Prediction of delayed graft function after kidney transplantation: comparison between logistic regression and machine learning methods

Cherry on Top or Real Need? A Review of Explainable Machine Learning in Kidney Transplantation

Research on solid organ transplantation has taken advantage of the substantial acquisition of medical data and the use of artificial intelligence (AI) and machine learning (ML) to answer diagnostic, prognostic, and therapeutic questions for many years. Nevertheless, despite the question of whether AI models add value to traditional modeling approaches, such as regression models, their "black box" nature is one of the factors that have hindered the translation from research to clinical practice. Several techniques that make such models understandable to humans were developed with the promise of increasing transparency in the support of medical decision-making. These techniques should help AI to close the gap between theory and practice by yielding trust in the model by doctors and patients, allowing model auditing, and facilitating compliance with emergent AI regulations. But is this also happening in the field of kidney transplantation? This review reports the use and explanation of "black box" models to diagnose and predict kidney allograft rejection, delayed graft function, graft failure, and other related outcomes after kidney transplantation. In particular, we emphasize the discussion on the need (or not) to explain ML models for biological discovery and clinical implementation in kidney transplantation. We also discuss promising future research paths for these computational tools.

Using Regression Equations to Enhance Interpretation of Histology Lesions of Kidney Transplant Rejection

BACKGROUND

The Banff system for histologic diagnosis of rejection in kidney transplant biopsies uses guidelines to assess designated features-lesions, donor-specific antibody (DSA), and C4d staining. We explored whether using regression equations to interpret the features as well as current guidelines could establish the relative importance of each feature and improve histologic interpretation.

METHODS

We developed logistic regression equations using the designated features to predict antibody-mediated rejection (AMR/mixed) and T-cell-mediated rejection (TCMR/mixed) in 1679 indication biopsies from the INTERCOMEX study ( ClinicalTrials.gov NCT01299168). Equations were trained on molecular diagnoses independent of the designated features.

RESULTS

In regression and random forests, the important features predicting molecular rejection were as follows: for AMR, ptc and g, followed by cg; for TCMR, t > i. V-lesions were relatively unimportant. C4d and DSA were also relatively unimportant for predicting AMR: by AUC, the model excluding them (0.853) was nearly as good as the model including them (0.860). Including time posttransplant slightly but significantly improved all models. By AUC, regression predicted molecular AMR and TCMR better than Banff histologic diagnoses. More importantly, in biopsies called "no rejection" by Banff guidelines, regression equations based on histology features identified histologic and molecular rejection-related changes in some biopsies and improved survival predictions. Thus, regression can screen for missed rejection.

CONCLUSIONS

Using lesion-based regression equations in addition to Banff histology guidelines defines the relative important of histology features for identifying rejection, allows screening for potential missed diagnoses, and permits early estimates of AMR when C4d and DSA are not available.

Machine learning approaches for early detection of non-alcoholic steatohepatitis based on clinical and blood parameters

This study aims to develop a machine learning approach leveraging clinical data and blood parameters to predict non-alcoholic steatohepatitis (NASH) based on the NAFLD Activity Score (NAS). Using a dataset of 181 patients, we performed preprocessing including normalization and categorical encoding. To identify predictive features, we applied sequential forward selection (SFS), chi-square, analysis of variance (ANOVA), and mutual information (MI). The selected features were used to train machine learning classifiers including SVM, random forest, AdaBoost, LightGBM, and XGBoost. Hyperparameter tuning was done for each classifier using randomized search. Model evaluation was performed using leave-one-out cross-validation over 100 repetitions. Among the classifiers, random forest, combined with SFS feature selection and 10 features, obtained the best performance: Accuracy: 81.32% ± 6.43%, Sensitivity: 86.04% ± 6.21%, Specificity: 70.49% ± 8.12% Precision: 81.59% ± 6.23%, and F1-score: 83.75% ± 6.23% percent. Our findings highlight the promise of machine learning in enhancing early diagnosis of NASH and provide a compelling alternative to conventional diagnostic techniques. Consequently, this study highlights the promise of machine learning techniques in enhancing early and non-invasive diagnosis of NASH based on readily available clinical and blood data. Our findings provide the basis for developing scalable approaches that can improve screening and monitoring of NASH progression.

