scholarly journals Dr. Agent: Clinical predictive model via mimicked second opinions

2020 ◽  
Vol 27 (7) ◽  
pp. 1084-1091
Author(s):  
Junyi Gao ◽  
Cao Xiao ◽  
Lucas M Glass ◽  
Jimeng Sun
Keyword(s):  
Health Status ◽  
Complex Disease ◽  
Medical Information ◽  
Prediction Models ◽  
Relevant Information ◽  
Disease Diagnosis ◽  
Mortality Prediction ◽  
Clinical Prediction ◽  
Phenotype Classification ◽  
Clinical Predictive Model

Abstract Objective Prediction of disease phenotypes and their outcomes is a difficult task. In practice, patients routinely seek second opinions from multiple clinical experts for complex disease diagnosis. Our objective is to mimic such a practice of seeking second opinions by training 2 agents with different focuses: the primary agent studies the most recent visit of the patient to learn the current health status, and then the second-opinion agent considers the entire patient history to obtain a more global view. Materials and Methods Our approach Dr. Agent augments recurrent neural networks with 2 policy gradient agents. Moreover, Dr. Agent is customized with various patient demographics information and learns a dynamic skip connection to focus on the relevant information over time. We trained Dr. Agent to perform 4 clinical prediction tasks on the publicly available MIMIC-III (Medical Information Mart for Intensive Care) database: (1) in-hospital mortality prediction, (2) acute care phenotype classification, (3) physiologic decompensation prediction, and (4) forecasting length of stay. We compared the performance of Dr. Agent against 4 baseline clinical predictive models. Results Dr. Agent outperforms baseline clinical prediction models across all 4 tasks in terms of all metrics. Compared with the best baseline model, Dr. Agent achieves up to 15% higher area under the precision-recall curve on different tasks. Conclusions Dr. Agent can comprehensively model the long-term dependencies of patients’ health status while considering patients’ demographics using 2 agents, and therefore achieves better prediction performance on different clinical prediction tasks.

scholarly journals MIMIC-IF: Interpretability and Fairness Evaluation of Deep Learning Models on MIMIC-IV Dataset

2021 ◽  
Author(s):  
Chuizheng Meng ◽  
Loc Trinh ◽  
Nan Xu ◽  
Yan Liu
Keyword(s):  
Deep Learning ◽  
Large Scale ◽  
Medical Information ◽  
Prediction Models ◽  
Short Term Memory ◽  
Mortality Prediction ◽  
Learning Models ◽  
Healthcare Applications ◽  
Disparate Treatment ◽  
Mortality Predictors

Abstract The recent release of large-scale healthcare datasets has greatly propelled the research of data-driven deep learning models for healthcare applications. However, due to the nature of such deep black-boxed models, concerns about interpretability, fairness, and biases in healthcare scenarios where human lives are at stake call for a careful and thorough examination of both datasets and models. In this work, we focus on MIMIC-IV (Medical Information Mart for Intensive Care, version IV), the largest publicly available healthcare dataset, and conduct comprehensive analyses of dataset representation bias as well as interpretability and prediction fairness of deep learning models for in-hospital mortality prediction. In terms of interpretability, we observe that (1) the best-performing interpretability method successfully identifies critical features for mortality prediction on various prediction models; (2) demographic features are important for prediction. In terms of fairness, we observe that (1) there exists disparate treatment in prescribing mechanical ventilation among patient groups across ethnicity, gender and age; (2) all of the studied mortality predictors are generally fair while the IMV-LSTM (Interpretable Multi-Variable Long Short-Term Memory) model provides the most accurate and unbiased predictions across all protected groups. We further draw concrete connections between interpretability methods and fairness metrics by showing how feature importance from interpretability methods can be beneficial in quantifying potential disparities in mortality predictors.

Risk Stratification in Acute Pulmonary Embolism: The Latest Algorithms

2021 ◽  
Vol 42 (02) ◽  
pp. 183-198
Author(s):  
Georgios A. Triantafyllou ◽  
Oisin O'Corragain ◽  
Belinda Rivera-Lebron ◽  
Parth Rali
Keyword(s):  
Pulmonary Embolism ◽  
Risk Stratification ◽  
Prediction Models ◽  
Bleeding Risk ◽  
Hemodynamic Instability ◽  
Reperfusion Therapy ◽  
Mortality Prediction ◽  
Risk Scores ◽  
Short Term Mortality ◽  
Urgent Patient

