Application of Machine Learning to Prediction of Surgical Site Infection

Fletcher, Roger; Olubeko, Olasubomi; Sonthalia, Harsh; Kateera, Fredrick; Nkurunziza, Theoneste; Ashby, Joanna; Riviello, Robert; Hedt‐Gauthier, Bethany

doi:10.1109/embc.2019.8857942

Cited by 13 publications

(26 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previously, a variety of ML models, including the support vector machine model, backward stepwise method, deep neural network, naive Bayes algorithm, Bayesian model, convolutional neural network and artificial neural network model have been used to predict SSI with an AUC ranging from .65 to .97. [13][14][15][16][17][18] However, while these models successfully predicted patients who would develop SSIs, they were not trained to discriminate between the type of SSI. Deep and organ space infections have a greater impact on patient outcome compared to superficial SSI, often require more invasive management with intravenous antibiotics, percutaneous drainage procedures, and reoperation, and may arise from a different disease process.…”

Section: Discussionmentioning

confidence: 99%

Machine Learning Model for Assessment of Risk Factors and Postoperative Day for Superficial vs Deep/Organ-Space Surgical Site Infections

et al. 2023

View full text Add to dashboard Cite

Background. Deep and organ space surgical site infections (SSI) require more intensive treatment, may result in more severe clinical disease and may have different risk factors when compared to superficial SSIs. Machine learning (ML) algorithms provide the opportunity to analyze multiple factors to predict of the type and time of development of SSI. Therefore, we developed a ML model to predict type and postoperative week of SSI. Methodology. A case-control study was conducted among patients who developed a SSI after undergoing general surgery procedures at a tertiary care hospital between 2019 to 2020. Patients were followed for 30 days. Six ML algorithms were trained as predictors of type of infection (superficial vs deep/organ space) and time of infection, and tested using area under the receiver operating characteristic curve (AUC-ROC). Results. Data for 113 patients with SSIs was available. Of these 62 (54.8%) had superficial and 51 had (45.2%) deep/organ space infections. Compared with other ML algorithms, the XG boost univariate model had highest AUC-ROC (.84) for prediction of type of SSI and Stochastic gradient boosting univariate, logistic regression univariate, XG boost univariate, and random forest classification univariate model had the highest AUC-ROC (.74) for prediction of week of infection. Conclusions. ML models offer reasonable accuracy in prediction of superficial vs deep SSI and time of developing infection. Follow-up duration and allocation of treatment strategies can be informed by ML predictions.

show abstract

Section: Discussionmentioning

confidence: 99%

Machine Learning Model for Assessment of Risk Factors and Postoperative Day for Superficial vs Deep/Organ-Space Surgical Site Infections

et al. 2023

View full text Add to dashboard Cite

show abstract

“…A third strategy is to develop algorithms using artificial intelligence, particularly machine learning. 47 This would require developing large libraries of images and diagnoses to train and evaluate algorithms, but if successful, these algorithms could improve accuracy and remove the person-to-person variability of diagnoses based on image review. 44 …”

Section: Discussionmentioning

confidence: 99%

“…This would allow comparisons between wound images from POD3 to the POD10 so that the GP can assess how the wound has progressed over time rather than at a static time point. A third strategy is to develop algorithms using artificial intelligence, particularly machine learning 47. This would require developing large libraries of images and diagnoses to train and evaluate algorithms, but if successful, these algorithms could improve accuracy and remove the person-to-person variability of diagnoses based on image review 44…”

Section: Discussionmentioning

confidence: 99%

mHealth-community health worker telemedicine intervention for surgical site infection diagnosis: a prospective study among women delivering via caesarean section in rural Rwanda

