Assessing illness severity in the ICU is crucial for early prediction of deterioration and prognosis. Traditional prognostic scores often treat organ systems separately, overlooking the body's interconnected nature. Network physiology offers a new approach to understanding these complex interactions. This study used the concept of transfer entropy (TE) to measure information flow between heart rate (HR), respiratory rate (RR), and capillary oxygen saturation (SpO2) in critically ill sepsis patients, hypothesizing that TE between these signals would correlate with disease outcome. The retrospective cohort study utilized the MIMIC III Clinical Database, including patients who met Sepsis-3 criteria on admission and had 30 minutes of continuous HR, RR, and SpO2 data. TE between the signals was calculated to create physiological network maps. Cox regression assessed the relationship between cardiorespiratory network indices and both deterioration (SOFA score increase of ≥2 points at 48 hours) and 30-day mortality. Among 164 patients, higher information flow from SpO2 to HR [TE(SpO2 → HR)] and reciprocal flow between HR and RR [TE(RR → HR) and TE(HR → RR)] were linked to reduced mortality, independent of age, mechanical ventilation, SOFA score, and comorbidity. Reductions in TE(HR → RR), TE(RR → HR), TE(SpO2 → RR), and TE(SpO2 → HR) were associated with increased risk of 48-hour deterioration. After adjustment for potential confounders, only TE(HR → RR) and TE(RR → HR) remained statistically significant. The study confirmed that physiological network mapping using routine signals in sepsis patients could indicate illness severity and that higher TE values were generally associated with improved outcomes.
{"title":"Decreased cardio-respiratory information transfer is associated with deterioration and a poor prognosis in critically ill patients with sepsis","authors":"Cecilia Morandotti, Matthew Wikner, Qijun Li, Emily Ito, Calix Tan, Pin-Yu Chen, Anika Cawthorn, Watjana Lilaonitkul, Alireza Mani","doi":"10.1101/2024.08.18.24312167","DOIUrl":"https://doi.org/10.1101/2024.08.18.24312167","url":null,"abstract":"Assessing illness severity in the ICU is crucial for early prediction of deterioration and prognosis. Traditional prognostic scores often treat organ systems separately, overlooking the body's interconnected nature. Network physiology offers a new approach to understanding these complex interactions. This study used the concept of transfer entropy (TE) to measure information flow between heart rate (HR), respiratory rate (RR), and capillary oxygen saturation (SpO2) in critically ill sepsis patients, hypothesizing that TE between these signals would correlate with disease outcome. The retrospective cohort study utilized the MIMIC III Clinical Database, including patients who met Sepsis-3 criteria on admission and had 30 minutes of continuous HR, RR, and SpO2 data. TE between the signals was calculated to create physiological network maps. Cox regression assessed the relationship between cardiorespiratory network indices and both deterioration (SOFA score increase of ≥2 points at 48 hours) and 30-day mortality. Among 164 patients, higher information flow from SpO2 to HR [TE(SpO2 → HR)] and reciprocal flow between HR and RR [TE(RR → HR) and TE(HR → RR)] were linked to reduced mortality, independent of age, mechanical ventilation, SOFA score, and comorbidity. Reductions in TE(HR → RR), TE(RR → HR), TE(SpO2 → RR), and TE(SpO2 → HR) were associated with increased risk of 48-hour deterioration. After adjustment for potential confounders, only TE(HR → RR) and TE(RR → HR) remained statistically significant. The study confirmed that physiological network mapping using routine signals in sepsis patients could indicate illness severity and that higher TE values were generally associated with improved outcomes.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179487","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 : 2024-08-16DOI: 10.1101/2024.08.12.24311770
Suraj Sudarsanan, Praveen Sivadasan, Prem Chandra, Amr S Omar, Kathy Lynn Gaviola Atuel, Hafeez Ulla Lone, Hany Osman Elsayed Ragab, Irshad Ehsan, Cornelia Sonia Carr, Abdul Rasheed Pattath, Abdulaziz Al Khulaifi, Yasser Mahfouz Eltokhy Shouman, Abdulwahid Al Mulla
Background: Assess the ability of APACHE II (acute physiology and chronic health evaluation II), SOFA (sequential organ failure assessment scores), Cardiac Surgery Score (CASUS) and SAVE (Survival after VA-ECMO) to predict outcomes in a cohort of patients undergoing Veno-Arterial ECMO (VA-ECMO) Methods: Observational retrospective study of all patients admitted to Cardiothoracic Intensive Care Unit (CTICU) for a minimum duration of 24 hours after undergoing VA-ECMO insertion between years 2015 to 2022. Scores for APACHE II, SOFA and CASUS were calculated at 24 after ICU admission. SAVE score was calculated from the last available patient details within 24 hours of ECMO insertion. Demographic, clinical, and laboratory data relevant for the study was retrieved from electronic patient records. Results: Pre-ECMO serum levels of lactates and creatinine were significantly associated with mortality. Lower ECMO flow rates at 4 hours and 12 hours after ECMO cannulation was significantly associated with survival to discharge. Development of arrythmias, acute kidney injury (AKI) and need of continuous renal replacement therapy (CRRT) while on ECMO were significantly associated with mortality. The APACHE-II, SOFA and CASUS, calculated at 24 hours of ICU admission were significantly higher amongst non-survivors. Following categorization of risk scores using ROC curve analysis, it was found that APACHE-II, SOFA and CASUS calculated at 24 hours of ICU admission after ECMO insertion demonstrated moderate predictive ability for mortality whereas SAVE score failed to predict mortality. APACHE-II >27 (AUC of 0.66) calculated at 24 hours of ICU admission after ECMO insertion, demonstrated the greatest predictive ability, for mortality. Multivariate logistic regression analysis of the four scores showed that APACHE-II > 27 and SOFA > 14 calculated at 24 hours of ICU admission after ECMO insertion, were independently significantly predictive of mortality Conclusions: The APACHE-II, SOFA and CASUS, calculated at 24 hours of ICU admission were significantly higher amongst non-survivors as compared to survivors. APACHE-II demonstrated the best mortality predictive ability. APACHE-II scores of 27 or above, and SOFA of 14 or above at 24 hours of ICU admission after ECMO cannulation can predict mortality and will aid physicians in decision making
