Pub Date : 2024-06-19DOI: 10.1016/S2589-7500(24)00086-4
Adrienne K Scott , Michelle L Oyen
{"title":"Virtual pregnancies: predicting and preventing pregnancy complications with digital twins","authors":"Adrienne K Scott , Michelle L Oyen","doi":"10.1016/S2589-7500(24)00086-4","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00086-4","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000864/pdfft?md5=4c537f02b6a0b5de364d1829924f9aa5&pid=1-s2.0-S2589750024000864-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141429102","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00121-3
{"title":"Correction to Lancet Digit Health 2024; 6: e33–43","authors":"","doi":"10.1016/S2589-7500(24)00121-3","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00121-3","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024001213/pdfft?md5=73e788169a1247c34b3f5649a47fbc2d&pid=1-s2.0-S2589750024001213-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428955","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00094-3
Karen K Wong MD , Thaddeus Segura MIDS , Gunnar Mein MIDS , Jia Lu PhD , Elizabeth J Hannapel MPH , Jasen M Kunz MPH , Troy Ritter PhD , Jessica C Smith MPH , Alberto Todeschini PhD , Fred Nugen PhD , Chris Edens PhD
Background
Cooling towers containing Legionella spp are a high-risk source of Legionnaires’ disease outbreaks. Manually locating cooling towers from aerial imagery during outbreak investigations requires expertise, is labour intensive, and can be prone to errors. We aimed to train a deep learning computer vision model to automatically detect cooling towers that are aerially visible.
Methods
Between Jan 1 and 31, 2021, we extracted satellite view images of Philadelphia (PN, USA) and New York state (NY, USA) from Google Maps and annotated cooling towers to create training datasets. We augmented training data with synthetic data and model-assisted labelling of additional cities. Using 2051 images containing 7292 cooling towers, we trained a two-stage model using YOLOv5, a model that detects objects in images, and EfficientNet-b5, a model that classifies images. We assessed the primary outcomes of sensitivity and positive predictive value (PPV) of the model against manual labelling on test datasets of 548 images, including from two cities not seen in training (Boston [MA, USA] and Athens [GA, USA]). We compared the search speed of the model with that of manual searching by four epidemiologists.
Findings
The model identified visible cooling towers with 95·1% sensitivity (95% CI 94·0–96·1) and a PPV of 90·1% (95% CI 90·0–90·2) in New York City and Philadelphia. In Boston, sensitivity was 91·6% (89·2–93·7) and PPV was 80·8% (80·5–81·2). In Athens, sensitivity was 86·9% (75·8–94·2) and PPV was 85·5% (84·2–86·7). For an area of New York City encompassing 45 blocks (0·26 square miles), the model searched more than 600 times faster (7·6 s; 351 potential cooling towers identified) than did human investigators (mean 83·75 min [SD 29·5]; mean 310·8 cooling towers [42·2]).
Interpretation
The model could be used to accelerate investigation and source control during outbreaks of Legionnaires’ disease through the identification of cooling towers from aerial imagery, potentially preventing additional disease spread. The model has already been used by public health teams for outbreak investigations and to initialise cooling tower registries, which are considered best practice for preventing and responding to outbreaks of Legionnaires’ disease.
