Application of Machine Learning to Prediction of Surgical Site Infection

Machine Learning Approaches for the Image-Based Identification of Surgical Wound Infections: Scoping Review

BACKGROUND

Surgical site infections (SSIs) occur frequently and impact patients and health care systems. Remote surveillance of surgical wounds is currently limited by the need for manual assessment by clinicians. Machine learning (ML)-based methods have recently been used to address various aspects of the postoperative wound healing process and may be used to improve the scalability and cost-effectiveness of remote surgical wound assessment.

OBJECTIVE

The objective of this review was to provide an overview of the ML methods that have been used to identify surgical wound infections from images.

METHODS

We conducted a scoping review of ML approaches for visual detection of SSIs following the JBI (Joanna Briggs Institute) methodology. Reports of participants in any postoperative context focusing on identification of surgical wound infections were included. Studies that did not address SSI identification, surgical wounds, or did not use image or video data were excluded. We searched MEDLINE, Embase, CINAHL, CENTRAL, Web of Science Core Collection, IEEE Xplore, Compendex, and arXiv for relevant studies in November 2022. The records retrieved were double screened for eligibility. A data extraction tool was used to chart the relevant data, which was described narratively and presented using tables. Employment of TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) guidelines was evaluated and PROBAST (Prediction Model Risk of Bias Assessment Tool) was used to assess risk of bias (RoB).

RESULTS

In total, 10 of the 715 unique records screened met the eligibility criteria. In these studies, the clinical contexts and surgical procedures were diverse. All papers developed diagnostic models, though none performed external validation. Both traditional ML and deep learning methods were used to identify SSIs from mostly color images, and the volume of images used ranged from under 50 to thousands. Further, 10 TRIPOD items were reported in at least 4 studies, though 15 items were reported in fewer than 4 studies. PROBAST assessment led to 9 studies being identified as having an overall high RoB, with 1 study having overall unclear RoB.

CONCLUSIONS

Research on the image-based identification of surgical wound infections using ML remains novel, and there is a need for standardized reporting. Limitations related to variability in image capture, model building, and data sources should be addressed in the future.

Machine learning models to detect and predict patient safety events using electronic health records: A systematic review

INTRODUCTION

Identifying patient safety events using electronic health records (EHRs) and automated machine learning-based detection methods can help improve the efficiency and quality of healthcare service provision.

OBJECTIVE

This study aimed to systematically review machine learning-based methods and techniques, as well as their results for patient safety event management using EHRs.

METHODS

We reviewed the studies that focused on machine learning techniques, including automatic prediction and detection of patient safety events and medical errors through EHR analysis to manage patient safety events. The data were collected by searching Scopus, PubMed (Medline), Web of Science, EMBASE, and IEEE Xplore databases.

RESULTS

After screening, 41 papers were reviewed. Support vector machine (SVM), random forest, conditional random field (CRF), and bidirectional long short-term memory with conditional random field (BiLSTM-CRF) algorithms were mostly applied to predict, identify, and classify patient safety events using EHRs; however, they had different performances. BiLSTM-CRF was employed in most of the studies to extract and identify concepts, e.g., adverse drug events (ADEs) and adverse drug reactions (ADRs), as well as relationships between drug and severity, drug and ADEs, drug and ADRs. Recurrent neural networks (RNN) and BiLSTM-CRF had the best results in detecting ADEs compared to other patient safety events. Linear classifiers and Naive Bayes (NB) had the highest performance for ADR detection. Logistic regression had the best results in detecting surgical site infections. According to the findings, the quality of articles has non-significantly improved in recent years, but they had low average scores.

CONCLUSIONS

Machine learning can be useful in automatic detection and prediction of patient safety events. However, most of these algorithms have not yet been externally validated or prospectively tested. Therefore, further studies are required to improve the performance of these automated systems.

Machine learning applied in maternal and fetal health: a narrative review focused on pregnancy diseases and complications

Introduction

Machine learning (ML) corresponds to a wide variety of methods that use mathematics, statistics and computational science to learn from multiple variables simultaneously. By means of pattern recognition, ML methods are able to find hidden correlations and accomplish accurate predictions regarding different conditions. ML has been successfully used to solve varied problems in different areas of science, such as psychology, economics, biology and chemistry. Therefore, we wondered how far it has penetrated into the field of obstetrics and gynecology.

Aim

To describe the state of art regarding the use of ML in the context of pregnancy diseases and complications.

Methodology

Publications were searched in PubMed, Web of Science and Google Scholar. Seven subjects of interest were considered: gestational diabetes mellitus, preeclampsia, perinatal death, spontaneous abortion, preterm birth, cesarean section, and fetal malformations.

Current state

ML has been widely applied in all the included subjects. Its uses are varied, the most common being the prediction of perinatal disorders. Other ML applications include (but are not restricted to) biomarker discovery, risk estimation, correlation assessment, pharmacological treatment prediction, drug screening, data acquisition and data extraction. Most of the reviewed articles were published in the last five years. The most employed ML methods in the field are non-linear. Except for logistic regression, linear methods are rarely used.