Predicting long-term outcomes of kidney transplantation in the era of artificial intelligence

The ability to accurately predict long-term kidney transplant survival can assist nephrologists in making therapeutic decisions. However, predicting kidney transplantation (KT) outcomes is challenging due to the complexity of the factors involved. Artificial intelligence (AI) has become an increasingly important tool in the prediction of medical outcomes. Our goal was to utilize both conventional and AI-based methods to predict long-term kidney transplant survival. Our study included 407 KTs divided into two groups (group A: with a graft lifespan greater than 5 years and group B: with poor graft survival). We first performed a traditional statistical analysis and then developed predictive models using machine learning (ML) techniques. Donors in group A were significantly younger. The use of Mycophenolate Mofetil (MMF) was the only immunosuppressive drug that was significantly associated with improved graft survival. The average estimated glomerular filtration rate (eGFR) in the 3rd month post-KT was significantly higher in group A. The number of hospital readmissions during the 1st year post-KT was a predictor of graft survival. In terms of early post-transplant complications, delayed graft function (DGF), acute kidney injury (AKI), and acute rejection (AR) were significantly associated with poor graft survival. Among the 35 AI models developed, the best model had an AUC of 89.7% (Se: 91.9%; Sp: 87.5%). It was based on ten variables selected by an ML algorithm, with the most important being hypertension and a history of red-blood-cell transfusion. The use of AI provided us with a robust model enabling fast and precise prediction of 5-year graft survival using early and easily collectible variables. Our model can be used as a decision-support tool to early detect graft status.

An innovative EEG-based emotion recognition using a single channel-specific feature from the brain rhythm code method

Introduction

Efficiently recognizing emotions is a critical pursuit in brain-computer interface (BCI), as it has many applications for intelligent healthcare services. In this work, an innovative approach inspired by the genetic code in bioinformatics, which utilizes brain rhythm code features consisting of δ, θ, α, β, or γ, is proposed for electroencephalography (EEG)-based emotion recognition.

Methods

These features are first extracted from the sequencing technique. After evaluating them using four conventional machine learning classifiers, an optimal channel-specific feature that produces the highest accuracy in each emotional case is identified, so emotion recognition through minimal data is realized. By doing so, the complexity of emotion recognition can be significantly reduced, making it more achievable for practical hardware setups.

Results

The best classification accuracies achieved for the DEAP and MAHNOB datasets range from 83-92%, and for the SEED dataset, it is 78%. The experimental results are impressive, considering the minimal data employed. Further investigation of the optimal features shows that their representative channels are primarily on the frontal region, and associated rhythmic characteristics are typical of multiple kinds. Additionally, individual differences are found, as the optimal feature varies with subjects.

Discussion

Compared to previous studies, this work provides insights into designing portable devices, as only one electrode is appropriate to generate satisfactory performances. Consequently, it would advance the understanding of brain rhythms, which offers an innovative solution for classifying EEG signals in diverse BCI applications, including emotion recognition.

A Machine Learning Prediction Model for Immediate Graft Function After Deceased Donor Kidney Transplantation

BACKGROUND

After kidney transplantation (KTx), the graft can evolve from excellent immediate graft function (IGF) to total absence of function requiring dialysis. Recipients with IGF do not seem to benefit from using machine perfusion, an expensive procedure, in the long term when compared with cold storage. This study proposes to develop a prediction model for IGF in KTx deceased donor patients using machine learning algorithms.

