Pub Date : 2024-02-21DOI: 10.1016/S2589-7500(24)00029-3
{"title":"Correction to Lancet Digit Health 2024; 6: e153","authors":"","doi":"10.1016/S2589-7500(24)00029-3","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00029-3","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000293/pdfft?md5=09903ded0c28213ca818fa579de64511&pid=1-s2.0-S2589750024000293-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139935421","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 : 2024-02-21DOI: 10.1016/S2589-7500(23)00242-X
Michiel H J Maas MD , Prof Helmut Neumann MD PhD , Prof Haim Shirin MD , Prof Lior H Katz MD , Ariel A Benson MD , Arslan Kahloon MD , Elsa Soons MD PhD , Rawi Hazzan MD , Marc J Landsman MD , Benjamin Lebwohl MD , Suzanne K Lewis MD , Visvakanth Sivanathan MD , Saowanee Ngamruengphong MD , Harold Jacob MD , Prof Peter D Siersema MD PhD
Background
Studies on the effect of computer-aided detection (CAD) in a daily clinical screening and surveillance colonoscopy population practice are scarce. The aim of this study was to evaluate a novel CAD system in a screening and surveillance colonoscopy population.
Methods
This multicentre, randomised, controlled trial was done in ten hospitals in Europe, the USA, and Israel by 31 endoscopists. Patients referred for non-immunochemical faecal occult blood test (iFOBT) screening or surveillance colonoscopy were included. Patients were randomomly assigned to CAD-assisted colonoscopy or conventional colonoscopy; a subset was further randomly assigned to undergo tandem colonoscopy: CAD followed by conventional colonoscopy or conventional colonoscopy followed by CAD. Primary objectives included adenoma per colonoscopy (APC) and adenoma per extraction (APE). Secondary objectives included adenoma miss rate (AMR) in the tandem colonoscopies. The study was registered at ClinicalTrials.gov, NCT04640792.
Findings
A total of 916 patients were included in the modified intention-to-treat analysis: 449 in the CAD group and 467 in the conventional colonoscopy group. APC was higher with CAD compared with conventional colonoscopy (0·70 vs 0·51, p=0·015; 314 adenomas per 449 colonoscopies vs 238 adenomas per 467 colonoscopies; poisson effect ratio 1·372 [95% CI 1·068–1·769]), while showing non-inferiority of APE compared with conventional colonoscopy (0·59 vs 0·66; p<0·001 for non-inferiority; 314 of 536 extractions vs 238 of 360 extractions). AMR in the 127 (61 with CAD first, 66 with conventional colonoscopy first) patients completing tandem colonoscopy was 19% (11 of 59 detected during the second pass) in the CAD first group and 36% (16 of 45 detected during the second pass) in the conventional colonoscopy first group (p=0·024).
Interpretation
CAD increased adenoma detection in non-iFOBT screening and surveillance colonoscopies and reduced adenoma miss rates compared with conventional colonoscopy, without an increase in the resection of non-adenomatous lesions.
Funding
Magentiq Eye.
