Severity and Outcome Detection for the Cancer Patients with COVID-19 Using Machine Learning Models

Abstract Background:The COVID-19 pandemic challenges the healthcare system to provide enough resources to battle the pandemic without jeopardizing routine treatments. As a result, this is important that we can predict the outcomes of patients at the time of admission. This study aims to apply different machine learning (ML) models for predicting Intensive Care Unit (ICU) admission and mortality of Cancer Patients infected with COVID-19.Methods:This study's data were collected from a referral cancer center in Iran. The study included all patients with cancer and a confirmed diagnosis of COVID-19.Different ML prediction algorithms like Logistic Regression (LR), Naïve Bayes (NB), k-Nearest Neighbours (kNN), Random Forest (RF), and Support Vector Machine (SVM) were used. Also, we applied the SelectKBest method to find the most important features for predicting ICU admission and mortality.Results:Three hundred thirty-nine patients enrolled in the study. One hundred fifteen were admitted to the Intensive Care Unit (ICU), and 118 patients died during the hospital admission. The Area Under Curve (AUC) for predicting mortality is 0.61 for LR, 0.74 for NB, 0.61 for kNN, 0.6 for SVM, and 0.79 for RF. The AUC for predicting ICU admission is 0.61 for LR, 0.74 for NB, 0.56 for kNN, 0.55 for SVM, and 0.7 for RF.C-reactive protein (CRP), Aspartate transaminase (AST), and Neutrophil-Lymphocyte Ratio (NLR) also are the most common features in predicting ICU admission and mortality.Conclusion:Our findings show the promise of different AI methods for predicting the risk of death or ICU in cancer patients infected with COVID-19, highlighting the importance of first laboratory results and patients' symptoms.

Download Full-text

Early Prediction of Seven-Day Mortality in Intensive Care Unit Using a Machine Learning Model: Results from the SPIN-UTI Project

Journal of Clinical Medicine ◽

10.3390/jcm10050992 ◽

2021 ◽

Vol 10 (5) ◽

pp. 992

Author(s):

Martina Barchitta ◽

Andrea Maugeri ◽

Giuliana Favara ◽

Paolo Marco Riela ◽

Giovanni Gallo ◽

...

Keyword(s):

Machine Learning ◽

Intensive Care ◽

Intensive Care Units ◽

Learning Algorithm ◽

Area Under The Curve ◽

Support Vector ◽

Icu Admission ◽

Risk Of Death ◽

Saps Ii ◽

Svm Algorithm

Patients in intensive care units (ICUs) were at higher risk of worsen prognosis and mortality. Here, we aimed to evaluate the ability of the Simplified Acute Physiology Score (SAPS II) to predict the risk of 7-day mortality, and to test a machine learning algorithm which combines the SAPS II with additional patients’ characteristics at ICU admission. We used data from the “Italian Nosocomial Infections Surveillance in Intensive Care Units” network. Support Vector Machines (SVM) algorithm was used to classify 3782 patients according to sex, patient’s origin, type of ICU admission, non-surgical treatment for acute coronary disease, surgical intervention, SAPS II, presence of invasive devices, trauma, impaired immunity, antibiotic therapy and onset of HAI. The accuracy of SAPS II for predicting patients who died from those who did not was 69.3%, with an Area Under the Curve (AUC) of 0.678. Using the SVM algorithm, instead, we achieved an accuracy of 83.5% and AUC of 0.896. Notably, SAPS II was the variable that weighted more on the model and its removal resulted in an AUC of 0.653 and an accuracy of 68.4%. Overall, these findings suggest the present SVM model as a useful tool to early predict patients at higher risk of death at ICU admission.

Download Full-text

Outcome of cancer patients considered for intensive care unit admission in two university hospitals in the Netherlands: the danger of delayed ICU admissions and off-hour triage decisions

Annals of Intensive Care ◽

10.1186/s13613-021-00898-2 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Esther N. van der Zee ◽

Dominique D. Benoit ◽

Marinus Hazenbroek ◽

Jan Bakker ◽

Erwin J. O. Kompanje ◽

...