METHODS

Unsensitized recipients who received their first KTx deceased donor between January 1, 2010, and December 31, 2019, were classified according to the conduct of renal function after transplantation. Variables related to the donor, recipient, kidney preservation, and immunology were used. The patients were randomly divided into 2 groups: 70% were assigned to the training and 30% to the test group. Popular machine learning algorithms were used: eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine, Gradient Boosting classifier, Logistic Regression, CatBoost classifier, AdaBoost classifier, and Random Forest classifier. Comparative performance analysis on the test dataset was performed using the results of the AUC values, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score.

RESULTS

Of the 859 patients, 21.7% (n = 186) had IGF. The best predictive performance resulted from the eXtreme Gradient Boosting model (AUC, 0.78; 95% CI, 0.71-0.84; sensitivity, 0.64; specificity, 0.78). Five variables with the highest predictive value were identified.

CONCLUSIONS

Our results indicated the possibility of creating a model for the prediction of IGF, enhancing the selection of patients who would benefit from an expensive treatment, as in the case of machine perfusion preservation.

Prediction of Prednisolone Dose Correction Using Machine Learning

Wrong dose, a common prescription error, can cause serious patient harm, especially in the case of high-risk drugs like oral corticosteroids. This study aims to build a machine learning model to predict dose-related prescription modifications for oral prednisolone tablets (i.e., highly imbalanced data with very few positive cases). Prescription data were obtained from the electronic medical records at a single institute. Cluster analysis classified the clinical departments into six clusters with similar patterns of prednisolone prescription. Two patterns of training datasets were created with/without preprocessing by the SMOTE method. Five ML models (SVM, KNN, GB, RF, and BRF) and logistic regression (LR) models were constructed by Python. The model was internally validated by five-fold stratified cross-validation and was validated with a 30% holdout test dataset. Eighty-two thousand five hundred fifty-three prescribing data for prednisolone tablets containing 135 dose-corrected positive cases were obtained. In the original dataset (without SMOTE), only the BRF model showed a good performance (in test dataset, ROC-AUC:0.917, recall: 0.951). In the training dataset preprocessed by SMOTE, performance was improved on all models. The highest performance models with SMOTE were SVM (in test dataset, ROC-AUC: 0.820, recall: 0.659) and BRF (ROC-AUC: 0.814, recall: 0.634). Although the prescribing data for dose-related collection are highly imbalanced, various techniques such as the following have allowed us to build high-performance prediction models: data preprocessing by SMOTE, stratified cross-validation, and BRF classifier corresponding to imbalanced data. ML is useful in complicated dose audits such as oral prednisolone.

Supplementary Information

The online version contains supplementary material available at 10.1007/s41666-023-00128-3.

Toward generalizing the use of artificial intelligence in nephrology and kidney transplantation

With its robust ability to integrate and learn from large sets of clinical data, artificial intelligence (AI) can now play a role in diagnosis, clinical decision making, and personalized medicine. It is probably the natural progression of traditional statistical techniques. Currently, there are many unmet needs in nephrology and, more particularly, in the kidney transplantation (KT) field. The complexity and increase in the amount of data, and the multitude of nephrology registries worldwide have enabled the explosive use of AI within the field. Nephrologists in many countries are already at the center of experiments and advances in this cutting-edge technology and our aim is to generalize the use of AI among nephrologists worldwide. In this paper, we provide an overview of AI from a medical perspective. We cover the core concepts of AI relevant to the practicing nephrologist in a consistent and simple way to help them get started, and we discuss the technical challenges. Finally, we focus on the KT field: the unmet needs and the potential role that AI can play to fill these gaps, then we summarize the published KT-related studies, including predictive factors used in each study, which will allow researchers to quickly focus on the most relevant issues.

A Machine Learning-Based Approach to Predict Prognosis and Length of Hospital Stay in Adults and Children With Traumatic Brain Injury: Retrospective Cohort Study

BACKGROUND

The treatment and care of adults and children with traumatic brain injury (TBI) constitute an intractable global health problem. Predicting the prognosis and length of hospital stay of patients with TBI may improve therapeutic effects and significantly reduce societal health care burden. Applying novel machine learning methods to the field of TBI may be valuable for determining the prognosis and cost-effectiveness of clinical treatment.