背景关于计算机辅助检测(CAD)在日常临床筛查和监视结肠镜检查中的应用效果的研究很少。这项研究的目的是评估新型 CAD 系统在筛查和监测结肠镜检查人群中的应用效果。方法这项多中心、随机对照试验是由 31 名内镜医师在欧洲、美国和以色列的 10 家医院进行的。研究对象包括接受非免疫化学粪便潜血试验(iFOBT)筛查或结肠镜监测的患者。患者被随机分配接受 CAD 辅助结肠镜检查或传统结肠镜检查;一部分患者被进一步随机分配接受串联结肠镜检查:子组进一步随机分配接受串联结肠镜检查:先接受 CAD 检查,再接受传统结肠镜检查,或先接受传统结肠镜检查,再接受 CAD 检查。主要目标包括每次结肠镜检查发现的腺瘤(APC)和每次摘除的腺瘤(APE)。次要目标包括串联结肠镜检查中的腺瘤漏检率(AMR)。该研究已在 ClinicalTrials.gov 注册,编号为 NCT04640792。研究结果共有 916 名患者被纳入修改后的意向治疗分析:CAD 组 449 人,传统结肠镜检查组 467 人。与传统结肠镜检查相比,CAD的APC更高(0-70 vs 0-51,p=0-015;每449次结肠镜检查发现314个腺瘤 vs 每467次结肠镜检查发现238个腺瘤;泊松效应比1-372 [95% CI 1-068-1-769]),而与传统结肠镜检查相比,APE显示出非劣性(0-59 vs 0-66;非劣性p<0-001;536次抽取中的314次 vs 360次抽取中的238次)。在完成串联结肠镜检查的 127 名患者中(61 名首先接受 CAD 检查,66 名首先接受传统结肠镜检查),首先接受 CAD 检查组的 AMR 为 19%(59 人中有 11 人在第二次检查中被发现),首先接受传统结肠镜检查组的 AMR 为 36%(45 人中有 16 人在第二次检查中被发现)(p=0-024)。与传统结肠镜检查相比,CAD提高了非iFOBT筛查和监测结肠镜检查的腺瘤检出率,降低了腺瘤漏检率,但非腺瘤病变的切除率并未增加。
{"title":"A computer-aided polyp detection system in screening and surveillance colonoscopy: an international, multicentre, randomised, tandem trial","authors":"Michiel H J Maas MD , Prof Helmut Neumann MD PhD , Prof Haim Shirin MD , Prof Lior H Katz MD , Ariel A Benson MD , Arslan Kahloon MD , Elsa Soons MD PhD , Rawi Hazzan MD , Marc J Landsman MD , Benjamin Lebwohl MD , Suzanne K Lewis MD , Visvakanth Sivanathan MD , Saowanee Ngamruengphong MD , Harold Jacob MD , Prof Peter D Siersema MD PhD","doi":"10.1016/S2589-7500(23)00242-X","DOIUrl":"10.1016/S2589-7500(23)00242-X","url":null,"abstract":"<div><h3>Background</h3><p>Studies on the effect of computer-aided detection (CAD) in a daily clinical screening and surveillance colonoscopy population practice are scarce. The aim of this study was to evaluate a novel CAD system in a screening and surveillance colonoscopy population.</p></div><div><h3>Methods</h3><p>This multicentre, randomised, controlled trial was done in ten hospitals in Europe, the USA, and Israel by 31 endoscopists. Patients referred for non-immunochemical faecal occult blood test (iFOBT) screening or surveillance colonoscopy were included. Patients were randomomly assigned to CAD-assisted colonoscopy or conventional colonoscopy; a subset was further randomly assigned to undergo tandem colonoscopy: CAD followed by conventional colonoscopy or conventional colonoscopy followed by CAD. Primary objectives included adenoma per colonoscopy (APC) and adenoma per extraction (APE). Secondary objectives included adenoma miss rate (AMR) in the tandem colonoscopies. The study was registered at <span>ClinicalTrials.gov</span><svg><path></path></svg>, <span>NCT04640792</span><svg><path></path></svg>.</p></div><div><h3>Findings</h3><p>A total of 916 patients were included in the modified intention-to-treat analysis: 449 in the CAD group and 467 in the conventional colonoscopy group. APC was higher with CAD compared with conventional colonoscopy (0·70 <em>vs</em> 0·51, p=0·015; 314 adenomas per 449 colonoscopies <em>vs</em> 238 adenomas per 467 colonoscopies; poisson effect ratio 1·372 [95% CI 1·068–1·769]), while showing non-inferiority of APE compared with conventional colonoscopy (0·59 <em>vs</em> 0·66; p<0·001 for non-inferiority; 314 of 536 extractions <em>vs</em> 238 of 360 extractions). AMR in the 127 (61 with CAD first, 66 with conventional colonoscopy first) patients completing tandem colonoscopy was 19% (11 of 59 detected during the second pass) in the CAD first group and 36% (16 of 45 detected during the second pass) in the conventional colonoscopy first group (p=0·024).