{"title":"Comparison of four intensive care scores in prediction of outcome after Veno-Arterial ECMO: A single center retrospective study","authors":"Suraj Sudarsanan, Praveen Sivadasan, Prem Chandra, Amr S Omar, Kathy Lynn Gaviola Atuel, Hafeez Ulla Lone, Hany Osman Elsayed Ragab, Irshad Ehsan, Cornelia Sonia Carr, Abdul Rasheed Pattath, Abdulaziz Al Khulaifi, Yasser Mahfouz Eltokhy Shouman, Abdulwahid Al Mulla","doi":"10.1101/2024.08.12.24311770","DOIUrl":"https://doi.org/10.1101/2024.08.12.24311770","url":null,"abstract":"Background: Assess the ability of APACHE II (acute physiology and chronic health evaluation II), SOFA (sequential organ failure assessment scores), Cardiac Surgery Score (CASUS) and SAVE (Survival after VA-ECMO) to predict outcomes in a cohort of patients undergoing Veno-Arterial ECMO (VA-ECMO)\u0000Methods: Observational retrospective study of all patients admitted to Cardiothoracic Intensive Care Unit (CTICU) for a minimum duration of 24 hours after undergoing VA-ECMO insertion between years 2015 to 2022. Scores for APACHE II, SOFA and CASUS were calculated at 24 after ICU admission. SAVE score was calculated from the last available patient details within 24 hours of ECMO insertion. Demographic, clinical, and laboratory data relevant for the study was retrieved from electronic patient records.\u0000Results: Pre-ECMO serum levels of lactates and creatinine were significantly associated with mortality. Lower ECMO flow rates at 4 hours and 12 hours after ECMO cannulation was significantly associated with survival to discharge. Development of arrythmias, acute kidney injury (AKI) and need of continuous renal replacement therapy (CRRT) while on ECMO were significantly associated with mortality. The APACHE-II, SOFA and CASUS, calculated at 24 hours of ICU admission were significantly higher amongst non-survivors. Following categorization of risk scores using ROC curve analysis, it was found that APACHE-II, SOFA and CASUS calculated at 24 hours of ICU admission after ECMO insertion demonstrated moderate predictive ability for mortality whereas SAVE score failed to predict mortality. APACHE-II >27 (AUC of 0.66) calculated at 24 hours of ICU admission after ECMO insertion, demonstrated the greatest predictive ability, for mortality. Multivariate logistic regression analysis of the four scores showed that APACHE-II > 27 and SOFA > 14 calculated at 24 hours of ICU admission after ECMO insertion, were independently significantly predictive of mortality\u0000Conclusions: The APACHE-II, SOFA and CASUS, calculated at 24 hours of ICU admission were significantly higher amongst non-survivors as compared to survivors. APACHE-II demonstrated the best mortality predictive ability. APACHE-II scores of 27 or above, and SOFA of 14 or above at 24 hours of ICU admission after ECMO cannulation can predict mortality and will aid physicians in decision making","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"29 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142179488","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 : 2024-08-07DOI: 10.1101/2024.08.06.24311556
Wonsuk Oh, Kullaya Takkavatakarn, Hannah Kittrell, Khaled Shawwa, Hernando Gomez, Ashwin S Sawant, Pranai Tandon, Gagan Kumar, Michael Sterling, Ira Hofer, Lili Chan, John Oropello, Roopa Kohli-Seth, Alexander W Charney, Monica Kraft, Patricia Kovatch, John A Kellum, Girish N Nadkarni, Ankit Sakhuja
Background: Intravenous fluids are mainstay of management of acute kidney injury (AKI) after sepsis but can cause fluid overload. Recent literature shows that restrictive fluid strategy may be beneficial in some patients with AKI, however, identifying these patients is challenging. We aimed to develop and validate a machine learning algorithm to identify patients who would benefit from a restrictive fluid strategy. Methods and Findings: We included patients with sepsis who developed AKI within 48 hours of ICU admission and defined restrictive fluid strategy as receiving <500mL fluids within 24 hours after AKI. Our primary outcome was early AKI reversal within 48 hours of AKI onset, and secondary outcomes included sustained AKI reversal and major adverse kidney events (MAKE) at discharge. We used a causal forest, a machine learning algorithm to estimate individual treatment effects and policy tree algorithm to identify patients who would benefit by restrictive fluid strategy. We developed the algorithm in MIMIC-IV and validated it in eICU database. Among 2,091 patients in the external validation cohort, policy tree recommended restrictive fluids for 88.2%. Among these, patients who received restrictive fluids demonstrated significantly higher rate of early AKI reversal (48.2% vs 39.6%, p<0.001), sustained AKI reversal (36.7% vs 27.4%, p<0.001) and lower rates of MAKE by discharge (29.3% vs 35.1%, p=0.019). These results were consistent in adjusted analysis. Conclusion: Policy tree based on causal machine learning can identify septic patients with AKI who benefit from a restrictive fluid strategy. This approach needs to be validated in prospective trials.