OBJECTIVE

We aimed to combine multiple machine learning approaches to build hybrid models for predicting the prognosis and length of hospital stay for adults and children with TBI.

METHODS

We collected relevant clinical information from patients treated at the Neurosurgery Center of the Second Affiliated Hospital of Anhui Medical University between May 2017 and May 2022, of which 80% was used for training the model and 20% for testing via screening and data splitting. We trained and tested the machine learning models using 5 cross-validations to avoid overfitting. In the machine learning models, 11 types of independent variables were used as input variables and Glasgow Outcome Scale score, used to evaluate patients' prognosis, and patient length of stay were used as output variables. Once the models were trained, we obtained and compared the errors of each machine learning model from 5 rounds of cross-validation to select the best predictive model. The model was then externally tested using clinical data of patients treated at the First Affiliated Hospital of Anhui Medical University from June 2021 to February 2022.

RESULTS

The final convolutional neural network-support vector machine (CNN-SVM) model predicted Glasgow Outcome Scale score with an accuracy of 93% and 93.69% in the test and external validation sets, respectively, and an area under the curve of 94.68% and 94.32% in the test and external validation sets, respectively. The mean absolute percentage error of the final built convolutional neural network-support vector regression (CNN-SVR) model predicting inpatient time in the test set and external validation set was 10.72% and 10.44%, respectively. The coefficient of determination (R²) was 0.93 and 0.92 in the test set and external validation set, respectively. Compared with back-propagation neural network, CNN, and SVM models built separately, our hybrid model was identified to be optimal and had high confidence.

CONCLUSIONS

This study demonstrates the clinical utility of 2 hybrid models built by combining multiple machine learning approaches to accurately predict the prognosis and length of stay in hospital for adults and children with TBI. Application of these models may reduce the burden on physicians when assessing TBI and assist clinicians in the medical decision-making process.

Predicting In-Hospital Mortality After Acute Myeloid Leukemia Therapy: Through Supervised Machine Learning Algorithms

PURPOSE

Despite careful patient selection, induction chemotherapy for acute myeloid leukemia (AML) is associated with a considerable risk for treatment-related mortality (5%-20%). We evaluated machine learning (ML) algorithms trained using factors available at the time of admission for AML therapy to predict death during the hospitalization.

METHODS

We included AML discharges with age > 17 years who received inpatient chemotherapy from State Inpatient Database from Arizona, Florida, New York, Maryland, Washington, and New Jersey for years 2008-2014. The primary objective was to predict inpatient mortality in patients undergoing chemotherapy using covariates present before initiation of chemotherapy. ML algorithms logistic regression (LR), decision tree, and random forest were compared.

RESULTS

29,613 hospitalizations for patients with AML were included in the analysis each with 4,177 features. The median age was 58.9 (18-101) years, 13,689 (53.7%) were male, and 20,203 (69%) were White. The mean time from admission to chemotherapy was 3 days (95% CI, 2.9 to 3.1), and 2,682 (9.1%) died during the hospitalization. Both LR and random forest models achieved an area under the curve (AUC) score of 0.78, whereas decision tree achieved an AUC of 0.70. The baseline LR model with age yielded an AUC of 0.62. To clinically balance and minimize false positives, we selected a decision threshold of 0.7 and at this threshold, 51 of our test set of 5,923 could have potentially averted treatment-related mortality.

CONCLUSION

Using readily accessible variables, inpatient mortality of patients on track for chemotherapy to treat AML can be predicted through ML algorithms. The model also predicted inpatient mortality when tested on different data representations and paves the way for future research.

Identification of the Optimal Model for the Prediction of Diabetic Retinopathy in Chinese Rural Population: Handan Eye Study

Background

To identify an optimal model for diabetic retinopathy (DR) prediction in Chinese rural population by establishing and comparing different algorithms based on the data from Handan Eye Study (HES).

