Figure - available via license: CC BY
Content may be subject to copyright.
Source publication
Objective
Machine learning methods may have better or comparable predictive ability than traditional analysis. We explore machine learning methods to predict the likelihood of acute kidney injury after liver cancer resection.
Methods
This is a secondary analysis cohort study. We reviewed data from patients who had undergone resection of primary he...
Context in source publication
Similar publications
Data analysis and prediction have gradually attracted more and more attention in the smart healthcare industry. The smart medical prediction system is of great importance to the enterprise strategy and business development, and it is also of great value to provide medical advices for patients and assist patient guidance. The research theme is the u...
Citations
... The most common pathological type of RCC is renal clear cell carcinoma (KIRC), which has poor prognosis and a high degree of malignancy. 3 mRNA-processing events which include alternative splicing, m6A methylation and alternative polyadenylation (APA), are crucial in the regulation of most human genes in various diseases, such as brain cancer, lung cancer, liver cancer and COVID-19. [4][5][6][7][8][9] However, there is limited research on the relationship between alternative polyadenylation (APA) and renal clear cell carcinoma. ...
Background
Alternative polyadenylation (APA) plays a vital regulatory role in various diseases. It is widely accepted that APA is regulated by APA regulatory factors.
Objective
Whether APA regulatory factors affect the prognosis of renal cell carcinoma remains unclear, and this is the main topic of this study.
Methods
We downloaded the transcriptome and clinical data from The Cancer Genome Atlas (TCGA) database. We used the Lasso regression system to construct an APA model for analyzing the relationship between common APA regulatory factors and renal cell carcinoma. We also validated our APA model using independent GEO datasets (GSE29609, GSE76207).
Results
It was found that the expression levels of 5 APA regulatory factors (CPSF1, CPSF2, CSTF2, PABPC1, and PABPC4) were significantly associated with tumor gene mutation burden (TMB) score in renal clear cell carcinoma, and the risk score constructed using the expression level of 5 key APA regulatory factors could be used to predict the outcome of renal clear cell carcinoma. The TMB score is associated with the remodeling of the immune microenvironment.
Conclusions
By identifying key APA regulatory factors in renal cell carcinoma and constructing risk scores for key APA regulatory factors, we showed that key APA regulators affect prognosis of renal clear cell carcinoma patients. In addition, the risk score level is associated with TMB, indicating that APA may affect the efficacy of immunotherapy through immune microenvironment-related genes. This helps us better understand the mRNA processing mechanism of renal clear cell carcinoma.
... With the rapid advancements in AI-assisted digital pathology, AIdriven automatic cell analysis has emerged as a burgeoning trend, widely used for cell identification, quantification, and disease staging. 22 Figure 3A). Notably, CPCs in PB often show fewer dysplastic ...
Background
The presence of circulating plasma cells (CPCs) is an important laboratory indicator for the diagnosis, staging, risk stratification, and progression monitoring of multiple myeloma (MM). Early detection of CPCs in the peripheral blood (PB) followed by timely interventions can significantly improve MM prognosis and delay its progression. Although the conventional cell morphology examination remains the predominant method for CPC detection because of accessibility, its sensitivity and reproducibility are limited by technician expertise and cell quantity constraints. This study aims to develop an artificial intelligence (AI)–based automated system for a more sensitive and efficient CPC morphology detection.
Methods
A total of 137 bone marrow smears and 72 PB smears from patients with at Zhongshan Hospital, Fudan University, were retrospectively reviewed. Using an AI‐powered digital pathology platform, Morphogo, 305,019 cell images were collected for training. Morphogo’s efficacy in CPC detection was evaluated with additional 184 PB smears (94 from patients with MM and 90 from those with other hematological malignancies) and compared with manual microscopy.
Results
Morphogo achieved 99.64% accuracy, 89.03% sensitivity, and 99.68% specificity in classifying CPCs. At a 0.60 threshold, Morphogo achieved a sensitivity of 96.15%, which was approximately twice that of manual microscopy, with a specificity of 78.03%. Patients with CPCs detected by AI scanning had a significantly shorter median progression‐free survival compared with those without CPC detection (18 months vs. 34 months, p< .01).