背景：静脉输液是治疗脓毒症后急性肾损伤（AKI）的主要方法，但可能导致液体超负荷。最近的文献显示，限制性输液策略可能对某些 AKI 患者有益，但识别这些患者却很困难。我们的目标是开发并验证一种机器学习算法，以识别可从限制性输液策略中获益的患者。方法和结果：我们纳入了在入住 ICU 48 小时内发生 AKI 的脓毒症患者，并将限制性输液策略定义为在发生 AKI 后 24 小时内接受 500 毫升液体。我们的主要结果是在 AKI 发生后 48 小时内早期逆转 AKI，次要结果包括持续逆转 AKI 和出院时的主要肾脏不良事件 (MAKE)。我们使用因果森林（一种估计个体治疗效果的机器学习算法）和政策树算法来确定哪些患者可从限制性输液策略中获益。我们在 MIMIC-IV 中开发了该算法，并在 eICU 数据库中进行了验证。在这些患者中，接受限制性输液的患者早期 AKI 逆转率（48.2% vs 39.6%，p<0.001）、持续 AKI 逆转率（36.7% vs 27.4%，p<0.001）和出院时 MAKE 率（29.3% vs 35.1%，p=0.019）均明显较高。这些结果在调整分析中保持一致。结论基于因果机器学习的策略树能识别出从限制性输液策略中获益的脓毒症 AKI 患者。这种方法需要在前瞻性试验中进行验证。
{"title":"Development and Validation of a Policy Tree Approach for Optimizing Intravenous Fluids in Critically Ill Patients with Sepsis and Acute Kidney Injury","authors":"Wonsuk Oh, Kullaya Takkavatakarn, Hannah Kittrell, Khaled Shawwa, Hernando Gomez, Ashwin S Sawant, Pranai Tandon, Gagan Kumar, Michael Sterling, Ira Hofer, Lili Chan, John Oropello, Roopa Kohli-Seth, Alexander W Charney, Monica Kraft, Patricia Kovatch, John A Kellum, Girish N Nadkarni, Ankit Sakhuja","doi":"10.1101/2024.08.06.24311556","DOIUrl":"https://doi.org/10.1101/2024.08.06.24311556","url":null,"abstract":"Background: Intravenous fluids are mainstay of management of acute kidney injury (AKI) after sepsis but can cause fluid overload. Recent literature shows that restrictive fluid strategy may be beneficial in some patients with AKI, however, identifying these patients is challenging. We aimed to develop and validate a machine learning algorithm to identify patients who would benefit from a restrictive fluid strategy. Methods and Findings: We included patients with sepsis who developed AKI within 48 hours of ICU admission and defined restrictive fluid strategy as receiving <500mL fluids within 24 hours after AKI. Our primary outcome was early AKI reversal within 48 hours of AKI onset, and secondary outcomes included sustained AKI reversal and major adverse kidney events (MAKE) at discharge. We used a causal forest, a machine learning algorithm to estimate individual treatment effects and policy tree algorithm to identify patients who would benefit by restrictive fluid strategy. We developed the algorithm in MIMIC-IV and validated it in eICU database.\u0000Among 2,091 patients in the external validation cohort, policy tree recommended restrictive fluids for 88.2%. Among these, patients who received restrictive fluids demonstrated significantly higher rate of early AKI reversal (48.2% vs 39.6%, p<0.001), sustained AKI reversal (36.7% vs 27.4%, p<0.001) and lower rates of MAKE by discharge (29.3% vs 35.1%, p=0.019). These results were consistent in adjusted analysis. Conclusion: Policy tree based on causal machine learning can identify septic patients with AKI who benefit from a restrictive fluid strategy. This approach needs to be validated in prospective trials.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"58 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141939388","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 : 2024-07-24DOI: 10.1101/2024.07.24.24310868
Goekmen Aktas, Felix Keller, Justyna Siwy, Agnieszka Latosinska, Harald Mischak, Jorge Mayor Ramirez, Jan-Dierk Clausen, Vesta Brauckmann, Michaela Wilhelmi, Stephan Sehmisch, Tarek Omar Pacha
Abstract Background: Treatment of severely injured patients represents a major challenge, in part due to the unpredictable risk of major adverse events, including death. Preemptive personalized treatment aimed at preventing these events is a key objective of patient management; however, the currently available scoring systems provide only moderate guidance. Molecular biomarkers from proteomics/peptidomics studies hold promise for improving the current situation, ultimately enabling precision medicine based on individual molecular profiles. Methods: To test the hypothesis that proteomics biomarkers could predict patient outcomes in severely injured patients, we initiated a pilot study involving consecutive urine sampling (on days 0, 2, 5, 10, and 14) and subsequent peptidome analysis using capillary electrophoresis coupled to mass spectrometry (CE-MS) of 14 severely injured patients and two additional ICU patients. The urine peptidomes of these patients were compared to the urine peptidomes of age- and sex-matched controls. Previously established urinary peptide-based classifiers, CKD274, AKI204, and CoV50, were applied to the obtained peptidome data, and the association of the scores with a combined endpoint (death and/or kidney failure and/or respiratory insufficiency) was investigated. Results: CE-MS peptidome analysis identified 281 peptides that were significantly altered in severely injured patients. Consistent upregulation was observed for peptides from A1AT, FETUA, and MYG, while peptides derived from CD99, PIGR and UROM were consistently reduced. Most of the significant peptides were from different collagens, and the majority were reduced in abundance. Two of the predefined peptidomic classifiers, CKD273 and AKI204, showed significant associations with the combined endpoint, which was not observed for the routine scores generally applied in the clinics. Conclusions: This prospective pilot study confirmed the hypothesis that urinary peptides provide information on patient outcomes and may guide personalized interventions based on individual molecular changes. The results obtained allow the planning of a well-powered prospective trial investigating the value of urinary peptides in this context in more detail. Keywords: urine, biomarker, trauma, polytrauma, intensive care, critical care, proteomics, peptides, prediction