Keyword(s):

Intensive Care Unit ◽

Intensive Care ◽

The Netherlands ◽

Cancer Patients ◽

Admission Policy ◽

P Value ◽

University Hospitals ◽

Icu Admission ◽

Risk Of Death ◽

Triage Decisions

Abstract Background Very few studies assessed the association between Intensive Care Unit (ICU) triage decisions and mortality. The aim of this study was to assess whether an association could be found between 30-day mortality, and ICU admission consultation conditions and triage decisions. Methods We conducted a retrospective cohort study in two large referral university hospitals in the Netherlands. We identified all adult cancer patients for whom ICU admission was requested from 2016 to 2019. Via a multivariable logistic regression analysis, we assessed the association between 30-day mortality, and ICU admission consultation conditions and triage decisions. Results Of the 780 cancer patients for whom ICU admission was requested, 332 patients (42.6%) were considered ‘too well to benefit’ from ICU admission, 382 (49%) patients were immediately admitted to the ICU and 66 patients (8.4%) were considered ‘too sick to benefit’ according to the consulting intensivist(s). The 30-day mortality in these subgroups was 30.1%, 36.9% and 81.8%, respectively. In the patient group considered ‘too well to benefit’, 258 patients were never admitted to the ICU and 74 patients (9.5% of the overall study population, 22.3% of the patients ‘too well to benefit’) were admitted to the ICU after a second ICU admission request (delayed ICU admission). Thirty-day mortality in these groups was 25.6% and 45.9%. After adjustment for confounders, ICU consultations during off-hours (OR 1.61, 95% CI 1.09–2.38, p-value 0.02) and delayed ICU admission (OR 1.83, 95% CI 1.00–3.33, p-value 0.048 compared to “ICU admission”) were independently associated with 30-day mortality. Conclusion The ICU denial rate in our study was high (51%). Sixty percent of the ICU triage decisions in cancer patients were made during off-hours, and 22.3% of the patients initially considered “too well to benefit” from ICU admission were subsequently admitted to the ICU. Both decisions during off-hours and a delayed ICU admission were associated with an increased risk of death at 30 days. Our study suggests that in cancer patients, ICU triage decisions should be discussed during on-hours, and ICU admission policy should be broadened, with a lower admission threshold for critically ill cancer patients.

Download Full-text

Predictors of survival in patients with gastrointestinal malignancies admitted to the intensive care unit.

Journal of Clinical Oncology ◽

10.1200/jco.2017.35.15_suppl.e18050 ◽

2017 ◽

Vol 35 (15_suppl) ◽

pp. e18050-e18050

Author(s):

Heidi Chwan Ko ◽

Melissa Yan ◽

Rohan Gupta ◽

Juhee Song ◽

Kayla Kebbel ◽

...

Keyword(s):

Intensive Care Unit ◽

Intensive Care ◽

Hospital Mortality ◽

Cancer Patients ◽

Sofa Score ◽

Icu Mortality ◽

Gastrointestinal Malignancies ◽

Icu Admission ◽

Risk Of Death ◽

Gi Cancer

e18050 Background: Cancer patients have a high use of healthcare utilization at the end of life which can frequently involve admissions to the intensive care unit (ICU). We sought to evaluate the predictors for outcome in gastrointestinal (GI) cancer patients admitted to the ICU for non-surgical conditions. Methods: The objective of this study was to determine the factors associated with ICU mortality, hospital mortality and overall survival (OS). A total of 200 patients with GI cancer admitted to the ICU at The University of Texas MD Anderson Cancer Center between November 2012 and February 2015 were retrospectively analyzed. Cancer characteristics, treatment characteristics, and Sequential Organ Failure Assessment (SOFA) scores defining severity based on 6 organ systems with scores ranging from 0 to 24 were analyzed for their effects on survival endpoints using multivariate logistic regression models and a multivariate Cox proportional hazards regression model. Results: The characteristics of the 200 patients were: 64.5% male, mean age of 60 years, median admission SOFA score of 6.0, and tumor types of primary intestinal (37.5%), hepatobiliary/pancreatic (36%), and gastroesophageal (GE) (24%). The ICU mortality was 26%, hospital mortality was 41%, and 6-month OS estimate was 25%. In multivariate analysis, ICU admission SOFA score > 10 (odds ratio (OR) 17.1, p < 0.0001), poorly differentiated tumor grade (OR 3.2, p = 0.02), and shorter duration of metastatic disease (OR 2.3, p = 0.07) were associated with increased odds of ICU mortality. These same variables were associated with increased odds of hospital mortality. In multivariate OS analysis, SOFA score 6-10 (hazard ratio (HR) 2.1, p = 0.0006) and SOFA score > 10 (HR 4.4, p < 0.0001), patients with GE primary (HR 2.2, p = 0.002) and patients with a poor outpatient performance status that precluded active chemotherapy (HR 2.2, p = 0.01) were associated with increased risk of death. Conclusions: The SOFA score was the most predictive factor for ICU mortality, hospital mortality, and OS for GI cancer patients admitted to the ICU. It should be utilized in all GI cancer patients upon ICU admission to improve both acute and longer-term prognostication.