</p></div><div><h3>Interpretation</h3><p>CAD increased adenoma detection in non-iFOBT screening and surveillance colonoscopies and reduced adenoma miss rates compared with conventional colonoscopy, without an increase in the resection of non-adenomatous lesions.</p></div><div><h3>Funding</h3><p>Magentiq Eye.</p></div>","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-02-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S258975002300242X/pdfft?md5=047493ed04e0aea400f66a1d0f300363&pid=1-s2.0-S258975002300242X-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139917829","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00227-3
Wei-Hung Weng MD PhD , Andrew Sellergen BS , Atilla P Kiraly PhD , Alexander D’Amour PhD , Jungyeon Park BA , Rory Pilgrim BE LLB , Stephen Pfohl PhD , Charles Lau MD , Vivek Natarajan MS , Shekoofeh Azizi PhD , Alan Karthikesalingam MD PhD , Heather Cole-Lewis PhD , Yossi Matias PhD , Greg S Corrado PhD , Dale R Webster PhD , Shravya Shetty MS , Shruthi Prabhakara PhD , Krish Eswaran PhD , Leo A G Celi MD MPH , Yun Liu PhD
Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components—GPPEs—from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.
{"title":"An intentional approach to managing bias in general purpose embedding models","authors":"Wei-Hung Weng MD PhD , Andrew Sellergen BS , Atilla P Kiraly PhD , Alexander D’Amour PhD , Jungyeon Park BA , Rory Pilgrim BE LLB , Stephen Pfohl PhD , Charles Lau MD , Vivek Natarajan MS , Shekoofeh Azizi PhD , Alan Karthikesalingam MD PhD , Heather Cole-Lewis PhD , Yossi Matias PhD , Greg S Corrado PhD , Dale R Webster PhD , Shravya Shetty MS , Shruthi Prabhakara PhD , Krish Eswaran PhD , Leo A G Celi MD MPH , Yun Liu PhD","doi":"10.1016/S2589-7500(23)00227-3","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00227-3","url":null,"abstract":"<div><p>Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components—GPPEs—from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.</p></div>","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002273/pdfft?md5=ad8033f384f4f510ef4a0b3ef9ad406f&pid=1-s2.0-S2589750023002273-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549643","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00226-1
Andrew A S Soltan MRCP , Anshul Thakur PhD , Jenny Yang MSc , Prof Anoop Chauhan FRCP , Leon G D’Cruz PhD , Phillip Dickson BSc , Marina A Soltan MRCP , Prof David R Thickett FRCP , Prof David W Eyre DPhil , Prof Tingting Zhu DPhil , Prof David A Clifton DPhil
Background
Multicentre training could reduce biases in medical artificial intelligence (AI); however, ethical, legal, and technical considerations can constrain the ability of hospitals to share data. Federated learning enables institutions to participate in algorithm development while retaining custody of their data but uptake in hospitals has been limited, possibly as deployment requires specialist software and technical expertise at each site. We previously developed an artificial intelligence-driven screening test for COVID-19 in emergency departments, known as CURIAL-Lab, which uses vital signs and blood tests that are routinely available within 1 h of a patient's arrival. Here we aimed to federate our COVID-19 screening test by developing an easy-to-use embedded system—which we introduce as full-stack federated learning—to train and evaluate machine learning models across four UK hospital groups without centralising patient data.