Conclusions
Morphogo is a highly sensitive system for the automated detection of CPCs, with potential applications in initial screening, prognosis prediction, and posttreatment monitoring for MM patients.
Plain Language Summary
Diagnosing and monitoring multiple myeloma (MM), a type of blood cancer, requires identifying and quantifying specific cells called circulating plasma cells (CPCs) in the blood. The conventional method for detecting CPCs is manual microscopic examination, which is time‐consuming and lacks sensitivity. This study introduces a highly sensitive CPC detection method using an artificial intelligence–based system, Morphogo. It demonstrated remarkable sensitivity and accuracy, surpassing conventional microscopy. This advanced approach suggests that early and accurate CPC detection is achievable by morphology examination, making efficient CPC screening more accessible for patients with MM. This innovative system has the potential to be used in the diagnosis and risk assessment of MM.
... Surgical resection is the primary treatment for hepatic malignancies [11][12][13], and intraoperative blood loss and transfusion requirements are closely related to perioperative morbidity and mortality in patients undergoing liver cancer surgery [14]. Hepatic portal blocking is usually necessary to control intraoperative bleeding [15]. In 1977, Foster and Berman published the results of a multicenter analysis of 621 patients who underwent hepatectomy for various indications, which showed that operative mortality rates exceeding 13% and 20% for hepatectomy and extended hepatic resection, respectively, with 20% of deaths resulting from bleeding [14,16]. ...
... In 1977, Foster and Berman published the results of a multicenter analysis of 621 patients who underwent hepatectomy for various indications, which showed that operative mortality rates exceeding 13% and 20% for hepatectomy and extended hepatic resection, respectively, with 20% of deaths resulting from bleeding [14,16]. Lei et al [15] conducted a retrospective study involving 643 consecutive patients who underwent hepatic resection for HCC. The study identified several risk factors associated with major intraoperative blood loss in these patients. ...
BACKGROUND
Surgical resection remains the primary treatment for hepatic malignancies, and intraoperative bleeding is associated with a significantly increased risk of death. Therefore, accurate prediction of intraoperative bleeding risk in patients with hepatic malignancies is essential to preventing bleeding in advance and providing safer and more effective treatment.
AIM
To develop a predictive model for intraoperative bleeding in primary hepatic malignancy patients for improving surgical planning and outcomes.
METHODS
The retrospective analysis enrolled patients diagnosed with primary hepatic malignancies who underwent surgery at the Hepatobiliary Surgery Department of the Fourth Hospital of Hebei Medical University between 2010 and 2020. Logistic regression analysis was performed to identify potential risk factors for intraoperative bleeding. A prediction model was developed using Python programming language, and its accuracy was evaluated using receiver operating characteristic (ROC) curve analysis.
RESULTS
Among 406 primary liver cancer patients, 16.0% (65/406) suffered massive intraoperative bleeding. Logistic regression analysis identified four variables as associated with intraoperative bleeding in these patients: ascites [odds ratio (OR): 22.839; P < 0.05], history of alcohol consumption (OR: 2.950; P < 0.015), TNM staging (OR: 2.441; P < 0.001), and albumin-bilirubin score (OR: 2.361; P < 0.001). These variables were used to construct the prediction model. The 406 patients were randomly assigned to a training set (70%) and a prediction set (30%). The area under the ROC curve values for the model’s ability to predict intraoperative bleeding were 0.844 in the training set and 0.80 in the prediction set.
CONCLUSION
The developed and validated model predicts significant intraoperative blood loss in primary hepatic malignancies using four preoperative clinical factors by considering four preoperative clinical factors: ascites, history of alcohol consumption, TNM staging, and albumin-bilirubin score. Consequently, this model holds promise for enhancing individualised surgical planning.