Download Full-text

Development of a machine learning model for predicting pediatric mortality in the early stages of intensive care unit admission

Scientific Reports ◽

10.1038/s41598-020-80474-z ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Bongjin Lee ◽

Kyunghoon Kim ◽

Hyejin Hwang ◽

You Sun Kim ◽

Eun Hee Chung ◽

...

Keyword(s):

Machine Learning ◽

Intensive Care Unit ◽

Intensive Care ◽

Validation Cohort ◽

External Validation ◽

Learning Model ◽

Derivation Cohort ◽

Icu Admission ◽

Early Stages ◽

Machine Learning Model

AbstractThe aim of this study was to develop a predictive model of pediatric mortality in the early stages of intensive care unit (ICU) admission using machine learning. Patients less than 18 years old who were admitted to ICUs at four tertiary referral hospitals were enrolled. Three hospitals were designated as the derivation cohort for machine learning model development and internal validation, and the other hospital was designated as the validation cohort for external validation. We developed a random forest (RF) model that predicts pediatric mortality within 72 h of ICU admission, evaluated its performance, and compared it with the Pediatric Index of Mortality 3 (PIM 3). The area under the receiver operating characteristic curve (AUROC) of RF model was 0.942 (95% confidence interval [CI] = 0.912–0.972) in the derivation cohort and 0.906 (95% CI = 0.900–0.912) in the validation cohort. In contrast, the AUROC of PIM 3 was 0.892 (95% CI = 0.878–0.906) in the derivation cohort and 0.845 (95% CI = 0.817–0.873) in the validation cohort. The RF model in our study showed improved predictive performance in terms of both internal and external validation and was superior even when compared to PIM 3.

Download Full-text

Development and Validation of Unplanned Extubation Prediction Models Using Intensive Care Unit Data: Comparative Machine Learning Study (Preprint)

10.2196/preprints.23508 ◽

2020 ◽

Author(s):

Sujeong Hur ◽

Ji Young Min ◽

Junsang Yoo ◽

Kyunga Kim ◽

Chi Ryang Chung ◽

...

Keyword(s):

Machine Learning ◽

Intensive Care Unit ◽

Patient Safety ◽

Intensive Care ◽

Predictive Value ◽

Prediction Models ◽

Support Vector ◽

Unplanned Extubation ◽

Electronic Health Record Data ◽

Icu Patients

BACKGROUND Patient safety in the intensive care unit (ICU) is one of the most critical issues, and unplanned extubation (UE) is considered as the most adverse event for patient safety. Prevention and early detection of such an event is an essential but difficult component of quality care. OBJECTIVE This study aimed to develop and validate prediction models for UE in ICU patients using machine learning. METHODS This study was conducted an academic tertiary hospital in Seoul. The hospital had approximately 2,000 inpatient beds and 120 intensive care unit (ICU) beds. The number of patients, on daily basis, was approximately 9,000 for the out-patient. The number of annual ICU admission was approximately 10,000. We conducted a retrospective study between January 1, 2010 and December 31, 2018. A total of 6,914 extubation cases were included. We developed an unplanned extubation prediction model using machine learning algorithms, which included random forest (RF), logistic regression (LR), artificial neural network (ANN), and support vector machine (SVM). For evaluating the model’s performance, we used area under the receiver operator characteristic curve (AUROC). Sensitivity, specificity, positive predictive value negative predictive value, and F1-score were also determined for each model. For performance evaluation, we also used calibration curve, the Brier score, and the Hosmer-Lemeshow goodness-of-fit statistic. RESULTS Among the 6,914 extubation cases, 248 underwent UE. In the UE group, there were more males than females, higher use of physical restraints, and fewer surgeries. The incidence of UE was more likely to occur during the night shift compared to the planned extubation group. The rate of reintubation within 24 hours and hospital mortality was higher in the UE group. The UE prediction algorithm was developed, and the AUROC for RF was 0.787, for LR was 0.762, for ANN was 0.762, and for SVM was 0.740. CONCLUSIONS We successfully developed and validated machine learning-based prediction models to predict UE in ICU patients using electronic health record data. The best AUROC was 0.787, which was obtained using RF. CLINICALTRIAL N/A