{"title":"Automated cooling tower detection through deep learning for Legionnaires’ disease outbreak investigations: a model development and validation study","authors":"Karen K Wong MD , Thaddeus Segura MIDS , Gunnar Mein MIDS , Jia Lu PhD , Elizabeth J Hannapel MPH , Jasen M Kunz MPH , Troy Ritter PhD , Jessica C Smith MPH , Alberto Todeschini PhD , Fred Nugen PhD , Chris Edens PhD","doi":"10.1016/S2589-7500(24)00094-3","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00094-3","url":null,"abstract":"<div><h3>Background</h3><p>Cooling towers containing <em>Legionella</em> spp are a high-risk source of Legionnaires’ disease outbreaks. Manually locating cooling towers from aerial imagery during outbreak investigations requires expertise, is labour intensive, and can be prone to errors. We aimed to train a deep learning computer vision model to automatically detect cooling towers that are aerially visible.</p></div><div><h3>Methods</h3><p>Between Jan 1 and 31, 2021, we extracted satellite view images of Philadelphia (PN, USA) and New York state (NY, USA) from Google Maps and annotated cooling towers to create training datasets. We augmented training data with synthetic data and model-assisted labelling of additional cities. Using 2051 images containing 7292 cooling towers, we trained a two-stage model using YOLOv5, a model that detects objects in images, and EfficientNet-b5, a model that classifies images. We assessed the primary outcomes of sensitivity and positive predictive value (PPV) of the model against manual labelling on test datasets of 548 images, including from two cities not seen in training (Boston [MA, USA] and Athens [GA, USA]). We compared the search speed of the model with that of manual searching by four epidemiologists.</p></div><div><h3>Findings</h3><p>The model identified visible cooling towers with 95·1% sensitivity (95% CI 94·0–96·1) and a PPV of 90·1% (95% CI 90·0–90·2) in New York City and Philadelphia. In Boston, sensitivity was 91·6% (89·2–93·7) and PPV was 80·8% (80·5–81·2). In Athens, sensitivity was 86·9% (75·8–94·2) and PPV was 85·5% (84·2–86·7). For an area of New York City encompassing 45 blocks (0·26 square miles), the model searched more than 600 times faster (7·6 s; 351 potential cooling towers identified) than did human investigators (mean 83·75 min [SD 29·5]; mean 310·8 cooling towers [42·2]).</p></div><div><h3>Interpretation</h3><p>The model could be used to accelerate investigation and source control during outbreaks of Legionnaires’ disease through the identification of cooling towers from aerial imagery, potentially preventing additional disease spread. The model has already been used by public health teams for outbreak investigations and to initialise cooling tower registries, which are considered best practice for preventing and responding to outbreaks of Legionnaires’ disease.</p></div><div><h3>Funding</h3><p>None.</p></div>","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000943/pdfft?md5=134e7afec7443d66f0fb73e4c1e6aabb&pid=1-s2.0-S2589750024000943-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428960","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00088-8
Pedro Lopez-Ayala MD , Jasper Boeddinghaus MD , Thomas Nestelberger MD , Luca Koechlin MD , Tobias Zimmermann MD , Paolo Bima MD , Jonas Glaeser MD , Carlos C Spagnuolo MD , Arnaud Champetier MSc , Oscar Miro MD PhD , Francisco Javier Martin-Sanchez MD PhD , Dagmar I Keller MD , Michael Christ MD , Karin Wildi MD PhD , Tobias Breidthardt MD , Ivo Strebel PhD , Prof Christian Mueller MD
Background
The myocardial-ischaemic-injury-index (MI3) is a novel machine learning algorithm for the early diagnosis of type 1 non-ST-segment elevation myocardial infarction (NSTEMI). The performance of MI3, both when using early serial blood draws (eg, at 1 h or 2 h) and in direct comparison with guideline-recommended algorithms, remains unknown. Our aim was to externally validate MI3 and compare its performance with that of the European Society of Cardiology (ESC) 0/1h-algorithm.
Methods
In this secondary analysis of a multicentre international diagnostic cohort study, adult patients (age >18 years) presenting to the emergency department with symptoms suggestive of myocardial infarction were prospectively enrolled from April 21, 2006, to Feb 27, 2019 in 12 centres from five European countries (Switzerland, Spain, Italy, Poland, and Czech Republic). Patients were excluded if they presented with ST-segment-elevation myocardial infarction, did not have at least two serial high-sensitivity cardiac troponin I (hs-cTnI) measurements, or if the final diagnosis remained unclear. The final diagnosis was centrally adjudicated by two independent cardiologists using all available medical records, including serial hs-cTnI measurements and cardiac imaging. The primary outcome was type 1 NSTEMI. The performance of MI3 was directly compared with that of the ESC 0/1h-algorithm.