... However, through continuous tunings of ML algorithms by algorithm engineers, new-proposed ML models have potentials to be exploited again through continuous learning. In a comparison of four ML algorithms incorporated by Lei [23] for predictions of acute kidney injury (AKI) in 1,173 postoperative hepatectomy patients, GBDT was found to have the best predictive ability for AKI, which may provide a reference for early identi cation of AKI after hepatectomy. Liu [24] showed that tuned Random Forest had the best predictive ability for patients at high risk of bone metastases in thyroid cancer from the seer database. ...
Objective: Post-hepatectomy liver failure (PHLF) is a terrible and serious complication after liver resection. Machine learning algorithms are emerging for data mining in recent years and have been shown to have greater advantages over traditional statistics. This study includes a comparison of different traditional machine learning algorithms and selects the best model for predicting PHLF.
Materials and Methods:Review the data of patients who had undergone resection of hepatocellular carcinoma from January 2013 to October 2022 in our hospital and randomly divide the data into a training set and a validation set at a 7:3 ratio. Using mutual information to screen 10 clinical characteristics with a higher correlation to PHLF. The data was trained and validated using Logistic Regression, Decision Tree, Gradient Boosting Decision Tree(GBDT), Random Forest, Extreme Gradient Boosting(XGBoost), LightGBM, multi-model fusion(hard voting), and multi-model fusion(soft voting). The hyperparameter of different machine learning was searched to achieve the best-fitting performance. Different traditional machine learning algorithms are evaluated comprehensively through accuracy rate, precision rate, recall rate, F1 score, and Receiver Operating Characteristic (ROC) and its area under the curve(AUC). Based on the feature importance ranking of the best model, clinical characteristics related to PHLF were ranked.
Results: A total of 319 patients’ data were included in this study, with 9.4% of the patients in the liver failure group(n=30). 10 clinical characteristics with higher correlation to PHLF are preoperative platelet count, preoperative prothrombin time, perioperative blood loss, perioperative transfusion(Yes/No), duration of surgery, clinically significant portal hypertension(Yes/No), preoperative aspartate aminotransferase, preoperative albumin, preoperative total bilirubin, and type of resection(minor/major). XGBoost and LightGBM showed the best performance on training set with an accuracy rate of 1. However, their performance decreased on validation set with an accuracy rate of 0.9375 and 0.9167, respectively. GBDT had the best anti-fitting performance in the training and validation sets, with an accuracy rate of 0.9462 and 0.9479, respectively. Preoperative albumin, perioperative blood loss, preoperative platelet count, duration of surgery, and preoperative alanine transaminase had higher weights in GBDT. The accuracy rate of the multi-model fusion(hard voting) was 0.9955 and 0.9583 in the training and validation cohort, respectively, while the accuracy rate of the multi-model fusion(soft voting) was 0.9731 and 0.9479 on training set and validation set, respectively.
Conclusion: GBDT performed the best among different traditional machine learning algorithms, and XGBoost and LightGBM still have great potential. Both multi-model fusion(hard voting) and multi-model fusion(soft voting) have improved the anti-fitting performance to some extent. Preoperative albumin, perioperative blood loss, preoperative platelet count, duration of surgery, and preoperative aspartate aminotransferase are the five most important clinical characteristics.
Retrospectively registered:Ethics Y(2022)130; 2022/09/17
... ML models have shown AUCs between 0.63 and 0.99 for predicting the course of disease, whereas accuracies have been demonstrated to range from 73% to 99%. Eight studies have applied ML to predict postoperative liver function and complications in patients that underwent hepatectomy [43][44][45][46][47][48][49][50] . In predicting postoperative liver function and complications, ML models have demonstrated AUCs ranging from 0.63 to 0.89, and accuracies between 73% and 89% have also been reported. ...
Aim: The aim of this systematic review was to provide an overview of Machine Learning applications within hepatopancreaticobiliary surgery. The secondary aim was to evaluate the predictive performances of applied Machine Learning models.
Methods: A systematic search was conducted in PubMed, EMBASE, Cochrane, and Web of Science. Studies were only eligible for inclusion when they described Machine Learning in hepatopancreaticobiliary surgery. The Cochrane and PROBAST risk of bias tools were used to evaluate the quality of studies and included Machine Learning models.
