to high-risk and first-episode psychosis to schizophrenia, exploring the explanatory potential of predictive processing as a transdiagnostic framework. We discuss the translational potential of predictive processing, including its use as a biomarker and in therapeutic interventions. We emphasize the need for standardized paradigms and longitudinal studies to advance predictive processing theories in clinical practice. By offering a unified theoretical perspective, this Review aims to inspire further research into the neuro-computational mechanisms underlying psychosis and enhance our understanding of psychiatric disorders. In this Review the authors integrate the latest evidence on predictive processing alterations across the continuum of psychosis and discuss its potential applications as a biomarker and in therapeutic interventions.
{"title":"Predictive processing accounts of psychosis: bottom-up or top-down disruptions","authors":"Isabella Goodwin, Kelly M. J. Diederen, Emily J. Hird, Veith Weilnhammer, Marta I. Garrido, Franziska Knolle","doi":"10.1038/s44220-025-00558-5","DOIUrl":"10.1038/s44220-025-00558-5","url":null,"abstract":"Predictive processing has revolutionized cognitive neuroscience, offering a comprehensive computational framework for understanding normative behavior and psychiatric illness. This narrative Review evaluates the role of predictive processing in understanding psychosis, revisiting the seminal work of Sterzer and colleagues. It consolidates recent experimental evidence on the alteration of priors and sensory likelihoods across different stages of psychosis in an attempt to reconcile top-down (that is, overly precise priors/noisy sensations) and bottom-up (that is, noisy priors/overly precise sensations) accounts. It evaluates predictive processing alterations across the continuum of psychosis, from non-clinical psychotic experiences to high-risk and first-episode psychosis to schizophrenia, exploring the explanatory potential of predictive processing as a transdiagnostic framework. We discuss the translational potential of predictive processing, including its use as a biomarker and in therapeutic interventions. We emphasize the need for standardized paradigms and longitudinal studies to advance predictive processing theories in clinical practice. By offering a unified theoretical perspective, this Review aims to inspire further research into the neuro-computational mechanisms underlying psychosis and enhance our understanding of psychiatric disorders. In this Review the authors integrate the latest evidence on predictive processing alterations across the continuum of psychosis and discuss its potential applications as a biomarker and in therapeutic interventions.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"60-84"},"PeriodicalIF":8.7,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931251","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}
Pub Date : 2026-01-02DOI: 10.1038/s44220-025-00563-8
Yinxian Chen, Qingyue Yuan, Lina Dimitrov, Benjamin Risk, Benson Ku, Anke Hüls
The genetic risk of persistent distressing psychotic-like experiences (PLE) in the multiancestral population is underinvestigated. The gene–neighborhood environment interaction in persistent distressing PLE is also unknown. This study included 6,449 participants from the Adolescent Brain and Cognitive Development Study. The genetic risk was measured by a multiancestral schizophrenia polygenic risk score (SCZ-PRS). The multidimensional neighborhood-level exposures were used to form the neighborhood exposome (NE). SCZ-PRS was not statistically significantly associated with odds of persistent distressing PLE (odds ratio (OR) of 1.04, 95% confidence intervals (CI) 0.97 to 1.13, P = 0.280), whereas the NE score was (OR of 1.15, 95% CI 1.05 to 1.26, P = 0.003). A significant negative multiplicative interaction between SCZ-PRS and NE was found (estimate of −0.08, 95% CI −0.15 to −0.00, P = 0.039). The additive interaction followed the same direction but was statistically insignificant (estimate of −0.06, 95% CI −0.15 to 0.03, P = 0.189). Persistent distressing PLE in children may be driven by detrimental neighborhood exposures in multiancestral populations, particularly among those with low genetic risk. Here the findings provide important evidence on persistent distressing PLE etiology attributed to genetic and environmental risks and identify potential susceptible populations for targeted interventions. Chen et al. examined how genetic risk interacts with neighborhood environmental exposures to influence psychotic-like experiences in children from the ABCD cohort study.