Download Full-text

Selection criteria for intensive care unit referral of lung cancer patients: a pilot study

European Respiratory Journal ◽

10.1183/09031936.00118114 ◽

2014 ◽

Vol 45 (2) ◽

pp. 491-500 ◽

Cited By ~ 19

Author(s):

Anne-Claire Toffart ◽

Carola Alegria Pizarro ◽

Carole Schwebel ◽

Linda Sakhri ◽

Clemence Minet ◽

...

Keyword(s):

Lung Cancer ◽

Intensive Care Unit ◽

Intensive Care ◽

Cancer Patients ◽

Selection Criteria ◽

Performance Status ◽

Lung Cancer Patients ◽

Icu Admission ◽

Organ Failures ◽

Intensity Of Care

The decision-making process for the intensity of care delivered to patients with lung cancer and organ failure is poorly understood, and does not always involve intensivists. Our objective was to describe the potential suitability for intensive care unit (ICU) referral of lung cancer in-patients with organ failures.We prospectively included consecutive lung cancer patients with failure of at least one organ admitted to the teaching hospital in Grenoble, France, between December 2010 and October 2012.Of 140 patients, 121 (86%) were evaluated by an oncologist and 49 (35%) were referred for ICU admission, with subsequent admission for 36 (73%) out of those 49. Factors independently associated with ICU referral were performance status ⩽2 (OR 10.07, 95% CI 3.85–26.32), nonprogressive malignancy (OR 7.00, 95% CI 2.24–21.80), and no explicit refusal of ICU admission by the patient and/or family (OR 7.95, 95% CI 2.39–26.37). Factors independently associated with ICU admission were the initial ward being other than the lung cancer unit (OR 6.02, 95% CI 1.11–32.80) and an available medical ICU bed (OR 8.19, 95% CI 1.48–45.35).Only one-third of lung cancer patients with organ failures were referred for ICU admission. The decision not to consider ICU admission was often taken by a non-intensivist, with advice from an oncologist rather than an intensivist.

Download Full-text

Multicenter outcome study of cancer patients admitted to the intensive care unit: a probability of mortality model.

Journal of Clinical Oncology ◽

10.1200/jco.1998.16.2.761 ◽

1998 ◽

Vol 16 (2) ◽

pp. 761-770 ◽

Cited By ~ 162

Author(s):

J S Groeger ◽

S Lemeshow ◽

K Price ◽

D M Nierman ◽

P White ◽

...

Keyword(s):

Intensive Care Unit ◽

Logistic Regression ◽

Intensive Care ◽

Hospital Mortality ◽

Cancer Patients ◽

The United States ◽

Categorical Variables ◽

Allogeneic Bone ◽

Icu Admission ◽

Hospital Survival

PURPOSE To develop prospectively and validate a model for probability of hospital survival at admission to the intensive care unit (ICU) of patients with malignancy. PATIENTS AND METHODS This was an inception cohort study in the setting of four ICUs of academic medical centers in the United States. Defined continuous and categorical variables were collected on consecutive patients with cancer admitted to the ICU. A preliminary model was developed from 1,483 patients and then validated on an additional 230 patients. Multiple logistic regression modeling was used to develop the models and subsequently evaluated by goodness-of-fit and receiver operating characteristic (ROC) analysis. The main outcome measure was hospital survival after ICU admission. RESULTS The observed hospital mortality rate was 42%. Continuous variables used in the ICU admission model are PaO2/FiO2 ratio, platelet count, respiratory rate, systolic blood pressure, and days of hospitalization pre-ICU. Categorical entries include presence of intracranial mass effect, allogeneic bone marrow transplantation, recurrent or progressive cancer, albumin less than 2.5 g/dL, bilirubin > or = 2 mg/dL, Glasgow Coma Score less than 6, prothrombin time greater than 15 seconds, blood urea nitrogen (BUN) greater than 50 mg/dL, intubation, performance status before hospitalization, and cardiopulmonary resuscitation (CPR). The P values for the fit of the preliminary and validation models are .939 and .314, respectively, and the areas under the ROC curves are .812 and .802. CONCLUSION We report a disease-specific multivariable logistic regression model to estimate the probability of hospital mortality in a cohort of critically ill cancer patients admitted to the ICU. The model consists of 16 unambiguous and readily available variables. This model should move the discussion regarding appropriate use of ICU resources forward. Additional validation in a community hospital setting is warranted.