Future challenges

To improve data recording, storage and update in medical and research settings from different realities. To develop more accurate and understandable ML models using data from cutting-edge instruments. To carry out validation and impact analysis studies of currently existing high-accuracy ML models.

Conclusion

The use of ML in pregnancy diseases and complications is quite recent, and has increased over the last few years. The applications are varied and point not only to the diagnosis, but also to the management, treatment, and pathophysiological understanding of perinatal alterations. Facing the challenges that come with working with different types of data, the handling of increasingly large amounts of information, the development of emerging technologies, and the need of translational studies, it is expected that the use of ML continue growing in the field of obstetrics and gynecology.

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

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.

Non-Fatal Drowning Risk Prediction Based on Stacking Ensemble Algorithm

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.

The Use of Mobile Thermal Imaging and Deep Learning for Prediction of Surgical Site Infection

The ability to detect surgical site infections (SSI) is a critical need for healthcare worldwide, but is especially important in low-income countries, where there is limited access to health facilities and trained clinical staff. In this paper, we present a new method of predicting SSI using a thermal image collected with a smart phone. Machine learning algorithms were developed using images collected as part of a clinical study that included 530 women in rural Rwanda who underwent cesarean section surgery. Thermal images were collected approximately 10 days after surgery, in conjunction with an examination by a trained doctor to determine the status of the wound (infected or not). Of the 530 women, 30 were found to have infected wounds. The data were used to develop two Convolutional Neural Net (CNN) models, with special care taken to avoid overfitting and address the problem of class imbalance in binary classification. The first model, a 6-layer naïve CNN model, demonstrated a median accuracy of AUC=0.84 with sensitivity=71% and specificity=87%. The transfer learning CNN model demonstrated a median accuracy of AUC=0.90 with sensitivity =95% and specificity=84%. To our knowledge, this is the first successful demonstration of a machine learning algorithm to predict surgical infection using thermal images alone.Clinical Relevance- This work establishes a promising new method for automated detection of surgical site infection.

Use of Convolutional Neural Nets and Transfer Learning for Prediction of Surgical Site Infection from Color Images

One of the greatest concerns in post-operative care is the infection of the surgical wound. Such infections are a particular concern in global health and low-resource areas, where microbial antibiotic resistance is often common. In order to help address this problem, there is a great interest in developing simple tools for early detection of surgical wounds. Motivated by this need, we describe the development of two Convolutional Neural Net (CNN) models designed to detect an infection in a surgical wound using a color image taken from a mobile device. These models were developed using image data collected from a clinical study with 572 women in Rural Rwanda, who underwent Cesarean section surgery and had photos taken approximately 10 days after surgery. Infected wounds (N=62) were diagnosed by a trained doctor through a physical exam. In our model development, we observed a trade-off between AUC accuracy and sensitivity, and we chose to optimize for sensitivity, to match its use as a screening tool. Our naïve CNN model, with a limited number of convolutions and parameters, achieved median AUC = 0.655, true positive rate sensitivity = 0.75, specificity = 0.58, classification accuracy = 0.86. The second CNN model, developed with transfer learning using the Resnet50 architecture, produced a median AUC = 0.639 sensitivity = 0.92, specificity = 0.18, and classification accuracy 0.82. We discuss the specific training and optimization methods used to compensate for significant class imbalance and maximize sensitivity.

Evaluating urban-rural access to pathology and laboratory medicine services in Tanzania

Placement of pathology and laboratory medicine (PALM) services requires balancing efficiency (maximizing test volume) with equitable urban-rural access. We compared the association between population density (proxy for efficiency) and travel time to the closest facility (proxy for equitable access) across levels of Tanzania's public sector health system. We linked geospatial data for Tanzania from multiple sources. Data on facility locations and other geographic measures were collected from government and non-governmental databases. We classified facilities assuming increasing PALM availability by tier: (1) dispensaries, (2) health centres, (3) district hospitals and (4) regional/referral hospitals. We used the AccessMod 5 algorithm to estimate travel time to the closest facility for each tier across Tanzania with 500-m resolution. District-level average population density and travel time to the closest facility were calculated and presented using medians and interquartile ranges. Spatial correlations between these variables were estimated using the global Moran's I and bivariate Local Indicator of Spatial Autocorrelation, specifying a queen's neighbourhood matrix. Spatial analysis was restricted to 171 contiguous districts. The study included 5406 dispensaries, 675 health centres, 186 district hospitals and 37 regional/referral hospitals. District-level travel times were shortest for Tier 1 (median: [IQR]: 45.4 min [30.0-74.7]) and longest for Tier 4 facilities (160.2 min [107.3-260.0]). There was a weak spatial autocorrelation across tiers (Tier 1: -0.289, Tier 2: -0.292, Tier 3: -0.271 and Tier 4: -0.258) and few districts were classified as significant spatial outliers. Across tiers, geographic patterns of populated districts surrounded by neighbours with short travel time and sparsely populated districts surrounded by neighbours with long travel time were observed. Similar spatial correlation measures across health system levels suggest that Tanzania's health system reflects equitable urban-rural access to different PALM services. Longer travel times to hospital-based care could be ameliorated by shifting specialized diagnostics to more accessible lower tiers.