Methods
We supplied a Raspberry Pi 4 Model B preloaded with our federated learning software pipeline to four National Health Service (NHS) hospital groups in the UK: Oxford University Hospitals NHS Foundation Trust (OUH; through the locally linked research University, University of Oxford), University Hospitals Birmingham NHS Foundation Trust (UHB), Bedfordshire Hospitals NHS Foundation Trust (BH), and Portsmouth Hospitals University NHS Trust (PUH). OUH, PUH, and UHB participated in federated training, training a deep neural network and logistic regressor over 150 rounds to form and calibrate a global model to predict COVID-19 status, using clinical data from patients admitted before the pandemic (COVID-19-negative) and testing positive for COVID-19 during the first wave of the pandemic. We conducted a federated evaluation of the global model for admissions during the second wave of the pandemic at OUH, PUH, and externally at BH. For OUH and PUH, we additionally performed local fine-tuning of the global model using the sites’ individual training data, forming a site-tuned model, and evaluated the resultant model for admissions during the second wave of the pandemic. This study included data collected between Dec 1, 2018, and March 1, 2021; the exact date ranges used varied by site. The primary outcome was overall model performance, measured as the area under the receiver operating characteristic curve (AUROC). Removable micro secure digital (microSD) storage was destroyed on study completion.
Findings
Clinical data from 130 941 patients (1772 COVID-19-positive), routinely collected across three hospital groups (OUH, PUH, and UHB), were included in federated training. The evaluation step included data from 32 986 patients (3549 COVID-19-positive) attending OUH, PUH, or BH during the second wave of the pandemic. Federated training of a global deep neural network classifier improved upon performance of models trained locally in terms of AUROC by a mean
{"title":"A scalable federated learning solution for secondary care using low-cost microcomputing: privacy-preserving development and evaluation of a COVID-19 screening test in UK hospitals","authors":"Andrew A S Soltan MRCP , Anshul Thakur PhD , Jenny Yang MSc , Prof Anoop Chauhan FRCP , Leon G D’Cruz PhD , Phillip Dickson BSc , Marina A Soltan MRCP , Prof David R Thickett FRCP , Prof David W Eyre DPhil , Prof Tingting Zhu DPhil , Prof David A Clifton DPhil","doi":"10.1016/S2589-7500(23)00226-1","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00226-1","url":null,"abstract":"<div><h3>Background</h3><p>Multicentre training could reduce biases in medical artificial intelligence (AI); however, ethical, legal, and technical considerations can constrain the ability of hospitals to share data. Federated learning enables institutions to participate in algorithm development while retaining custody of their data but uptake in hospitals has been limited, possibly as deployment requires specialist software and technical expertise at each site. We previously developed an artificial intelligence-driven screening test for COVID-19 in emergency departments, known as CURIAL-Lab, which uses vital signs and blood tests that are routinely available within 1 h of a patient's arrival. Here we aimed to federate our COVID-19 screening test by developing an easy-to-use embedded system—which we introduce as full-stack federated learning—to train and evaluate machine learning models across four UK hospital groups without centralising patient data.</p></div><div><h3>Methods</h3><p>We supplied a Raspberry Pi 4 Model B preloaded with our federated learning software pipeline to four National Health Service (NHS) hospital groups in the UK: Oxford University Hospitals NHS Foundation Trust (OUH; through the locally linked research University, University of Oxford), University Hospitals Birmingham NHS Foundation Trust (UHB), Bedfordshire Hospitals NHS Foundation Trust (BH), and Portsmouth Hospitals University NHS Trust (PUH). OUH, PUH, and UHB participated in federated training, training a deep neural network and logistic regressor over 150 rounds to form and calibrate a global model to predict COVID-19 status, using clinical data from patients admitted before the pandemic (COVID-19-negative) and testing positive for COVID-19 during the first wave of the pandemic. We conducted a federated evaluation of the global model for admissions during the second wave of the pandemic at OUH, PUH, and externally at BH. For OUH and PUH, we additionally performed local fine-tuning of the global model using the sites’ individual training data, forming a site-tuned model, and evaluated the resultant model for admissions during the second wave of the pandemic. This study included data collected between Dec 1, 2018, and March 1, 2021; the exact date ranges used varied by site. The primary outcome was overall model performance, measured as the area under the receiver operating characteristic curve (AUROC). Removable micro secure digital (microSD) storage was destroyed on study completion.