Nkurunziza

Williams

Kateera³

et al. 2022

BMJ Glob Health

Self Cite

View full text Add to dashboard Cite

BackgroundSurgical site infections (SSIs) cause a significant global public health burden in low and middle-income countries. Most SSIs develop after patient discharge and may go undetected. We assessed the feasibility and diagnostic accuracy of an mHealth-community health worker (CHW) home-based telemedicine intervention to diagnose SSIs in women who delivered via caesarean section in rural Rwanda.MethodsThis prospective cohort study included women who underwent a caesarean section at Kirehe District Hospital between September 2019 and March 2020. At postoperative day 10 (±3 days), a trained CHW visited the woman at home, provided wound care and transmitted a photo of the wound to a remote general practitioner (GP) via WhatsApp. The GP reviewed the photo and made an SSI diagnosis. The next day, the woman returned to the hospital for physical examination by an independent GP, whose SSI diagnosis was considered the gold standard for our analysis. We describe the intervention process indicators and report the sensitivity and specificity of the telemedicine-based diagnosis.ResultsOf 787 women included in the study, 91.4% (n=719) were located at their home by the CHW and all of them (n=719, 100%) accepted the intervention. The full intervention was completed, including receipt of GP telemedicine diagnosis within 1 hour, for 79.0% (n=623). The GPs diagnosed 30 SSIs (4.2%) through telemedicine and 38 SSIs (5.4%) through physical examination. The telemedicine sensitivity was 36.8% and specificity was 97.6%. The negative predictive value was 96.4%.ConclusionsImplementation of an mHealth-CHW home-based intervention in rural Rwanda and similar settings is feasible. Patients’ acceptance of the intervention was key to its success. The telemedicine-based SSI diagnosis had a high negative predictive value but a low sensitivity. Further studies must explore strategies to improve accuracy, such as accompanying wound images with clinical data or developing algorithms using machine learning.

show abstract

“…The classification of learning algorithms aims to build a classifier from a training dataset in order to accurately predict testing samples [ 31 ]. The performance of the prediction models can be evaluated using different metrics, such as sensitivity, specificity, accuracy and the area under the ROC curve (AUC) [ 32 ]. Studies show that the AUC is more statistically discriminant than other methods when evaluating the predictive ability of the classification algorithm [ 31 ].…”

Section: Discussionmentioning

confidence: 99%

Non-Fatal Drowning Risk Prediction Based on Stacking Ensemble Algorithm

Xie

et al. 2022

Children

View full text Add to dashboard Cite

Drowning is a major public health problem and a leading cause of death in children living in developing countries. We seek better machine learning (ML) algorithms to provide a novel risk-assessment insight on non-fatal drowning prediction. The data on non-fatal drowning were collected in Qingyuan city, Guangdong Province, China. We developed four ML models to predict the non-fatal drowning risk, including a logistic regression model (LR), random forest model (RF), support vector machine model (SVM), and stacking-based model, on three primary learners (LR, RF, SVM). The area under the curve (AUC), F1 value, accuracy, sensitivity, and specificity were calculated to evaluate the predictive ability of the different learning algorithms. This study included a total of 8390 children. Of those, 12.07% (1013) had experienced non-fatal drowning. We found the following risk factors are closely associated with the risk of non-fatal drowning: the frequency of swimming in open water, distance between the school and the surrounding open waters, swimming skills, personality (introvert) and relationality with family members. Compared to the other three base models, the stacking generalization model achieved a superior performance in the non-fatal drowning dataset (AUC = 0.741, sensitivity = 0.625, F1 value = 0.359, accuracy = 0.739 and specificity = 0.754). This study indicates that applying stacking ensemble algorithms in the non-fatal drowning dataset may outperform other ML models.

show abstract

Application of Machine Learning to Prediction of Surgical Site Infection

Cited by 13 publications

References 5 publications

Machine Learning Model for Assessment of Risk Factors and Postoperative Day for Superficial vs Deep/Organ-Space Surgical Site Infections

Machine Learning Model for Assessment of Risk Factors and Postoperative Day for Superficial vs Deep/Organ-Space Surgical Site Infections

mHealth-community health worker telemedicine intervention for surgical site infection diagnosis: a prospective study among women delivering via caesarean section in rural Rwanda

Non-Fatal Drowning Risk Prediction Based on Stacking Ensemble Algorithm

Contact Info

Product

Resources

About