{"title":"Interaction between neighborhood exposome and genetic risk in persistent distressing psychotic-like experiences in children","authors":"Yinxian Chen, Qingyue Yuan, Lina Dimitrov, Benjamin Risk, Benson Ku, Anke Hüls","doi":"10.1038/s44220-025-00563-8","DOIUrl":"10.1038/s44220-025-00563-8","url":null,"abstract":"The genetic risk of persistent distressing psychotic-like experiences (PLE) in the multiancestral population is underinvestigated. The gene–neighborhood environment interaction in persistent distressing PLE is also unknown. This study included 6,449 participants from the Adolescent Brain and Cognitive Development Study. The genetic risk was measured by a multiancestral schizophrenia polygenic risk score (SCZ-PRS). The multidimensional neighborhood-level exposures were used to form the neighborhood exposome (NE). SCZ-PRS was not statistically significantly associated with odds of persistent distressing PLE (odds ratio (OR) of 1.04, 95% confidence intervals (CI) 0.97 to 1.13, P = 0.280), whereas the NE score was (OR of 1.15, 95% CI 1.05 to 1.26, P = 0.003). A significant negative multiplicative interaction between SCZ-PRS and NE was found (estimate of −0.08, 95% CI −0.15 to −0.00, P = 0.039). The additive interaction followed the same direction but was statistically insignificant (estimate of −0.06, 95% CI −0.15 to 0.03, P = 0.189). Persistent distressing PLE in children may be driven by detrimental neighborhood exposures in multiancestral populations, particularly among those with low genetic risk. Here the findings provide important evidence on persistent distressing PLE etiology attributed to genetic and environmental risks and identify potential susceptible populations for targeted interventions. Chen et al. examined how genetic risk interacts with neighborhood environmental exposures to influence psychotic-like experiences in children from the ABCD cohort study.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"136-145"},"PeriodicalIF":8.7,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931246","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}
Pub Date : 2025-12-31DOI: 10.1038/s44220-025-00559-4
Claire Houtsma, Chris J. Kennedy, Howard Liu, Emily R. Edwards, Nancy A. Sampson, Joe C. Geraci, Brian P. Marx, Matthew K. Nock, James Wagner, Murray B. Stein, Robert J. Ursano, Ronald C. Kessler
US veterans are significantly more likely than civilians to die by suicide. Machine-learning models have been developed to target high-risk transitioning service members for suicide prevention interventions to reduce veteran suicides. These models are suicide method-agnostic. However, firearms are involved in most veteran suicides, and firearm-specific preventions exist. We used data from US Army veterans from 2010 to 2019 (N = 800,579) to develop and compare firearm-specific machine-learning models with a method-agnostic model to predict firearm suicides among transitioning Army veterans up to 10 years after discharge. The models performed comparably overall (area under the receiver operating characteristic curve = 0.710–0.708; integrated calibration index = 0.0003–0.0005% for firearm-specific and method-agnostic models, respectively), with the best model depending on the intervention threshold. Results from this study show the method-agnostic model was better at predicting firearm suicides at the highest intervention threshold, whereas the firearm-specific model was better at lower thresholds. When considering fairness with respect to sex and race/ethnicity, the firearm-specific model was best across all thresholds. Thus, model choice depends on weighing numerous factors, and optimal thresholds might differ for coordinated firearm-specific and method-agnostic interventions. This research developed and compared firearm-specific and method-agnostic machine-learning models using data from 800,579 Army veterans, revealing that model choice and intervention thresholds impact predictive accuracy and fairness, guiding tailored suicide prevention efforts.
{"title":"Predicting firearm suicide among US Army veterans transitioning from active service","authors":"Claire Houtsma, Chris J. Kennedy, Howard Liu, Emily R. Edwards, Nancy A. Sampson, Joe C. Geraci, Brian P. Marx, Matthew K. Nock, James Wagner, Murray B. Stein, Robert J. Ursano, Ronald C. Kessler","doi":"10.1038/s44220-025-00559-4","DOIUrl":"10.1038/s44220-025-00559-4","url":null,"abstract":"US veterans are significantly more likely than civilians to die by suicide. Machine-learning models have been developed to target high-risk transitioning service members for suicide prevention interventions to reduce veteran suicides. These models are suicide method-agnostic. However, firearms are involved in most veteran suicides, and firearm-specific preventions exist. We used data from US Army veterans from 2010 to 2019 (N = 800,579) to develop and compare firearm-specific machine-learning models with a method-agnostic model to predict firearm suicides among transitioning Army veterans up to 10 years after discharge. The models performed comparably overall (area under the receiver operating characteristic curve = 0.710–0.708; integrated calibration index = 0.0003–0.0005% for firearm-specific and method-agnostic models, respectively), with the best model depending on the intervention threshold. Results from this study show the method-agnostic model was better at predicting firearm suicides at the highest intervention threshold, whereas the firearm-specific model was better at lower thresholds. When considering fairness with respect to sex and race/ethnicity, the firearm-specific model was best across all thresholds. Thus, model choice depends on weighing numerous factors, and optimal thresholds might differ for coordinated firearm-specific and method-agnostic interventions. This research developed and compared firearm-specific and method-agnostic machine-learning models using data from 800,579 Army veterans, revealing that model choice and intervention thresholds impact predictive accuracy and fairness, guiding tailored suicide prevention efforts.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"125-135"},"PeriodicalIF":8.7,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931212","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}
Pub Date : 2025-12-22DOI: 10.1038/s44220-025-00552-x
Evidence from national medical records of over 8 million people in the Netherlands shows that autism is associated with increased risk of cardiometabolic conditions. These associations emerged in adolescents and young adults, suggesting earlier onset of such conditions in individuals with autism than in individuals without it.
{"title":"People with autism are at increased risk of cardiometabolic conditions","authors":"","doi":"10.1038/s44220-025-00552-x","DOIUrl":"10.1038/s44220-025-00552-x","url":null,"abstract":"Evidence from national medical records of over 8 million people in the Netherlands shows that autism is associated with increased risk of cardiometabolic conditions. These associations emerged in adolescents and young adults, suggesting earlier onset of such conditions in individuals with autism than in individuals without it.","PeriodicalId":74247,"journal":{"name":"Nature mental health","volume":"4 1","pages":"13-14"},"PeriodicalIF":8.7,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145931249","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：