{"title":"Application of Urinary Peptide-Biomarkers in Trauma Patients as a Predictive Tool for Prognostic Assessment, Treatment Interventions, and Intervention Timing: Prospective Nonrandomized Pilot Study","authors":"Goekmen Aktas, Felix Keller, Justyna Siwy, Agnieszka Latosinska, Harald Mischak, Jorge Mayor Ramirez, Jan-Dierk Clausen, Vesta Brauckmann, Michaela Wilhelmi, Stephan Sehmisch, Tarek Omar Pacha","doi":"10.1101/2024.07.24.24310868","DOIUrl":"https://doi.org/10.1101/2024.07.24.24310868","url":null,"abstract":"Abstract Background: Treatment of severely injured patients represents a major challenge, in part due to the unpredictable risk of major adverse events, including death. Preemptive personalized treatment aimed at preventing these events is a key objective of patient management; however, the currently available scoring systems provide only moderate guidance. Molecular biomarkers from proteomics/peptidomics studies hold promise for improving the current situation, ultimately enabling precision medicine based on individual molecular profiles.\u0000Methods: To test the hypothesis that proteomics biomarkers could predict patient outcomes in severely injured patients, we initiated a pilot study involving consecutive urine sampling (on days 0, 2, 5, 10, and 14) and subsequent peptidome analysis using capillary electrophoresis coupled to mass spectrometry (CE-MS) of 14 severely injured patients and two additional ICU patients. The urine peptidomes of these patients were compared to the urine peptidomes of age- and sex-matched controls. Previously established urinary peptide-based classifiers, CKD274, AKI204, and CoV50, were applied to the obtained peptidome data, and the association of the scores with a combined endpoint (death and/or kidney failure and/or respiratory insufficiency) was investigated.\u0000Results: CE-MS peptidome analysis identified 281 peptides that were significantly altered in severely injured patients. Consistent upregulation was observed for peptides from A1AT, FETUA, and MYG, while peptides derived from CD99, PIGR and UROM were consistently reduced. Most of the significant peptides were from different collagens, and the majority were reduced in abundance. Two of the predefined peptidomic classifiers, CKD273 and AKI204, showed significant associations with the combined endpoint, which was not observed for the routine scores generally applied in the clinics.\u0000Conclusions: This prospective pilot study confirmed the hypothesis that urinary peptides provide information on patient outcomes and may guide personalized interventions based on individual molecular changes. The results obtained allow the planning of a well-powered prospective trial investigating the value of urinary peptides in this context in more detail.\u0000Keywords: urine, biomarker, trauma, polytrauma, intensive care, critical care, proteomics, peptides, prediction","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"40 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778102","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 : 2024-07-24DOI: 10.1101/2024.07.23.24310792
Wout J. Claassen, Marloes van den Berg, Rianne J Baelde, Sylvia J.P. Bogaards, Luuk Bonis, Heleen C. Hakkeling, Gerben Schaaf, Albertus Beishuizen, Chris Dickhoff, Reinier A. Boon, Leo Heunks, Tyler J. Kirby, Coen A.C. Ottenheijm
Abstract (236 words) Rationale. Intensive care unit (ICU) acquired diaphragm weakness is a common consequence of mechanical ventilation (MV). It contributes to difficult weaning, which is associated with increased morbidity and mortality. Diaphragm weakness is caused by a combination of atrophy and dysfunction of myofibers, large syncytial cells that are maintained by a population of myonuclei. Each myonucleus provides gene transcripts to a finite fiber volume, termed the myonuclear domain. Myonuclear loss in myofibers undergoing atrophy is subject to debate. Myonuclear number is a determinant of transcriptional capacity, and therefore critical for muscle regeneration after atrophy. Objectives. Our objective was to investigate if and how myonuclear number is altered in the diaphragm of mechanically ventilated ICU patients. Methods. We used a combination of confocal microscopy, transcriptomics, and immunohistochemistry techniques to study myonuclear alterations in diaphragm and quadriceps biopsies from MV ICU patients. Measurements and Main Results. Patients with established diaphragm atrophy had a reduced myonuclear number and myonuclear domain. Intrinsic apoptotic pathway activation was identified as a potential mechanism underlying myonuclear removal in the diaphragm of mechanically ventilated ICU patients. Total transcription of myofibers decreased with myonuclear loss. Furthermore, muscle stem cell number was reduced in the patients with diaphragm atrophy. Conclusion. We identified myonuclear loss due to intrinsic apoptotic pathway activation as a mechanism underlying diaphragm atrophy in mechanically ventilated patients. The loss of myonuclei may contribute to difficult weaning due to impaired regrowth of myofibers after atrophy.