Download Full-text

Clinical Characteristics and Prognostic Factors for Intensive Care Unit Admission of Patients With COVID-19: Retrospective Study Using Machine Learning and Natural Language Processing (Preprint)

10.2196/preprints.21801 ◽

2020 ◽

Author(s):

Jose Luis Izquierdo ◽

Julio Ancochea ◽

Joan B Soriano ◽

Keyword(s):

Machine Learning ◽

Intensive Care Unit ◽

Intensive Care ◽

Natural Language Processing ◽

Natural Language ◽

Language Processing ◽

Clinical Characteristics ◽

Epidemiological Methods ◽

Icu Admission ◽

La Mancha

BACKGROUND Many factors involved in the onset and clinical course of the ongoing COVID-19 pandemic are still unknown. Although big data analytics and artificial intelligence are widely used in the realms of health and medicine, researchers are only beginning to use these tools to explore the clinical characteristics and predictive factors of patients with COVID-19. OBJECTIVE Our primary objectives are to describe the clinical characteristics and determine the factors that predict intensive care unit (ICU) admission of patients with COVID-19. Determining these factors using a well-defined population can increase our understanding of the real-world epidemiology of the disease. METHODS We used a combination of classic epidemiological methods, natural language processing (NLP), and machine learning (for predictive modeling) to analyze the electronic health records (EHRs) of patients with COVID-19. We explored the unstructured free text in the EHRs within the Servicio de Salud de Castilla-La Mancha (SESCAM) Health Care Network (Castilla-La Mancha, Spain) from the entire population with available EHRs (1,364,924 patients) from January 1 to March 29, 2020. We extracted related clinical information regarding diagnosis, progression, and outcome for all COVID-19 cases. RESULTS A total of 10,504 patients with a clinical or polymerase chain reaction–confirmed diagnosis of COVID-19 were identified; 5519 (52.5%) were male, with a mean age of 58.2 years (SD 19.7). Upon admission, the most common symptoms were cough, fever, and dyspnea; however, all three symptoms occurred in fewer than half of the cases. Overall, 6.1% (83/1353) of hospitalized patients required ICU admission. Using a machine-learning, data-driven algorithm, we identified that a combination of age, fever, and tachypnea was the most parsimonious predictor of ICU admission; patients younger than 56 years, without tachypnea, and temperature <39 degrees Celsius (or >39 ºC without respiratory crackles) were not admitted to the ICU. In contrast, patients with COVID-19 aged 40 to 79 years were likely to be admitted to the ICU if they had tachypnea and delayed their visit to the emergency department after being seen in primary care. CONCLUSIONS Our results show that a combination of easily obtainable clinical variables (age, fever, and tachypnea with or without respiratory crackles) predicts whether patients with COVID-19 will require ICU admission.

Download Full-text

Disparities in Intensive Care Unit Admission and Mortality Among Patients With Schizophrenia and COVID-19: A National Cohort Study

Schizophrenia Bulletin ◽

10.1093/schbul/sbaa158 ◽

2020 ◽

Cited By ~ 1

Author(s):

Guillaume Fond ◽

Vanessa Pauly ◽

Marc Leone ◽

Pierre-Michel Llorca ◽

Veronica Orleans ◽

...