AbstractPulmonary embolism (PE) is a common clinical entity, which most clinicians will encounter. Appropriate risk stratification of patients is key to identify those who may benefit from reperfusion therapy. The first step in risk assessment should be the identification of hemodynamic instability and, if present, urgent patient consideration for systemic thrombolytics. In the absence of shock, there is a plethora of imaging studies, biochemical markers, and clinical scores that can be used to further assess the patients' short-term mortality risk. Integrated prediction models incorporate more information toward an individualized and precise mortality prediction. Additionally, bleeding risk scores should be utilized prior to initiation of anticoagulation and/or reperfusion therapy administration. Here, we review the latest algorithms for a comprehensive risk stratification of the patient with acute PE.

scholarly journals Developing and validating COVID-19 adverse outcome risk prediction models from a bi-national European cohort of 5594 patients

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Espen Jimenez-Solem ◽  
Tonny S. Petersen ◽  
Casper Hansen ◽  
Christian Hansen ◽  
Christina Lioma ◽  
...  
Keyword(s):  
Prediction Models ◽  
Severe Disease ◽  
External Validation ◽  
Area Under The Curve ◽  
Predictive Performance ◽  
Mortality Prediction ◽  
Icu Admission ◽  
Risk Of Death ◽  
Online Risk ◽  
The United Kingdom

AbstractPatients with severe COVID-19 have overwhelmed healthcare systems worldwide. We hypothesized that machine learning (ML) models could be used to predict risks at different stages of management and thereby provide insights into drivers and prognostic markers of disease progression and death. From a cohort of approx. 2.6 million citizens in Denmark, SARS-CoV-2 PCR tests were performed on subjects suspected for COVID-19 disease; 3944 cases had at least one positive test and were subjected to further analysis. SARS-CoV-2 positive cases from the United Kingdom Biobank was used for external validation. The ML models predicted the risk of death (Receiver Operation Characteristics—Area Under the Curve, ROC-AUC) of 0.906 at diagnosis, 0.818, at hospital admission and 0.721 at Intensive Care Unit (ICU) admission. Similar metrics were achieved for predicted risks of hospital and ICU admission and use of mechanical ventilation. Common risk factors, included age, body mass index and hypertension, although the top risk features shifted towards markers of shock and organ dysfunction in ICU patients. The external validation indicated fair predictive performance for mortality prediction, but suboptimal performance for predicting ICU admission. ML may be used to identify drivers of progression to more severe disease and for prognostication patients in patients with COVID-19. We provide access to an online risk calculator based on these findings.

scholarly journals High mortality rate of obstetric critically ill women in Rwanda and its predictability

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Alcade Rudakemwa ◽  
Amyl Lucille Cassidy ◽  
Théogène Twagirumugabe
Keyword(s):  
Prediction Models ◽  
Average Length ◽  
Mortality Prediction ◽  
Routine Practice ◽  
Icu Mortality ◽  
Resource Limited ◽  
Obstetric Haemorrhage ◽  
Tertiary Hospitals ◽  
Mortality Prediction Models ◽  
Qsofa Score

Abstract Background Reasons for admission to intensive care units (ICUs) for obstetric patients vary from one setting to another. Outcomes from ICU and prediction models are not well explored in Rwanda owing to lack of appropriate scores. This study aimed to assess reasons for admission and accuracy of prediction models for mortality of obstetric patients admitted to ICUs of two public tertiary hospitals in Rwanda. Methods We prospectively collected data from all obstetric patients admitted to the ICUs of the two public tertiary hospitals in Rwanda from March 2017 to February 2018 to identify reasons for admission, demographic and clinical characteristics, outcome including death and its predictability by both the Modified Early Obstetric Warning Score (MEOWS) and quick Sequential Organ Failure Assessment (qSOFA). We analysed the accuracy of mortality prediction models by MEOWS or qSOFA by using logistic regression adjusting for factors associated with mortality. Area under the Receiver Operating characteristic (AUROC) curves is used to show the predicting capacity for each individual tool. Results Obstetric patients (n = 94) represented 12.8 % of all 747 ICU admissions which is 1.8 % of all 4.999 admitted women for pregnancy or labor. Sepsis (n = 30; 31.9 %) and obstetric haemorrhage (n = 24; 25.5 %) were the two commonest reasons for ICU admission. Overall ICU mortality for obstetric patients was 54.3 % (n = 51) with average length of stay of 6.6 ± 7.525 days. MEOWS score was an independent predictor of mortality (adjusted (a)OR 1.25; 95 % CI 1.07–1.46) and so was qSOFA score (aOR 2.81; 95 % CI 1.25–6.30) with an adjusted AUROC of 0.773 (95 % CI 0.67–0.88) and 0.764 (95 % CI 0.65–0.87), indicating fair accuracy for ICU mortality prediction in these settings of both MEOWS and qSOFA scores. Conclusions Sepsis and obstetric haemorrhage were the commonest reasons for obstetric admissions to ICU in Rwanda. MEOWS and qSOFA scores could accurately predict ICU mortality of obstetric patients in resource-limited settings, but larger studies are needed before a recommendation for their use in routine practice in similar settings.