Methods

Five algorithms, including multivariable logistic regression (MLR), classification and regression trees (C&RT), support vector machine (SVM), random forests (RF), and gradient boosting machine (GBM), were used to establish DR prediction models with HES data. The performance of the models was assessed based on the adjusted area under the ROC curve (AUROC), sensitivity, specificity, and accuracy.

Results

The data on 4752 subjects were used to build the DR prediction model, and among them, 198 patients were diagnosed with DR. The age of the included subjects ranged from 30 to 85 years old, with an average age of 50.9 years (SD = 3.04). The kappa coefficient of the diagnosis between the two ophthalmologists was 0.857. The MLR model revealed that blood glucose, systolic blood pressure, and body mass index were independently associated with the development of DR. The AUROC obtained by GBM (0.952), RF (0.949), and MLR (0.936) was similar and statistically larger than that of CART (0.682) and SVM (0.765).

Conclusions

The MLR model exhibited excellent prediction performance and visible equation and thus was the optimal model for DR prediction. Therefore, the MLR model may have the potential to serve as a complementary screening tool for the early detection of DR, especially in remote and underserved areas.

Predicting Characteristics Associated with Breast Cancer Survival Using Multiple Machine Learning Approaches

Breast cancer is one of the most commonly diagnosed female disorders globally. Numerous studies have been conducted to predict survival markers, although the majority of these analyses were conducted using simple statistical techniques. In lieu of that, this research employed machine learning approaches to develop models for identifying and visualizing relevant prognostic indications of breast cancer survival rates. A comprehensive hospital-based breast cancer dataset was collected from the National Cancer Institute's SEER Program's November 2017 update, which offers population-based cancer statistics. The dataset included female patients diagnosed between 2006 and 2010 with infiltrating duct and lobular carcinoma breast cancer (SEER primary cites recode NOS histology codes 8522/3). The dataset included nine predictor factors and one predictor variable that were linked to the patients' survival status (alive or dead). To identify important prognostic markers associated with breast cancer survival rates, prediction models were constructed using K-nearest neighbor (K-NN), decision tree (DT), gradient boosting (GB), random forest (RF), AdaBoost, logistic regression (LR), voting classifier, and support vector machine (SVM). All methods yielded close results in terms of model accuracy and calibration measures, with the lowest achieved from logistic regression (accuracy = 80.57 percent) and the greatest acquired from the random forest (accuracy = 94.64 percent). Notably, the multiple machine learning algorithms utilized in this research achieved high accuracy, suggesting that these approaches might be used as alternative prognostic tools in breast cancer survival studies, especially in the Asian area.

Pre-Transplant Expression of CCR-2 in Kidney Transplant Recipients Is Associated With the Development of Delayed Graft Function

Background

Delayed graft function (DGF) leads to a reduced graft survival. Donors’ features have been always considered as key pathogenic factors in this setting. The aim of our study was to evaluate the recipients’ characteristics in the development of DGF.

Methods

We enrolled 932 kidney graft recipients from 466 donors; 226 recipients experienced DGF. In 290 donors, both recipients presented with early graft function (EGF, group A), in 50 both recipients experienced DGF (group B), and in 126 one recipient presented with DGF and the other with EGF (group C). In group C, we selected 7 couples of DGF/EGF recipients and we evaluated the transcriptomic profile by microarray on circulating mononuclear cells harvested before transplantation. Results were validated by qPCR in an independent group of 25 EGF/DGF couples.

Findings

In the whole study group, DGF was associated with clinical characteristics related to both donors and recipient. In group C, DGF was significantly associated with body mass index, hemodialysis, and number of mismatches. In the same group, we identified 411 genes differently expressed before transplantation between recipients discordant for the transplant outcome. Those genes were involved in immune dysfunction and inflammation. In particular, we observed a significant increase in DGF patients in the expression of C–C chemokine receptor type 2 (CCR2), the monocyte chemoattractant protein-1 (MCP-1) receptor. CCR-2 upregulation was confirmed in an independent cohort of patients.