{"title":"Myonuclear apoptosis underlies diaphragm atrophy in mechanically ventilated ICU patients.","authors":"Wout J. Claassen, Marloes van den Berg, Rianne J Baelde, Sylvia J.P. Bogaards, Luuk Bonis, Heleen C. Hakkeling, Gerben Schaaf, Albertus Beishuizen, Chris Dickhoff, Reinier A. Boon, Leo Heunks, Tyler J. Kirby, Coen A.C. Ottenheijm","doi":"10.1101/2024.07.23.24310792","DOIUrl":"https://doi.org/10.1101/2024.07.23.24310792","url":null,"abstract":"Abstract (236 words)\u0000Rationale. Intensive care unit (ICU) acquired diaphragm weakness is a common consequence of mechanical ventilation (MV). It contributes to difficult weaning, which is associated with increased morbidity and mortality. Diaphragm weakness is caused by a combination of atrophy and dysfunction of myofibers, large syncytial cells that are maintained by a population of myonuclei. Each myonucleus provides gene transcripts to a finite fiber volume, termed the myonuclear domain. Myonuclear loss in myofibers undergoing atrophy is subject to debate. Myonuclear number is a determinant of transcriptional capacity, and therefore critical for muscle regeneration after atrophy. Objectives. Our objective was to investigate if and how myonuclear number is altered in the diaphragm of mechanically ventilated ICU patients. Methods. We used a combination of confocal microscopy, transcriptomics, and immunohistochemistry techniques to study myonuclear alterations in diaphragm and quadriceps biopsies from MV ICU patients. Measurements and Main Results. Patients with established diaphragm atrophy had a reduced myonuclear number and myonuclear domain. Intrinsic apoptotic pathway activation was identified as a potential mechanism underlying myonuclear removal in the diaphragm of mechanically ventilated ICU patients. Total transcription of myofibers decreased with myonuclear loss. Furthermore, muscle stem cell number was reduced in the patients with diaphragm atrophy.\u0000Conclusion. We identified myonuclear loss due to intrinsic apoptotic pathway activation as a mechanism underlying diaphragm atrophy in mechanically ventilated patients. The loss of myonuclei may contribute to difficult weaning due to impaired regrowth of myofibers after atrophy.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"69 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778101","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 : 2024-07-23DOI: 10.1101/2024.07.22.24310841
Ivor Popovich
Abstract Background: The discordance between the macrocirculation and microcirculation in septic shock has been recognised but never explained. I present a novel mathematical hypothesis as to how heterogenous microcirculatory flow distribution directly induces a hyperdynamic circulation and how elevated central venous pressure induces microcirculatory dysfunction. Methods: I explore the tube law and modified Poiseuille resistance for compliant blood vessels. Using these equations a new equation is developed incorporating time constants, elastance of the vessel, unstressed volume and wave reflections that demonstrates the relationship between volume of a microcirculatory vessel and total flow through it. Results: The relationship is demonstrated to be constant at zero until the unstressed volume is reached after which it increases exponentially. By considering n of these vessels in parallel, I demonstrate that the summed flow is minimised when flow is equally distributed among the n vessels, while it is maximised when all flow goes through one vessel alone, thereby demonstrating that heterogenous microvascular perfusion leads to increased total flow. It is shown that if conditions of wave reflection are right then a hyperdynamic circulation with high cardiac output develops. It is also demonstrated that high central venous pressure increases wave reflections and necessarily leads to microvascular perfusion heterogeneity if cardiac output is to be maintained. Conclusions: Microvascular impairment in septic shock directly leads to a hyperdynamic circulation with high cardiac output. High central venous pressures impair the microcirculation. Decades of clinical findings can now be explained mathematically. Implications for hemodynamic therapy for septic shock are discussed.
摘要 背景：脓毒性休克的大循环和微循环之间的不协调已得到公认，但从未得到解释。我提出了一个新的数学假说，说明微循环血流分布不均是如何直接诱发高动力循环的，以及中心静脉压升高是如何诱发微循环功能障碍的。方法：我探讨了顺应性血管的管子定律和修正的普瓦休伊阻力。利用这些方程，结合时间常数、血管弹性、非受压容积和波反射，建立了一个新的方程，证明了微循环血管的容积与通过该血管的总流量之间的关系。结果：结果表明，该关系恒定为零，直到达到非应力容积，之后呈指数增长。通过平行考虑 n 个这样的血管，我证明了当流量在 n 个血管中平均分配时，总流量最小，而当所有流量仅通过一个血管时，总流量最大，从而证明了异质微血管灌注会导致总流量增加。研究表明，如果波反射条件合适，就会形成高心输出量的超动力循环。研究还证明，如果要保持心输出量，高中心静脉压会增加波反射，必然导致微血管灌注异质性。结论：脓毒性休克的微血管损伤直接导致高心输出量的高动力循环。中心静脉压力过高会损害微循环。几十年的临床发现现在可以用数学来解释了。讨论了脓毒性休克血液动力学治疗的意义。
{"title":"Microvascular dysfunction induces a hyperdynamic circulation; a mathematical exploration","authors":"Ivor Popovich","doi":"10.1101/2024.07.22.24310841","DOIUrl":"https://doi.org/10.1101/2024.07.22.24310841","url":null,"abstract":"Abstract Background: The discordance between the macrocirculation and microcirculation in septic shock has been recognised but never explained. I present a novel mathematical hypothesis as to how heterogenous microcirculatory flow distribution directly induces a hyperdynamic circulation and how elevated central venous pressure induces microcirculatory dysfunction. Methods: I explore the tube law and modified Poiseuille resistance for compliant blood vessels. Using these equations a new equation is developed incorporating time constants, elastance of the vessel, unstressed volume and wave reflections that demonstrates the relationship between volume of a microcirculatory vessel and total flow through it. Results: The relationship is demonstrated to be constant at zero until the unstressed volume is reached after which it increases exponentially. By considering n of these vessels in parallel, I demonstrate that the summed flow is minimised when flow is equally distributed among the n vessels, while it is maximised when all flow goes through one vessel alone, thereby demonstrating that heterogenous microvascular perfusion leads to increased total flow. It is shown that if conditions of wave reflection are right then a hyperdynamic circulation with high cardiac output develops. It is also demonstrated that high central venous pressure increases wave reflections and necessarily leads to microvascular perfusion heterogeneity if cardiac output is to be maintained. Conclusions: Microvascular impairment in septic shock directly leads to a hyperdynamic circulation with high cardiac output. High central venous pressures impair the microcirculation. Decades of clinical findings can now be explained mathematically. Implications for hemodynamic therapy for septic shock are discussed.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"78 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778038","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 : 2024-07-11DOI: 10.1101/2024.07.11.24310285
Abigail Samuelsen, Parker Burrows, Erik Lehman, Anthony S Bonavia
Immunoparalysis is a significant concern in patients with sepsis and critical illness, potentially leading to increased risk of secondary infections. This study aimed to perform a longitudinal assessment of immune function over the initial two weeks following the onset of sepsis and critical illness. We compared ex vivo stimulated cytokine release to traditional markers of immunoparalysis, including monocyte Human Leukocyte Antigen (mHLA)-DR expression and absolute lymphocyte count (ALC). A total of 64 critically ill patients were recruited in a tertiary care academic medical setting, including 31 septic and 33 non-septic patients. Results showed that while mHLA-DR expression significantly increased over time, this was primarily driven by the non-septic subset of critically ill patients. ALC recovery was more prominent in septic patients. Ex vivo stimulation revealed significant increases in TNF and IL-6 production over time in septic patients. However, IFNg production varied with the stimulant used and did not show significant recovery when normalized to cell count. No significant correlation was found between mHLA-DR expression and other immunoparalysis biomarkers. These findings suggest the need for more nuanced immune monitoring approaches beyond the traditional 'sepsis' versus 'non-sepsis' classifications in critically ill patients. It also provided further evidence of a potential window for targeted immunotherapeutic interventions in the first week of critical illness.