Results: Out of 1821 articles, 52 studies have met the inclusion criteria. The majority of Machine Learning models were developed to predict the course of disease, and postoperative complications. The course of disease has been predicted with accuracies up to 99%, and postoperative complications with accuracies up to 89%. Most studies had a retrospective study design, in which external validation was absent for Machine Learning models.
Conclusion: Machine learning models have shown promising accuracies in the prediction of short-term and long-term surgical outcomes after hepatopancreaticobiliary surgery. External validation of Machine Learning models is required to facilitate the clinical introduction of Machine Learning.
... Moreover, classical regression analysis was generally used which does not account for non-linear relationships between the variables and the expected outcomes. 8 Recently, machine learning (ML) algorithms can be applied to large clinical datasets to gain extensive insight into the correlation between the different features and risk factors that influence disease progression and recurrence. As a result, the most crucial variables to the possibility of recurrence can be identified. ...
... Presently, many intubation difficulties in thyroid surgery patients cannot be predicted in advance, due to limited predictive tools (2,3). Machine learning has been applied to several medical fields, including cancer, pulmonary complications, chronic pain, and mental health (4)(5)(6)(7). A study of patients with obesity has demonstrated that machine learning can help predict difficult intubations: Among the six machine learning algorithms, only three can predict intubation difficulty in patients with obesity, and the Xgbc algorithm has the best comprehensive performance, with an accuracy rate exceeding 80% (8). ...
Background
In this paper, we examine whether machine learning and deep learning can be used to predict difficult airway intubation in patients undergoing thyroid surgery.
Methods
We used 10 machine learning and deep learning algorithms to establish a corresponding model through a training group, and then verify the results in a test group. We used R for the statistical analysis and constructed the machine learning prediction model in Python.
Results
The top 5 weighting factors for difficult airways identified by the average algorithm in machine learning were age, sex, weight, height, and BMI. In the training group, the AUC values and accuracy and the Gradient Boosting precision were 0.932, 0.929, and 100%, respectively. As for the modeled effects of predicting difficult airways in test groups, among the models constructed by the 10 algorithms, the three algorithms with the highest AUC values were Gradient Boosting, CNN, and LGBM, with values of 0.848, 0.836, and 0.812, respectively; In addition, among the algorithms, Gradient Boosting had the highest accuracy with a value of 0.913; Additionally, among the algorithms, the Gradient Boosting algorithm had the highest precision with a value of 100%.
Conclusion
According to our results, Gradient Boosting performed best overall, with an AUC >0.8, an accuracy >90%, and a precision of 100%. Besides, the top 5 weighting factors identified by the average algorithm in machine learning for difficult airways were age, sex, weight, height, and BMI.
... This data abundance enables the use of machine learning in medicine and healthcare. The uses of machine learning in medical research are as follows: prediction of Alzheimer's disease [7], prediction of kidney transplant survival [8], prediction of late-onset preeclampsia [9], prediction of pulmonary hypertension [10], detection of a skin lesion in the diagnosis of melanoma [11], detection of subclinical depressive symptoms among healthy individuals [12], and prediction of kidney injury after liver cancer resection [13]. Additionally, various types of breast cancer research have been conducted, and the possible use of machine learning in the area is endless. ...
Background
Rapid advancement in computing technology and digital information leads to the possible use of machine learning on breast cancer.
Objective
This study aimed to evaluate the research output of the top 100 publications and further identify a research theme of breast cancer and machine-learning studies.
Methods
Databases of Scopus and Web of Science were used to extract the top 100 publications. These publications were filtered based on the total citation of each paper. Additionally, a bibliometric analysis was applied to the top 100 publications.
Results
The top 100 publications were published between 1993 and 2019. The most productive author was Giger ML, and the top two institutions were the University of Chicago and the National University of Singapore. The most active countries were the USA, Germany and China. Ten clusters were identified as both basic and specialised themes of breast cancer and machine learning.
Conclusion
Various countries demonstrated comparable interest in breast cancer and machine-learning research. A few Asian countries, such as China, India and Singapore, were listed in the top 10 countries based on the total citation. Additionally, the use of deep learning and breast imaging data was trending in the past 10 years in the field of breast cancer and machine-learning research.