Findings
Among 6487 patients, (median age 61·0 years [IQR 49·0–73·0]; 2122 [33%] female and 4365 [67%] male), 882 (13·6%) patients had type 1 NSTEMI. The median time difference between the first and second hs-cTnI measurement was 60·0 mins (IQR 57·0–70·0). MI3 performance was very good, with an area under the receiver-operating-characteristic curve of 0·961 (95% CI 0·957 to 0·965) and a good overall calibration (intercept –0·09 [–0·2 to 0·02]; slope 1·02 [0·97 to 1·08]). The originally defined MI3 score of less than 1·6 identified 4186 (64·5%) patients as low probability of having a type 1 NSTEMI (sensitivity 99·1% [95% CI 98·2 to 99·5]; negative predictive value [NPV] 99·8% [95% CI 99·6 to 99·9]) and an MI3 score of 49·7 or more identified 915 (14·1%) patients as high probability of having a type 1 NSTEMI (specificity 95·0% [94·3 to 95·5]; positive predictive value [PPV] 69·1% [66·0–72·0]). The sensitivity and NPV of the ESC 0/1h-algorithm were higher than that of MI3 (difference for sensitivity 0·88% [0·19 to 1·60], p=0·0082; difference for NPV 0·18% [0·05 to 0·32], p=0·016), and the rule-out efficacy was higher for MI3 (11% difference, p<0·0001). Specificity and PPV for MI3 were superior (difference for specificity 3·80% [3·24 to 4·36], p<0·0001; difference for PPV 7·84% [5·86 to 9·97], p<0·0001), and the rule-in efficacy was higher for the ESC 0/1h-algorithm (5·4% difference, p<0·0001).
{"title":"External validation of the myocardial-ischaemic-injury-index machine learning algorithm for the early diagnosis of myocardial infarction: a multicentre cohort study","authors":"Pedro Lopez-Ayala MD , Jasper Boeddinghaus MD , Thomas Nestelberger MD , Luca Koechlin MD , Tobias Zimmermann MD , Paolo Bima MD , Jonas Glaeser MD , Carlos C Spagnuolo MD , Arnaud Champetier MSc , Oscar Miro MD PhD , Francisco Javier Martin-Sanchez MD PhD , Dagmar I Keller MD , Michael Christ MD , Karin Wildi MD PhD , Tobias Breidthardt MD , Ivo Strebel PhD , Prof Christian Mueller MD","doi":"10.1016/S2589-7500(24)00088-8","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00088-8","url":null,"abstract":"<div><h3>Background</h3><p>The myocardial-ischaemic-injury-index (MI<sup>3</sup>) is a novel machine learning algorithm for the early diagnosis of type 1 non-ST-segment elevation myocardial infarction (NSTEMI). The performance of MI<sup>3</sup>, both when using early serial blood draws (eg, at 1 h or 2 h) and in direct comparison with guideline-recommended algorithms, remains unknown. Our aim was to externally validate MI<sup>3</sup> and compare its performance with that of the European Society of Cardiology (ESC) 0/1h-algorithm.</p></div><div><h3>Methods</h3><p>In this secondary analysis of a multicentre international diagnostic cohort study, adult patients (age >18 years) presenting to the emergency department with symptoms suggestive of myocardial infarction were prospectively enrolled from April 21, 2006, to Feb 27, 2019 in 12 centres from five European countries (Switzerland, Spain, Italy, Poland, and Czech Republic). Patients were excluded if they presented with ST-segment-elevation myocardial infarction, did not have at least two serial high-sensitivity cardiac troponin I (hs-cTnI) measurements, or if the final diagnosis remained unclear. The final diagnosis was centrally adjudicated by two independent cardiologists using all available medical records, including serial hs-cTnI measurements and cardiac imaging. The primary outcome was type 1 NSTEMI. The performance of MI<sup>3</sup> was directly compared with that of the ESC 0/1h-algorithm.</p></div><div><h3>Findings</h3><p>Among 6487 patients, (median age 61·0 years [IQR 49·0–73·0]; 2122 [33%] female and 4365 [67%] male), 882 (13·6%) patients had type 1 NSTEMI. The median time difference between the first and second hs-cTnI measurement was 60·0 mins (IQR 57·0–70·0). MI<sup>3</sup> performance was very good, with an area under the receiver-operating-characteristic curve of 0·961 (95% CI 0·957 to 0·965) and a good overall calibration (intercept –0·09 [–0·2 to 0·02]; slope 1·02 [0·97 to 1·08]). The originally defined MI<sup>3</sup> score of less than 1·6 identified 4186 (64·5%) patients as low probability of having a type 1 NSTEMI (sensitivity 99·1% [95% CI 98·2 to 99·5]; negative predictive value [NPV] 99·8% [95% CI 99·6 to 99·9]) and an MI<sup>3</sup> score of 49·7 or more identified 915 (14·1%) patients as high probability of having a type 1 NSTEMI (specificity 95·0% [94·3 to 95·5]; positive predictive value [PPV] 69·1% [66·0–72·0]). The sensitivity and NPV of the ESC 0/1h-algorithm were higher than that of MI<sup>3</sup> (difference for sensitivity 0·88% [0·19 to 1·60], p=0·0082; difference for NPV 0·18% [0·05 to 0·32], p=0·016), and the rule-out efficacy was higher for MI<sup>3</sup> (11% difference, p<0·0001). Specificity and PPV for MI<sup>3</sup> were superior (difference for specificity 3·80% [3·24 to 4·36], p<0·0001; difference for PPV 7·84% [5·86 to 9·97], p<0·0001), and the rule-in efficacy was higher for the ESC 0/1h-algorithm (5·4% difference, p<0·0001).</p></div>","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000888/pdfft?md5=4687bfa0693df8237a23349722a85e46&pid=1-s2.0-S2589750024000888-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428958","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00100-6
Grace B Hatton , Christie Brooks
{"title":"A response to evaluating national data flows","authors":"Grace B Hatton , Christie Brooks","doi":"10.1016/S2589-7500(24)00100-6","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00100-6","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024001006/pdfft?md5=71db10e05782e151870bf4bed71961b6&pid=1-s2.0-S2589750024001006-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428953","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00122-5
The Lancet Digital Health
{"title":"The lofty heights of digital health","authors":"The Lancet Digital Health","doi":"10.1016/S2589-7500(24)00122-5","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00122-5","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024001225/pdfft?md5=0f5b1d023170bc6a9b888b87db25ca61&pid=1-s2.0-S2589750024001225-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428730","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 : 2024-06-19DOI: 10.1016/S2589-7500(24)00070-0
Prof Sebastian Kohlmann PhD , Franziska Sikorski MSc , Prof Hans-Helmut König MD , Marion Schütt MSc , Prof Antonia Zapf PhD , Prof Bernd Löwe MD
Background
Despite the availability of effective treatments, most depressive disorders remain undetected and untreated. Internet-based depression screening combined with automated feedback of screening results could reach people with depression and lead to evidence-based care. We aimed to test the efficacy of two versions of automated feedback after internet-based screening on depression severity compared with no feedback.
Methods
DISCOVER was an observer-masked, three-armed, randomised controlled trial in Germany. We recruited individuals (aged ≥18 years) who were undiagnosed with depression and screened positive for depression on an internet-based self-report depression rating scale (Patient Health Questionnaire-9 [PHQ-9] ≥10 points). Participants were randomly assigned 1:1:1 to automatically receive no feedback, non-tailored feedback, or tailored feedback on the depression screening result. Randomisation was stratified by depression severity (moderate: PHQ-9 score 10–14 points; severe: PHQ-9 score ≥15 points). Participants could not be masked but were kept unaware of trial hypotheses to minimise expectancy bias. The non-tailored feedback included the depression screening result, a recommendation to seek professional diagnostic advice, and brief general information about depression and its treatment. The tailored feedback included the same basic information but individually framed according to the participants’ symptom profiles, treatment preferences, causal symptom attributions, health insurance, and local residence. Research staff were masked to group allocation and outcome assessment as these were done using online questionnaires. The primary outcome was change in depression severity, defined as change in PHQ-9 score 6 months after random assignment. Analyses were conducted following the intention-to-treat principle for participants with at least one follow-up visit. This trial was registered at ClinicalTrials.gov, NCT04633096.