{"title":"Time-Dependent Variation in Immunoparalysis Biomarkers Among Patients with Sepsis and Critical Illness","authors":"Abigail Samuelsen, Parker Burrows, Erik Lehman, Anthony S Bonavia","doi":"10.1101/2024.07.11.24310285","DOIUrl":"https://doi.org/10.1101/2024.07.11.24310285","url":null,"abstract":"Immunoparalysis is a significant concern in patients with sepsis and critical illness, potentially leading to increased risk of secondary infections. This study aimed to perform a longitudinal assessment of immune function over the initial two weeks following the onset of sepsis and critical illness. We compared ex vivo stimulated cytokine release to traditional markers of immunoparalysis, including monocyte Human Leukocyte Antigen (mHLA)-DR expression and absolute lymphocyte count (ALC). A total of 64 critically ill patients were recruited in a tertiary care academic medical setting, including 31 septic and 33 non-septic patients. Results showed that while mHLA-DR expression significantly increased over time, this was primarily driven by the non-septic subset of critically ill patients. ALC recovery was more prominent in septic patients. Ex vivo stimulation revealed significant increases in TNF and IL-6 production over time in septic patients. However, IFNg production varied with the stimulant used and did not show significant recovery when normalized to cell count. No significant correlation was found between mHLA-DR expression and other immunoparalysis biomarkers. These findings suggest the need for more nuanced immune monitoring approaches beyond the traditional 'sepsis' versus 'non-sepsis' classifications in critically ill patients. It also provided further evidence of a potential window for targeted immunotherapeutic interventions in the first week of critical illness.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"80 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141611936","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 : 2024-07-01DOI: 10.1101/2024.06.30.24309722
Phong Nguyen Thanh, Duc Hong Du, Ho Bich Hai, Nguyen Thanh Nguyen, Le Dinh Van Khoa, Le Thuy Thuy Khanh, Luu Hoai Bao Tran, Nguyen Thi My Linh, Cao Thi Cam Van, Dang Phuong Thao, Nguyen Thi Diem Trinh, Pham Tieu Kieu, Nguyen Thanh Truong, Vo Tan Hoang, Nguyen Thanh Ngoc, Tran Thi Dong Vien, Vo Trieu Ly, Tran Dang Khoa, Abi Beane, James T Anibal, Guy Thwaites, Ronald B Geskus, David Clifton, Nguyen Thi Phuong Dung, Evelyne Kestelyn, Guy Glover, Le Van Tan, Lam Minh Yen, Nguyen Le Nhu Tung, Nguyen Thanh Dung, C. Louise Thwaites
Objectives: We evaluated the efficacy and acceptability of awake-prone positioning (APP) in a randomised controlled trial, using a dedicated APP implementation team and wearable continuous-monitoring devices to monitor position and oximetry. Methods: The trial was performed at a tertiary level hospital in Ho Chi Minh City, Vietnam, recruiting adults (≥18 years) hospitalised with moderate or severe COVID-19 and receiving supplemental oxygen therapy via nasal/facemask systems or high-flow nasal canulae. Participants were randomized (1:1) to standard care or APP. The primary outcome was escalation of respiratory support within 28 days of randomisation. Results: Ninety-three patients were enrolled between March 2022 and March 2023; 80 (86%) had received ≥2 doses of SARS-CoV2 vaccine. Significantly greater mean daily APP times were achieved in those allocated to APP, although most did not achieve the target 8 hours/day. We did not detect significant differences in the primary outcome (RR 0.85, 95% CI 0.40-1.78, p=0.67) or secondary outcomes, including intubation rate and 28-day mortality. Particpants reported prone positioning was comfortable, although almost all preferred supine positioning. No adverse events associated with the intervention were reported. Conclusions: APP was not associated with benefit, but was safe. Continuous monitoring with wearable devices was feasible and acceptable to patients.