Performance of the Surprise Question Compared to Prediction Models in Hemodialysis Patients: A Prospective Study

2017 ◽  
Vol 46 (5) ◽  
pp. 390-396 ◽  
Author(s):  
Rakesh Malhotra ◽  
Xia Tao ◽  
Yuedong Wang ◽  
Yuqi Chen ◽  
Rebecca H. Apruzzese ◽  
...  
Keyword(s):  
Linear Models ◽  
Prediction Models ◽  
Clinical Laboratory ◽  
Mortality Prediction ◽  
Operating Characteristics ◽  
Dialysis Treatment ◽  
Mortality Probability ◽  
A Prospective Study ◽  
Surprise Question

Background: The surprise question (SQ) (“Would you be surprised if this patient were still alive in 6 or 12 months?”) is used as a mortality prognostication tool in hemodialysis (HD) patients. We compared the performance of the SQ with that of prediction models (PMs) for 6- and 12-month mortality prediction. Methods: Demographic, clinical, laboratory, and dialysis treatment indicators were used to model 6- and 12-month mortality probability in a HD patients training cohort (n = 6,633) using generalized linear models (GLMs). A total of 10 nephrologists from 5 HD clinics responded to the SQ in 215 patients followed prospectively for 12 months. The performance of PM was evaluated in the validation (n = 6,634) and SQ cohorts (n = 215) using the areas under receiver operating characteristics curves. We compared sensitivities and specificities of PM and SQ. Results: The PM and SQ cohorts comprised 13,267 (mean age 61 years, 55% men, 54% whites) and 215 (mean age 62 years, 59% men, 50% whites) patients, respectively. During the 12-month follow-up, 1,313 patients died in the prediction model cohort and 22 in the SQ cohort. For 6-month mortality prediction, the GLM had areas under the curve of 0.77 in the validation cohort and 0.77 in the SQ cohort. As for 12-month mortality, areas under the curve were 0.77 and 0.80 in the validation and SQ cohorts, respectively. The 6- and 12-month PMs had sensitivities of 0.62 (95% CI 0.35–0.88) and 0.75 (95% CI 0.56–0.94), respectively. The 6- and 12-month SQ sensitivities were 0.23 (95% CI 0.002–0.46) and 0.35 (95% CI 0.14–0.56), respectively. Conclusion: PMs exhibit superior sensitivity compared to the SQ for mortality prognostication in HD patients.

scholarly journals Building prediction models for coronary heart disease by synthesizing multiple longitudinal research findings

2005 ◽  
Vol 12 (5) ◽  
pp. 459-464 ◽  
Author(s):  
Guizhou Hu ◽  
Martin M. Root
Keyword(s):  
Coronary Heart Disease ◽  
Heart Disease ◽  
Prediction Model ◽  
Empirical Model ◽  
Complex Disease ◽  
Prediction Models ◽  
Longitudinal Research ◽  
Study Data ◽  
Individual Risk ◽  
Data Set

Background No methodology is currently available to allow the combining of individual risk factor information derived from different longitudinal studies for a chronic disease in a multivariate fashion. This paper introduces such a methodology, named Synthesis Analysis, which is essentially a multivariate meta-analytic technique. Design The construction and validation of statistical models using available data sets. Methods and results Two analyses are presented. (1) With the same data, Synthesis Analysis produced a similar prediction model to the conventional regression approach when using the same risk variables. Synthesis Analysis produced better prediction models when additional risk variables were added. (2) A four-variable empirical logistic model for death from coronary heart disease was developed with data from the Framingham Heart Study. A synthesized prediction model with five new variables added to this empirical model was developed using Synthesis Analysis and literature information. This model was then compared with the four-variable empirical model using the first National Health and Nutrition Examination Survey (NHANES I) Epidemiologic Follow-up Study data set. The synthesized model had significantly improved predictive power ( x2 = 43.8, P < 0.00001). Conclusions Synthesis Analysis provides a new means of developing complex disease predictive models from the medical literature.

Sign in / Sign up

Export Citation Format

Share Document