Keyword(s):

Health Care ◽

Intensive Care Unit ◽

Mental Illness ◽

Cohort Study ◽

Intensive Care ◽

Hospital Mortality ◽

Severe Mental Illness ◽

Vulnerable Population ◽

Icu Admission ◽

Risk Of Death

Abstract Patients with schizophrenia (SCZ) represent a vulnerable population who have been understudied in COVID-19 research. We aimed to establish whether health outcomes and care differed between patients with SCZ and patients without a diagnosis of severe mental illness. We conducted a population-based cohort study of all patients with identified COVID-19 and respiratory symptoms who were hospitalized in France between February and June 2020. Cases were patients who had a diagnosis of SCZ. Controls were patients who did not have a diagnosis of severe mental illness. The outcomes were in-hospital mortality and intensive care unit (ICU) admission. A total of 50 750 patients were included, of whom 823 were SCZ patients (1.6%). The SCZ patients had an increased in-hospital mortality (25.6% vs 21.7%; adjusted OR 1.30 [95% CI, 1.08–1.56], P = .0093) and a decreased ICU admission rate (23.7% vs 28.4%; adjusted OR, 0.75 [95% CI, 0.62–0.91], P = .0062) compared with controls. Significant interactions between SCZ and age for mortality and ICU admission were observed (P = .0006 and P < .0001). SCZ patients between 65 and 80 years had a significantly higher risk of death than controls of the same age (+7.89%). SCZ patients younger than 55 years had more ICU admissions (+13.93%) and SCZ patients between 65 and 80 years and older than 80 years had less ICU admissions than controls of the same age (−15.44% and −5.93%, respectively). Our findings report the existence of disparities in health and health care between SCZ patients and patients without a diagnosis of severe mental illness. These disparities differed according to the age and clinical profile of SCZ patients, suggesting the importance of personalized COVID-19 clinical management and health care strategies before, during, and after hospitalization for reducing health disparities in this vulnerable population.

Download Full-text

Advance care planning and terminal intensive care unit admissions for cancer patients: We can do better.

Journal of Clinical Oncology ◽

10.1200/jco.2019.37.15_suppl.e23000 ◽

2019 ◽

Vol 37 (15_suppl) ◽

pp. e23000-e23000

Author(s):

Joseph Heng ◽

Ramy Sedhom ◽

Thomas J. Smith

Keyword(s):

Intensive Care Unit ◽

Intensive Care ◽

High Risk ◽

Advance Care Planning ◽

Hematologic Malignancies ◽

Cancer Center ◽

Care Planning ◽

Code Status ◽

Risk Of Death ◽

Advance Care

e23000 Background: Terminal oncology intensive care unit (ICU) admissions are associated with high healthcare costs and decreased quality of life. Chemotherapy can be given in non-curative settings to optimize symptom control, but use of it at the end of life does not improve longevity. In addition, goals of care are too often not addressed for patients at high risk of death. Methods: We carried out a retrospective review identifying patients of a large academic cancer center who were admitted to and expired in an ICU between January 1, 2017 to December 31, 2018. Results: 120 patients met inclusion criteria. Median age was 58 years. Only 15.0% (n = 18) of all patients had advance directives. The majority of patients (94.1%, n = 113) were FULL CODE on admission. Median duration of admission was 10 days. Median time to death from ICU admission was 7.5 days. 65.0% (n = 78) of all patients were intubated, while 15.0% (n = 15) received CPR. 58.3% (n = 70) of the study population had solid malignancies; of note, 97.1% (n = 68) of these patients were metastatic at presentation and had a median ECOG performance status of 2. Patients with metastatic solid tumors typically have a more indolent course of progression compared to patients with hematologic malignancies. However, only 23.5% (n = 16) had discussed goals of care or code status with their outpatient oncologists, despite many seeing them within the last month prior to admission (83.8%, n = 57). Similarly, only 4.0% (n = 2) of patients with hematologic malignancies had advance care planning discussions with their oncologists prior to their terminal ICU admission. 27.5% (n = 33) of all patients had an inpatient palliative care consult. The inpatient pulmonary/critical care team had a high rate of inpatient code status transitions, with 85.6% (n = 97) of FULL CODE admissions transitioning to DNR/DNI. Conclusions: These findings reflect contemporary practice at a major academic cancer center. Despite most patients having regular contact with their outpatient oncologists, the intensity of health care utilization noted highlights a need to optimize recognition of patients at high risk of death and to engage patients in advance care planning discussions to avoid terminal ICU admissions.

Download Full-text