{"title":"Awake prone positioning effectiveness in moderate to severe COVID-19 a randomized controlled trial.","authors":"Phong Nguyen Thanh, Duc Hong Du, Ho Bich Hai, Nguyen Thanh Nguyen, Le Dinh Van Khoa, Le Thuy Thuy Khanh, Luu Hoai Bao Tran, Nguyen Thi My Linh, Cao Thi Cam Van, Dang Phuong Thao, Nguyen Thi Diem Trinh, Pham Tieu Kieu, Nguyen Thanh Truong, Vo Tan Hoang, Nguyen Thanh Ngoc, Tran Thi Dong Vien, Vo Trieu Ly, Tran Dang Khoa, Abi Beane, James T Anibal, Guy Thwaites, Ronald B Geskus, David Clifton, Nguyen Thi Phuong Dung, Evelyne Kestelyn, Guy Glover, Le Van Tan, Lam Minh Yen, Nguyen Le Nhu Tung, Nguyen Thanh Dung, C. Louise Thwaites","doi":"10.1101/2024.06.30.24309722","DOIUrl":"https://doi.org/10.1101/2024.06.30.24309722","url":null,"abstract":"Objectives: We evaluated the efficacy and acceptability of awake-prone positioning (APP) in a randomised controlled trial, using a dedicated APP implementation team and wearable continuous-monitoring devices to monitor position and oximetry.\u0000Methods: The trial was performed at a tertiary level hospital in Ho Chi Minh City, Vietnam, recruiting adults (≥18 years) hospitalised with moderate or severe COVID-19 and receiving supplemental oxygen therapy via nasal/facemask systems or high-flow nasal canulae. Participants were randomized (1:1) to standard care or APP. The primary outcome was escalation of respiratory support within 28 days of randomisation.\u0000Results: Ninety-three patients were enrolled between March 2022 and March 2023; 80 (86%) had received ≥2 doses of SARS-CoV2 vaccine. Significantly greater mean daily APP times were achieved in those allocated to APP, although most did not achieve the target 8 hours/day. We did not detect significant differences in the primary outcome (RR 0.85, 95% CI 0.40-1.78, p=0.67) or secondary outcomes, including intubation rate and 28-day mortality. Particpants reported prone positioning was comfortable, although almost all preferred supine positioning. No adverse events associated with the intervention were reported.\u0000Conclusions: APP was not associated with benefit, but was safe. Continuous monitoring with wearable devices was feasible and acceptable to patients.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500656","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 : 2024-06-30DOI: 10.1101/2024.06.28.24309547
Samuel W Fenske, Alec Peltekian, Mengija Kang, Nikolay S Markov, Mengou Zhu, Kevin Grudzinski, Melissa J Bak, Anna Pawlowski, Vishu Gupta, Yuwei Mao, Stanislav Bratchikov, Thomas Stoeger, Luke V Rasmussen, Alok N Choudhary, Alexander V Misharin, Benjamin D Singer, GR Scott Budinger, Richard D Wunderink, Ankit Agrawal, Catherine A Gao, NU Script Study Investigators
Background: Criteria to identify patients who are ready to be liberated from mechanical ventilation are imprecise, often resulting in prolonged mechanical ventilation or reintubation, both of which are associated with adverse outcomes. Daily protocol-driven assessment of the need for mechanical ventilation leads to earlier extubation but requires dedicated personnel. We sought to determine whether machine learning applied to the electronic health record could predict successful extubation. Methods: We examined 37 clinical features from patients from a single-center prospective cohort study of patients in our quaternary care medical ICU who required mechanical ventilation and underwent a bronchoalveolar lavage for known or suspected pneumonia. We also tested our models on an external test set from a community hospital ICU in our health care system. We curated electronic health record data aggregated from midnight to 8AM and labeled extubation status. We deployed three data encoding/imputation strategies and built XGBoost, LightGBM, logistic regression, LSTM, and RNN models to predict successful next-day extubation. We evaluated each model's performance using Area Under the Receiver Operating Characteristic (AUROC), Area Under the Precision Recall Curve (AUPRC), Sensitivity (Recall), Specificity, PPV (Precision), Accuracy, and F1-Score. Results: Our internal cohort included 696 patients and 9,828 ICU days, and our external cohort had 333 patients and 2,835 ICU days. The best model (LSTM) predicted successful extubation on a given ICU day with an AUROC 0.87 (95% CI 0.834- 0.902) and the internal test set and 0.87 (95% CI 0.848-0.885) on the external test set. A Logistic Regression model performed similarly (AUROC 0.86 internal test, 0.83 external test). Across multiple model types, measures previously demonstrated to be important in determining readiness for extubation were found to be most informative, including plateau pressure and Richmond Agitation Sedation Scale (RASS) score. Our model often predicted patients to be stable for extubation in the days preceding their actual extubation, with 63.8% of predicted extubations occurring within three days of true extubation. We also tested the best model on cases of failed extubations (requiring reintubation within two days) not seen by the model during training. Our best model would have identified 35.4% (17/48) of these cases in the internal test set and 48.1% (13/27) cases in the external test set as unlikely to be successfully extubated. Conclusions: Machine learning models can accurately predict the likelihood of extubation on a given ICU day from data available in the electronic health record. Predictions from these models are driven by clinical features that have been associated with successful extubation in clinical trials.