Conclusions

Our results suggest that recipients’ clinical/immunological features, potentially modulated by dialysis, are associated with the development of DGF independently of donors’ features.

Artificial Intelligence-A Tool for Risk Assessment of Delayed-Graft Function in Kidney Transplant

Delayed-graft function (DGF) might be responsible for shorter graft survival. Therefore, a clinical tool predicting its occurrence is vital for the risk assessment of transplant outcomes. In a single-center study, we conducted data mining and machine learning experiments, resulting in DGF predictive models based on random forest classifiers (RF) and an artificial neural network called multi-layer perceptron (MLP). All designed models had four common input parameters, determining the best accuracy and discriminant ability: donor’s eGFR, recipient’s BMI, donor’s BMI, and recipient–donor weight difference. RF and MLP designs, using these parameters, achieved an accuracy of 84.38% and an area under curve (AUC) 0.84. The model additionally implementing a donor’s age, gender, and Kidney Donor Profile Index (KDPI) accomplished an accuracy of 93.75% and an AUC of 0.91. The other configuration with the estimated post-transplant survival (EPTS) and the kidney donor risk profile (KDRI) achieved an accuracy of 93.75% and an AUC of 0.92. Using machine learning, we were able to assess the risk of DGF in recipients after kidney transplant from a deceased donor. Our solution is scalable and can be improved during subsequent transplants. Based on the new data, the models can achieve better outcomes.

Predicting Kidney Discard Using Machine Learning

BACKGROUND

Despite the kidney supply shortage, 18%-20% of deceased donor kidneys are discarded annually in the United States. In 2018, 3569 kidneys were discarded.

METHODS

We compared machine learning (ML) techniques to identify kidneys at risk of discard at the time of match run and after biopsy and machine perfusion results become available. The cohort consisted of adult deceased donor kidneys donated between December 4, 2014, and July 1, 2019. The studied ML models included Random Forests (RF), Adaptive Boosting (AdaBoost), Neural Networks (NNet), Support Vector Machines (SVM), and K-nearest Neighbors (KNN). In addition, a Logistic Regression (LR) model was fitted and used for comparison with the ML models' performance.

RESULTS

RF outperformed other ML models. Of 8036 discarded kidneys in the test dataset, LR correctly classified 3422 kidneys, whereas RF correctly classified 4762 kidneys (area under the receiver operative curve [AUC]: 0.85 versus 0.888, and balanced accuracy: 0.681 versus 0.759). For the kidneys with kidney donor profile index of >85% (6079 total), RF significantly outperformed LR in classifying discard and transplant prediction (AUC: 0.814 versus 0.717, and balanced accuracy: 0.732 versus 0.657). More than 388 kidneys were correctly classified using RF. Including biopsy and machine perfusion variables improved the performance of LR and RF (LR's AUC: 0.888 and balanced accuracy: 0.74 versus RF's AUC: 0.904 and balanced accuracy: 0.775).

CONCLUSIONS

Kidneys that are at risk of discard can be more accurately identified using ML techniques such as RF.

The Future Role of Machine Learning in Clinical Transplantation

The use of artificial intelligence and machine learning (ML) has revolutionized our daily lives and will soon be instrumental in healthcare delivery. The rise of ML is due to multiple factors: increasing access to massive datasets, exponential increases in processing power, and key algorithmic developments that allow ML models to tackle increasingly challenging questions. Progressively more transplantation research is exploring the potential utility of ML models throughout the patient journey, although this has not yet widely transitioned into the clinical domain. In this review, we explore common approaches used in ML in solid organ clinical transplantation and consider opportunities for ML to help clinicians and patients. We discuss ways in which ML can aid leverage of large complex datasets, generate cutting-edge prediction models, perform clinical image analysis, discover novel markers in molecular data, and fuse datasets to generate novel insights in modern transplantation practice. We focus on key areas in transplantation in which ML is driving progress, explore the future potential roles of ML, and discuss the challenges and limitations of these powerful tools.