... Notably, AUCs for algorithms based on machine learning were in the range of 0.628-0.772 [39]. On the contrary, an AUC of nearly 0.90 was found for urine neutrophil gelatinase-associated lipocalin (NGAL) concentration assessed 12 h postoperatively, yet these results seem to be related to early detection of AKI rather than prediction of its development [40]. ...
Background
Skin autofluorescence (SAF) reflects accumulation of advanced glycation end-products (AGEs). The aim of this study was to evaluate predictive usefulness of SAF measurement in prediction of acute kidney injury (AKI) after liver resection.
Methods
This prospective observational study included 130 patients undergoing liver resection. The primary outcome measure was AKI. SAF was measured preoperatively and expressed in arbitrary units (AU).
Results
AKI was observed in 32 of 130 patients (24.6%). SAF independently predicted AKI (p = 0.047), along with extent of resection (p = 0.019) and operative time (p = 0.046). Optimal cut-off for SAF in prediction of AKI was 2.7 AU (area under the curve [AUC] 0.611), with AKI rates of 38.7% and 20.2% in patients with high and low SAF, respectively (p = 0.037). Score based on 3 independent predictors (SAF, extent of resection, and operative time) well stratified the risk of AKI (AUC 0.756), with positive and negative predictive values of 59.3% and 84.0%, respectively. In particular, SAF predicted AKI in patients undergoing major and prolonged resections (p = 0.010, AUC 0.733) with positive and negative predictive values of 81.8%, and 62.5%, respectively.
Conclusions
AGEs accumulation negatively affects renal function in patients undergoing liver resection. SAF measurement may be used to predict AKI after liver resection, particularly in high-risk patients.
... According to the global estimation in 2018, liver cancer ranked the 6th most generally diagnosed cancers and the 4th most malignant carcinomas causing death, 2 among which 46.6% of cases were located in China. 3 Though not featured among the top 10 cancers, gallbladder cancer also had a high annual rate in Chile, Canada and some Asian countries. 4 Most of the hepatobiliary cancer patients experienced different antitumor treatments depending on their cancer progression, including surgeries (like hepatectomy, liver transplantation and cholecystectomy), chemotherapy, immunotherapy and palliative care. ...
Background
Based on the admission data, we applied the XGBoost algorithm to create a prediction model to estimate the AKI risk in patients with hepatobiliary malignancies and then compare its prediction capacity with the logistic model.
Methods
We reviewed clinical data of 7968 and 589 liver/gallbladder cancer patients admitted to Zhongshan Hospital during 2014 and 2015. They were randomly divided into the training set and test set. Data were collected from the electronic medical record system. XGBoost and LASSO-logistic were used to develop prediction models, respectively. The performance measures included the classification matrix, the area under the receiver operating characteristic curve (AUC), lift chart and learning curve.
Results
Of 6846 participants in the training set, 792 (11.6%) cases developed AKI. In XGBoost model, the top 3 most important variables for AKI were serum creatinine (SCr), glomerular filtration rate (eGFR) and antitumor treatment in liver cancer patients. Similarly, SCr and eGFR also ranked second and third most important variables in the gallbladder cancer-related AKI model just after phosphorus. In the classification matrix, XGBoost model possessed a comparably better agreement between the actual observations and the predictions than LASSO-logistic model. The Youden’s index of XGBoost model was 47.5% and 59.3%, respectively, which was significantly higher than that of LASSO-logistic model (41.6% and 32.7%). The AUCs of XGBoost model were 0.822 in liver cancer and 0.850 in gallbladder cancer. By comparison, the AUC values of Logistic models were significantly lower as 0.793 and 0.740 (p=0.024 and 0.018). With the accumulation of training samples, XGBoost model maintained greater robustness in the learning curve.
Conclusion
XGBoost model based on admission data has higher accuracy and stronger robustness in predicting AKI. It will benefit AKI risk classification management in clinical practice and take an advanced intervention among patients with hepatobiliary malignancies.