</p></div><div><h3>Findings</h3><p>Clinical data from 130 941 patients (1772 COVID-19-positive), routinely collected across three hospital groups (OUH, PUH, and UHB), were included in federated training. The evaluation step included data from 32 986 patients (3549 COVID-19-positive) attending OUH, PUH, or BH during the second wave of the pandemic. Federated training of a global deep neural network classifier improved upon performance of models trained locally in terms of AUROC by a mean","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002261/pdfft?md5=ec0102307b58e60cd24b5090ba16c75a&pid=1-s2.0-S2589750023002261-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139548931","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00266-2
Joshua D Kaggie
{"title":"Addressing machine learning challenges with microcomputing and federated learning","authors":"Joshua D Kaggie","doi":"10.1016/S2589-7500(23)00266-2","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00266-2","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002662/pdfft?md5=0947749a966095e17fd209095cd803d9&pid=1-s2.0-S2589750023002662-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549683","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 : 2024-01-24DOI: 10.1016/S2589-7500(24)00003-7
The Lancet Digital Health
{"title":"Taking responsibility for child safety online","authors":"The Lancet Digital Health","doi":"10.1016/S2589-7500(24)00003-7","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00003-7","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000037/pdfft?md5=9c7037dd6bb510136ebbba3b4b76fd51&pid=1-s2.0-S2589750024000037-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549682","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00241-8
Rebecca Giddings MFPH , Anabel Joseph BSc , Thomas Callender MFPH , Prof Sam M Janes PhD , Prof Mihaela van der Schaar PhD , Jessica Sheringham FFPH , Neal Navani PhD
Machine learning (ML)-based risk prediction models hold the potential to support the health-care setting in several ways; however, use of such models is scarce. We aimed to review health-care professional (HCP) and patient perceptions of ML risk prediction models in published literature, to inform future risk prediction model development. Following database and citation searches, we identified 41 articles suitable for inclusion. Article quality varied with qualitative studies performing strongest. Overall, perceptions of ML risk prediction models were positive. HCPs and patients considered that models have the potential to add benefit in the health-care setting. However, reservations remain; for example, concerns regarding data quality for model development and fears of unintended consequences following ML model use. We identified that public views regarding these models might be more negative than HCPs and that concerns (eg, extra demands on workload) were not always borne out in practice. Conclusions are tempered by the low number of patient and public studies, the absence of participant ethnic diversity, and variation in article quality. We identified gaps in knowledge (particularly views from under-represented groups) and optimum methods for model explanation and alerts, which require future research.
基于机器学习(ML)的风险预测模型有可能在多个方面为医疗机构提供支持;然而,此类模型的使用却很少。我们旨在回顾已发表文献中医疗保健专业人员(HCP)和患者对 ML 风险预测模型的看法,为未来风险预测模型的开发提供参考。经过数据库和引文检索,我们确定了 41 篇适合纳入的文章。文章质量参差不齐,其中定性研究的效果最好。总体而言,人们对 ML 风险预测模型的看法是积极的。医疗保健人员和患者认为模型有可能为医疗保健环境带来更多益处。但仍有一些保留意见,例如,对模型开发数据质量的担忧,以及对使用 ML 模型后意外后果的恐惧。我们发现,公众对这些模型的看法可能比医护人员更消极,而且所担心的问题(如对工作量的额外要求)并不总是在实践中得到证实。由于患者和公众研究的数量较少、缺乏参与者的种族多样性以及文章质量的差异,我们得出的结论并不全面。我们发现了知识方面的差距(尤其是代表性不足群体的观点)以及模型解释和警报的最佳方法,这些都需要未来的研究。