Findings
Between Jan 12, 2021, and Jan 31, 2022, 4878 individuals completed the internet-based screening. Of these, 1178 (24%) screened positive for depression (mean age 37·1 [SD 14·2] years; 824 [70%] woman, 344 [29%] men, and 10 [1%] other gender identity). 6 months after random assignment, depression severity decreased by 3·4 PHQ-9 points in the no feedback group (95% CI 2·9–4·0; within-group d 0·67; 325 participants), by 3·5 points in the non-tailored feedback group (3·0–4·0; within-group d 0·74; 319 participants), and by 3·7 points in the tailored feedback group (3·2–4·3; within-group d 0·71; 321 participants), with no significant differences among the three groups (p=0·72). The number of participants seeking help for depression or initiating psychotherapy or antidepressant treatment did not differ among study groups. The results remained consistent when adjusted for fulfilli
{"title":"The efficacy of automated feedback after internet-based depression screening (DISCOVER): an observer-masked, three-armed, randomised controlled trial in Germany","authors":"Prof Sebastian Kohlmann PhD , Franziska Sikorski MSc , Prof Hans-Helmut König MD , Marion Schütt MSc , Prof Antonia Zapf PhD , Prof Bernd Löwe MD","doi":"10.1016/S2589-7500(24)00070-0","DOIUrl":"https://doi.org/10.1016/S2589-7500(24)00070-0","url":null,"abstract":"<div><h3>Background</h3><p>Despite the availability of effective treatments, most depressive disorders remain undetected and untreated. Internet-based depression screening combined with automated feedback of screening results could reach people with depression and lead to evidence-based care. We aimed to test the efficacy of two versions of automated feedback after internet-based screening on depression severity compared with no feedback.</p></div><div><h3>Methods</h3><p>DISCOVER was an observer-masked, three-armed, randomised controlled trial in Germany. We recruited individuals (aged ≥18 years) who were undiagnosed with depression and screened positive for depression on an internet-based self-report depression rating scale (Patient Health Questionnaire-9 [PHQ-9] ≥10 points). Participants were randomly assigned 1:1:1 to automatically receive no feedback, non-tailored feedback, or tailored feedback on the depression screening result. Randomisation was stratified by depression severity (moderate: PHQ-9 score 10–14 points; severe: PHQ-9 score ≥15 points). Participants could not be masked but were kept unaware of trial hypotheses to minimise expectancy bias. The non-tailored feedback included the depression screening result, a recommendation to seek professional diagnostic advice, and brief general information about depression and its treatment. The tailored feedback included the same basic information but individually framed according to the participants’ symptom profiles, treatment preferences, causal symptom attributions, health insurance, and local residence. Research staff were masked to group allocation and outcome assessment as these were done using online questionnaires. The primary outcome was change in depression severity, defined as change in PHQ-9 score 6 months after random assignment. Analyses were conducted following the intention-to-treat principle for participants with at least one follow-up visit. This trial was registered at <span>ClinicalTrials.gov</span><svg><path></path></svg>, <span>NCT04633096</span><svg><path></path></svg>.</p></div><div><h3>Findings</h3><p>Between Jan 12, 2021, and Jan 31, 2022, 4878 individuals completed the internet-based screening. Of these, 1178 (24%) screened positive for depression (mean age 37·1 [SD 14·2] years; 824 [70%] woman, 344 [29%] men, and 10 [1%] other gender identity). 6 months after random assignment, depression severity decreased by 3·4 PHQ-9 points in the no feedback group (95% CI 2·9–4·0; within-group d 0·67; 325 participants), by 3·5 points in the non-tailored feedback group (3·0–4·0; within-group d 0·74; 319 participants), and by 3·7 points in the tailored feedback group (3·2–4·3; within-group d 0·71; 321 participants), with no significant differences among the three groups (p=0·72). The number of participants seeking help for depression or initiating psychotherapy or antidepressant treatment did not differ among study groups. The results remained consistent when adjusted for fulfilli","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000700/pdfft?md5=339cfa330e95efe94536fa8c159d0a77&pid=1-s2.0-S2589750024000700-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428956","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 : 2024-06-06DOI: 10.1016/S2589-7500(24)00085-2
Jue Wang MS , Nafen Zheng BSc , Huan Wan BSc , Qinyue Yao MS , Shijun Jia MS , Xin Zhang MS , Sha Fu MD , Jingliang Ruan MD , Gui He BSc , Xulin Chen MS , Suiping Li MS , Rui Chen BSc , Boan Lai BSc , Jin Wang PhD , Prof Qingping Jiang MD , Prof Nengtai Ouyang MD , Yin Zhang PhD
Background
Accurately distinguishing between malignant and benign thyroid nodules through fine-needle aspiration cytopathology is crucial for appropriate therapeutic intervention. However, cytopathologic diagnosis is time consuming and hindered by the shortage of experienced cytopathologists. Reliable assistive tools could improve cytopathologic diagnosis efficiency and accuracy. We aimed to develop and test an artificial intelligence (AI)-assistive system for thyroid cytopathologic diagnosis according to the Thyroid Bethesda Reporting System.