{"title":"Developing and validating a machine learning model to predict successful next-day extubation in the ICU","authors":"Samuel W Fenske, Alec Peltekian, Mengija Kang, Nikolay S Markov, Mengou Zhu, Kevin Grudzinski, Melissa J Bak, Anna Pawlowski, Vishu Gupta, Yuwei Mao, Stanislav Bratchikov, Thomas Stoeger, Luke V Rasmussen, Alok N Choudhary, Alexander V Misharin, Benjamin D Singer, GR Scott Budinger, Richard D Wunderink, Ankit Agrawal, Catherine A Gao, NU Script Study Investigators","doi":"10.1101/2024.06.28.24309547","DOIUrl":"https://doi.org/10.1101/2024.06.28.24309547","url":null,"abstract":"Background: Criteria to identify patients who are ready to be liberated from mechanical ventilation are imprecise, often\u0000resulting in prolonged mechanical ventilation or reintubation, both of which are associated with adverse outcomes. Daily\u0000protocol-driven assessment of the need for mechanical ventilation leads to earlier extubation but requires dedicated\u0000personnel. We sought to determine whether machine learning applied to the electronic health record could predict\u0000successful extubation.\u0000Methods: We examined 37 clinical features from patients from a single-center prospective cohort study of patients in our\u0000quaternary care medical ICU who required mechanical ventilation and underwent a bronchoalveolar lavage for known or\u0000suspected pneumonia. We also tested our models on an external test set from a community hospital ICU in our health care\u0000system. We curated electronic health record data aggregated from midnight to 8AM and labeled extubation status. We\u0000deployed three data encoding/imputation strategies and built XGBoost, LightGBM, logistic regression, LSTM, and RNN\u0000models to predict successful next-day extubation. We evaluated each model's performance using Area Under the Receiver\u0000Operating Characteristic (AUROC), Area Under the Precision Recall Curve (AUPRC), Sensitivity (Recall), Specificity, PPV\u0000(Precision), Accuracy, and F1-Score.\u0000Results: Our internal cohort included 696 patients and 9,828 ICU days, and our external cohort had 333 patients and 2,835\u0000ICU days. The best model (LSTM) predicted successful extubation on a given ICU day with an AUROC 0.87 (95% CI 0.834-\u00000.902) and the internal test set and 0.87 (95% CI 0.848-0.885) on the external test set. A Logistic Regression model\u0000performed similarly (AUROC 0.86 internal test, 0.83 external test). Across multiple model types, measures previously\u0000demonstrated to be important in determining readiness for extubation were found to be most informative, including plateau\u0000pressure and Richmond Agitation Sedation Scale (RASS) score. Our model often predicted patients to be stable for\u0000extubation in the days preceding their actual extubation, with 63.8% of predicted extubations occurring within three days of\u0000true extubation. We also tested the best model on cases of failed extubations (requiring reintubation within two days) not\u0000seen by the model during training. Our best model would have identified 35.4% (17/48) of these cases in the internal test\u0000set and 48.1% (13/27) cases in the external test set as unlikely to be successfully extubated.\u0000Conclusions: Machine learning models can accurately predict the likelihood of extubation on a given ICU day from data\u0000available in the electronic health record. Predictions from these models are driven by clinical features that have been\u0000associated with successful extubation in clinical trials.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"13 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141524683","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 : 2024-06-20DOI: 10.1101/2024.06.19.24309166
Yu Ma, Azadeh Tabari, Jesus Alfonso Juarez Palazuelos, Anthony Gebran, Haytham Kaafarani, Dimitris Bertsimas, Dania Daye
Introduction: Endovenous thermal ablation (EVTA) stands as one of the primary treatments for superficial venous insufficiency. Concern exists about the potential for thromboembolic complications following this procedure. Although rare, those complications can be severe, necessitating early identification of patients prone to increased thrombotic risks. This study aims to leverage AI-based algorithms to forecast patients' likelihood of developing deep vein thrombosis (DVT) within 30 days following EVTA. Materials and Methods: From 2007 to 2017, all patients who underwent EVTA were identified using the American College of Surgeons National Surgical Quality Improvement Program database. We developed and validated 4 machine learning models using demographics, comorbidities, and laboratory values to predict the risk of postoperative deep vein thrombosis: Classification and Regression Trees (CART), Optimal Classification Trees (OCT), Random Forests, and Extreme Gradient Boosting (XGBoost). The models were trained using all the available variables. SHAP analysis was adopted to interpret model outcomes and offer medical insights into feature importance and interactions. Results: A total of 21,549 patients were included (mean age of 54 +- SD years, 67% female). In this cohort, 1.59% developed DVT. The XGBoost model had good discriminative power for predicting DVT risk with AUC of 0.711 in the hold-out test set for all-variable model. Stratification of the test set by age, BMI, preoperative white blood cell and platelet count shows that the model performs equally well across these groups. Conclusion: We developed and validated an interpretable model that enables physicians to predict which patients with superficial venous insufficiency has higher risk of developing deep vein thrombosis within 30 days following endovenous thermal ablation.
{"title":"An artificial-intelligence interpretable tool to predict risk of deep vein thrombosis after endovenous thermal ablation","authors":"Yu Ma, Azadeh Tabari, Jesus Alfonso Juarez Palazuelos, Anthony Gebran, Haytham Kaafarani, Dimitris Bertsimas, Dania Daye","doi":"10.1101/2024.06.19.24309166","DOIUrl":"https://doi.org/10.1101/2024.06.19.24309166","url":null,"abstract":"Introduction: Endovenous thermal ablation (EVTA) stands as one of the primary treatments for superficial venous insufficiency. Concern exists about the potential for thromboembolic complications following this procedure. Although rare, those complications can be severe, necessitating early identification of patients prone to increased thrombotic risks. This study aims to leverage AI-based algorithms to forecast patients' likelihood of developing deep vein thrombosis (DVT) within 30 days following EVTA.\u0000Materials and Methods: From 2007 to 2017, all patients who underwent EVTA were identified using the American College of Surgeons National Surgical Quality Improvement Program database. We developed and validated 4 machine learning models using demographics, comorbidities, and laboratory values to predict the risk of postoperative deep vein thrombosis: Classification and Regression Trees (CART), Optimal Classification Trees (OCT), Random Forests, and Extreme Gradient Boosting (XGBoost). The models were trained using all the available variables. SHAP analysis was adopted to interpret model outcomes and offer medical insights into feature importance and interactions.\u0000Results: A total of 21,549 patients were included (mean age of 54 +- SD years, 67% female). In this cohort, 1.59% developed DVT. The XGBoost model had good discriminative power for predicting DVT risk with AUC of 0.711 in the hold-out test set for all-variable model. Stratification of the test set by age, BMI, preoperative white blood cell and platelet count shows that the model performs equally well across these groups. Conclusion: We developed and validated an interpretable model that enables physicians to predict which patients with superficial venous insufficiency has higher risk of developing deep vein thrombosis within 30 days following endovenous thermal ablation.","PeriodicalId":501249,"journal":{"name":"medRxiv - Intensive Care and Critical Care Medicine","volume":"16 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500780","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：