{"title":"Factors influencing clinician and patient interaction with machine learning-based risk prediction models: a systematic review","authors":"Rebecca Giddings MFPH , Anabel Joseph BSc , Thomas Callender MFPH , Prof Sam M Janes PhD , Prof Mihaela van der Schaar PhD , Jessica Sheringham FFPH , Neal Navani PhD","doi":"10.1016/S2589-7500(23)00241-8","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00241-8","url":null,"abstract":"<div><p>Machine learning (ML)-based risk prediction models hold the potential to support the health-care setting in several ways; however, use of such models is scarce. We aimed to review health-care professional (HCP) and patient perceptions of ML risk prediction models in published literature, to inform future risk prediction model development. Following database and citation searches, we identified 41 articles suitable for inclusion. Article quality varied with qualitative studies performing strongest. Overall, perceptions of ML risk prediction models were positive. HCPs and patients considered that models have the potential to add benefit in the health-care setting. However, reservations remain; for example, concerns regarding data quality for model development and fears of unintended consequences following ML model use. We identified that public views regarding these models might be more negative than HCPs and that concerns (eg, extra demands on workload) were not always borne out in practice. Conclusions are tempered by the low number of patient and public studies, the absence of participant ethnic diversity, and variation in article quality. We identified gaps in knowledge (particularly views from under-represented groups) and optimum methods for model explanation and alerts, which require future research.</p></div>","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002418/pdfft?md5=f3d2d6147b16146f6b3fd719afcb42f0&pid=1-s2.0-S2589750023002418-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549644","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00228-5
Torsten Dahlén MD PhD , Jingcheng Zhao MD PhD , Prof Michael P Busch MD PhD , Gustaf Edgren MD PhD
Background
Identification and prevention of transfusion-transmitted disease is essential for blood transfusion safety. However, current surveillance systems are largely driven by reports of sentinel events, which is an approach that might be inadequate for identifying transmission of pathogens not known to be transmissible or pathogens with long incubation periods. Using a combination of health-data registers and blood-bank databases, we aimed to perform an agnostic search for potential transfusion-transmitted diseases and to identify unknown threats to the blood supply.
Methods
In this nationwide, agnostic retrospective cohort study, we developed a systematic algorithm for performing a phenome-wide search for transfusion-transmitted disease without consideration of any a-priori suspicion of blood-borne transmissibility. We applied this algorithm to a nationwide Swedish transfusion database (SCANDAT-3S) to test for possible transmission of 1155 disease entities based on all relevant diagnostic coding systems in use during the period. We ascertained health outcomes of blood donors and transfusion recipients from the Swedish National Inpatient Register, Swedish Cause of Death Register, and Swedish Cancer Register. Analyses were two-pronged, studying both disease diagnosis concordance between donors and recipients and a possible shared increased disease risk among all recipients of a given donor. For both approaches, we used Cox proportional hazards regression models with time-dependent covariates. Adjustment for multiple comparisons was done using a false discovery rate method.
Findings
The analyses included data on 1·72 million patients who had received 18·97 million transfusions (red blood cell, plasma, platelet, or whole blood units) between Jan 1, 1968, and Dec 31, 2017, from 1·04 million blood donors. The median follow-up was 4·5 (IQR 0·9–11·4) years for recipients and 18·5 (8·3–26·2) years for donors. We found evidence of transfusion-transmission for 15 diseases, of which 13 were validated using a second conceptually different approach. We identified transmission of viral hepatitis and its complications (eg, oesophageal varices) but also transmission of other conditions (eg, pneumonia of unknown origin). The diseases that could not be validated in this second approach, HIV and abnormal findings in specimens from male genital organs, were not statistically significant after adjustment for multiple testing. The effect sizes were small (close to 1) for other conditions.
Interpretation
We find no strong evidence of unexpected, widespread transfusion-transmitted disease. This novel approach serves as a proof-of-concept for agnostic, data-driven surveillance for transfusion-transmitted disease using routinely collected blood-bank and health-care data.