Methods
11 254 whole-slide images (WSIs) from 4037 patients were used to train deep learning models. Among the selected WSIs, cell level was manually annotated by cytopathologists according to The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) guidelines of the second edition (2017 version). A retrospective dataset of 5638 WSIs of 2914 patients from four medical centres was used for validation. 469 patients were recruited for the prospective study of the performance of AI models and their 537 thyroid nodule samples were used. Cohorts for training and validation were enrolled between Jan 1, 2016, and Aug 1, 2022, and the prospective dataset was recruited between Aug 1, 2022, and Jan 1, 2023. The performance of our AI models was estimated as the area under the receiver operating characteristic (AUROC), sensitivity, specificity, accuracy, positive predictive value, and negative predictive value. The primary outcomes were the prediction sensitivity and specificity of the model to assist cyto-diagnosis of thyroid nodules.
Findings
The AUROC of TBSRTC III+ (which distinguishes benign from TBSRTC classes III, IV, V, and VI) was 0·930 (95% CI 0·921–0·939) for Sun Yat-sen Memorial Hospital of Sun Yat-sen University (SYSMH) internal validation and 0·944 (0·929 – 0·959), 0·939 (0·924–0·955), 0·971 (0·938–1·000) for The First People's Hospital of Foshan (FPHF), Sichuan Cancer Hospital & Institute (SCHI), and The Third Affiliated Hospital of Guangzhou Medical University (TAHGMU) medical centres, respectively. The AUROC of TBSRTC V+ (which distinguishes benign from TBSRTC classes V and VI) was 0·990 (95% CI 0·986–0·995) for SYSMH internal validation and 0·988 (0·980–0·995), 0·965 (0·953–0·977), and 0·991 (0·972–1·000) for FPHF, SCHI, and TAHGMU medical centres, respectively. For the prospective study at SYSMH, the AUROC of TBSRTC III+ and TBSRTC V+ was 0·977 and 0·981, respectively. With the assistance of AI, the specificity of junior cytopathologists was boosted from 0·887 (95% CI 0·8440–0·922) to 0·993 (0·974–0·999) and the accuracy was improved from 0·877 (0·846–0·904) to 0·948 (0·926–0·965). 186 atypia of undetermined significance samples from 186 patients with BRAF mutation information were collected; 43 of them harbour the BRAFV600E mutation. 91% (39/43) of BRAFV600E-positive atypia o
{"title":"Deep learning models for thyroid nodules diagnosis of fine-needle aspiration biopsy: a retrospective, prospective, multicentre study in China","authors":"Jue Wang MS , Nafen Zheng BSc , Huan Wan BSc , Qinyue Yao MS , Shijun Jia MS , Xin Zhang MS , Sha Fu MD , Jingliang Ruan MD , Gui He BSc , Xulin Chen MS , Suiping Li MS , Rui Chen BSc , Boan Lai BSc , Jin Wang PhD , Prof Qingping Jiang MD , Prof Nengtai Ouyang MD , Yin Zhang PhD","doi":"10.1016/S2589-7500(24)00085-2","DOIUrl":"10.1016/S2589-7500(24)00085-2","url":null,"abstract":"<div><h3>Background</h3><p>Accurately distinguishing between malignant and benign thyroid nodules through fine-needle aspiration cytopathology is crucial for appropriate therapeutic intervention. However, cytopathologic diagnosis is time consuming and hindered by the shortage of experienced cytopathologists. Reliable assistive tools could improve cytopathologic diagnosis efficiency and accuracy. We aimed to develop and test an artificial intelligence (AI)-assistive system for thyroid cytopathologic diagnosis according to the Thyroid Bethesda Reporting System.