Funding
Department of Health and Human Services, US National Heart, Lung, and Blood
{"title":"Using routine health-care data to search for unknown transfusion-transmitted disease: a nationwide, agnostic retrospective cohort study","authors":"Torsten Dahlén MD PhD , Jingcheng Zhao MD PhD , Prof Michael P Busch MD PhD , Gustaf Edgren MD PhD","doi":"10.1016/S2589-7500(23)00228-5","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00228-5","url":null,"abstract":"<div><h3>Background</h3><p>Identification and prevention of transfusion-transmitted disease is essential for blood transfusion safety. However, current surveillance systems are largely driven by reports of sentinel events, which is an approach that might be inadequate for identifying transmission of pathogens not known to be transmissible or pathogens with long incubation periods. Using a combination of health-data registers and blood-bank databases, we aimed to perform an agnostic search for potential transfusion-transmitted diseases and to identify unknown threats to the blood supply.</p></div><div><h3>Methods</h3><p>In this nationwide, agnostic retrospective cohort study, we developed a systematic algorithm for performing a phenome-wide search for transfusion-transmitted disease without consideration of any a-priori suspicion of blood-borne transmissibility. We applied this algorithm to a nationwide Swedish transfusion database (SCANDAT-3S) to test for possible transmission of 1155 disease entities based on all relevant diagnostic coding systems in use during the period. We ascertained health outcomes of blood donors and transfusion recipients from the Swedish National Inpatient Register, Swedish Cause of Death Register, and Swedish Cancer Register. Analyses were two-pronged, studying both disease diagnosis concordance between donors and recipients and a possible shared increased disease risk among all recipients of a given donor. For both approaches, we used Cox proportional hazards regression models with time-dependent covariates. Adjustment for multiple comparisons was done using a false discovery rate method.</p></div><div><h3>Findings</h3><p>The analyses included data on 1·72 million patients who had received 18·97 million transfusions (red blood cell, plasma, platelet, or whole blood units) between Jan 1, 1968, and Dec 31, 2017, from 1·04 million blood donors. The median follow-up was 4·5 (IQR 0·9–11·4) years for recipients and 18·5 (8·3–26·2) years for donors. We found evidence of transfusion-transmission for 15 diseases, of which 13 were validated using a second conceptually different approach. We identified transmission of viral hepatitis and its complications (eg, oesophageal varices) but also transmission of other conditions (eg, pneumonia of unknown origin). The diseases that could not be validated in this second approach, HIV and abnormal findings in specimens from male genital organs, were not statistically significant after adjustment for multiple testing. The effect sizes were small (close to 1) for other conditions.</p></div><div><h3>Interpretation</h3><p>We find no strong evidence of unexpected, widespread transfusion-transmitted disease. This novel approach serves as a proof-of-concept for agnostic, data-driven surveillance for transfusion-transmitted disease using routinely collected blood-bank and health-care data.</p></div><div><h3>Funding</h3><p>Department of Health and Human Services, US National Heart, Lung, and Blood ","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002285/pdfft?md5=c332bc81a4376a3b54037caa6f5b96e7&pid=1-s2.0-S2589750023002285-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549642","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 : 2024-01-24DOI: 10.1016/S2589-7500(23)00251-0
Vishnu Khanal , Timothy Shaw , Elaine Wills , John Wakerman , Deborah J Russell
{"title":"Co-design of digital health technologies in Australian First Nations communities","authors":"Vishnu Khanal , Timothy Shaw , Elaine Wills , John Wakerman , Deborah J Russell","doi":"10.1016/S2589-7500(23)00251-0","DOIUrl":"https://doi.org/10.1016/S2589-7500(23)00251-0","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002510/pdfft?md5=b1a55e589d2badc8f45d5921c5b231b7&pid=1-s2.0-S2589750023002510-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139549684","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 : 2024-01-22DOI: 10.1016/S2589-7500(23)00253-4
Bo M van der Weijden , Sanne W C M Janssen , William E Benitz , Michael W Kuzniewicz , Karen M Puopolo , Frans B Plötz , Niek B Achten
{"title":"Open-source code to extend early-onset sepsis calculator accessibility","authors":"Bo M van der Weijden , Sanne W C M Janssen , William E Benitz , Michael W Kuzniewicz , Karen M Puopolo , Frans B Plötz , Niek B Achten","doi":"10.1016/S2589-7500(23)00253-4","DOIUrl":"10.1016/S2589-7500(23)00253-4","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750023002534/pdfft?md5=7e972f009afd72341d3df0eb85928954&pid=1-s2.0-S2589750023002534-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139543384","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}