</p></div><div><h3>Methods</h3><p>11 254 whole-slide images (WSIs) from 4037 patients were used to train deep learning models. Among the selected WSIs, cell level was manually annotated by cytopathologists according to The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) guidelines of the second edition (2017 version). A retrospective dataset of 5638 WSIs of 2914 patients from four medical centres was used for validation. 469 patients were recruited for the prospective study of the performance of AI models and their 537 thyroid nodule samples were used. Cohorts for training and validation were enrolled between Jan 1, 2016, and Aug 1, 2022, and the prospective dataset was recruited between Aug 1, 2022, and Jan 1, 2023. The performance of our AI models was estimated as the area under the receiver operating characteristic (AUROC), sensitivity, specificity, accuracy, positive predictive value, and negative predictive value. The primary outcomes were the prediction sensitivity and specificity of the model to assist cyto-diagnosis of thyroid nodules.</p></div><div><h3>Findings</h3><p>The AUROC of TBSRTC III+ (which distinguishes benign from TBSRTC classes III, IV, V, and VI) was 0·930 (95% CI 0·921–0·939) for Sun Yat-sen Memorial Hospital of Sun Yat-sen University (SYSMH) internal validation and 0·944 (0·929 – 0·959), 0·939 (0·924–0·955), 0·971 (0·938–1·000) for The First People's Hospital of Foshan (FPHF), Sichuan Cancer Hospital & Institute (SCHI), and The Third Affiliated Hospital of Guangzhou Medical University (TAHGMU) medical centres, respectively. The AUROC of TBSRTC V+ (which distinguishes benign from TBSRTC classes V and VI) was 0·990 (95% CI 0·986–0·995) for SYSMH internal validation and 0·988 (0·980–0·995), 0·965 (0·953–0·977), and 0·991 (0·972–1·000) for FPHF, SCHI, and TAHGMU medical centres, respectively. For the prospective study at SYSMH, the AUROC of TBSRTC III+ and TBSRTC V+ was 0·977 and 0·981, respectively. With the assistance of AI, the specificity of junior cytopathologists was boosted from 0·887 (95% CI 0·8440–0·922) to 0·993 (0·974–0·999) and the accuracy was improved from 0·877 (0·846–0·904) to 0·948 (0·926–0·965). 186 atypia of undetermined significance samples from 186 patients with <em>BRAF</em> mutation information were collected; 43 of them harbour the <em>BRAF</em><sup>V600E</sup> mutation. 91% (39/43) of <em>BRAF</em><sup>V600E</sup>-positive atypia o","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000852/pdfft?md5=d718eec693d6690f3aa369916941141d&pid=1-s2.0-S2589750024000852-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141288784","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 : 2024-05-29DOI: 10.1016/S2589-7500(24)00090-6
Melany Gaetani , Mjaye Mazwi , Hadrian Balaci , Robert Greer , Christina Maratta
{"title":"Artificial intelligence in medicine and the pursuit of environmentally responsible science","authors":"Melany Gaetani , Mjaye Mazwi , Hadrian Balaci , Robert Greer , Christina Maratta","doi":"10.1016/S2589-7500(24)00090-6","DOIUrl":"10.1016/S2589-7500(24)00090-6","url":null,"abstract":"","PeriodicalId":48534,"journal":{"name":"Lancet Digital Health","volume":null,"pages":null},"PeriodicalIF":30.8,"publicationDate":"2024-05-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589750024000906/pdfft?md5=1aaa78fbfb7a49f99226ead1aa0c786c&pid=1-s2.0-S2589750024000906-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141180100","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}