Artificial intelligence and kidney transplantation

Artificial intelligence and its primary subfield, machine learning, have started to gain widespread use in medicine, including the field of kidney transplantation. We made a review of the literature that used artificial intelligence techniques in kidney transplantation. We located six main areas of kidney transplantation that artificial intelligence studies are focused on: Radiological evaluation of the allograft, pathological evaluation including molecular evaluation of the tissue, prediction of graft survival, optimizing the dose of immunosuppression, diagnosis of rejection, and prediction of early graft function. Machine learning techniques provide increased automation leading to faster evaluation and standardization, and show better performance compared to traditional statistical analysis. Artificial intelligence leads to improved computer-aided diagnostics and quantifiable personalized predictions that will improve personalized patient care.

Machine Learning and Surgical Outcomes Prediction: A Systematic Review

BACKGROUND

Machine learning (ML) has garnered increasing attention as a means to quantitatively analyze the growing and complex medical data to improve individualized patient care. We herein aim to critically examine the current state of ML in predicting surgical outcomes, evaluate the quality of currently available research, and propose areas of improvement for future uses of ML in surgery.

METHODS

A systematic review was conducted in accordance with the Preferred Reporting Items for a Systematic Review and Meta-Analysis (PRISMA) checklist. PubMed, MEDLINE, and Embase databases were reviewed under search syntax "machine learning" and "surgery" for papers published between 2015 and 2020.

RESULTS

Of the initial 2677 studies, 45 papers met inclusion and exclusion criteria. Fourteen different subspecialties were represented with neurosurgery being most common. The most frequently used ML algorithms were random forest (n = 19), artificial neural network (n = 17), and logistic regression (n = 17). Common outcomes included postoperative mortality, complications, patient reported quality of life and pain improvement. All studies which compared ML algorithms to conventional studies which used area under the curve (AUC) to measure accuracy found improved outcome prediction with ML models.

CONCLUSIONS

While still in its early stages, ML models offer surgeons an opportunity to capitalize on the myriad of clinical data available and improve individualized patient care. Limitations included heterogeneous outcome and imperfect quality of some of the papers. We therefore urge future research to agree upon methods of outcome reporting and require basic quality standards.

Toward Advancing Long-Term Outcomes of Kidney Transplantation with Artificial Intelligence

After decades of pioneering advances and improvements, kidney transplantation is now the renal replacement therapy of choice for most patients with end-stage kidney disease (ESKD). Despite this success, the high risk of premature death and frequent occurrence of graft failure remain important clinical and research challenges. The current burst of studies and other innovative initiatives using artificial intelligence (AI) for a wide range of analytical and practical applications in biomedical areas seems to correlate with the same trend observed in publications in the kidney transplantation field, and points toward the potential of such novel approaches to address the aforementioned aim of improving long-term outcomes of kidney transplant recipients (KTR). However, at the same time, this trend underscores now more than ever the old methodological challenges and potential threats that the research and clinical community needs to be aware of and actively look after with regard to AI-driven evidence. The purpose of this narrative mini-review is to explore challenges for obtaining applicable and adequate kidney transplant data for analyses using AI techniques to develop prediction models, and to propose next steps in the field. We make a call to act toward establishing the strong collaborations needed to bring innovative synergies further augmented by AI, which have the potential to impact the long-term care of KTR. We encourage researchers and clinicians to submit their invaluable research, including original clinical and imaging studies, database studies from registries, meta-analyses, and AI research in the kidney transplantation field.

Risk prediction for malignant intraductal papillary mucinous neoplasm of the pancreas: logistic regression versus machine learning

Most models for predicting malignant pancreatic intraductal papillary mucinous neoplasms were developed based on logistic regression (LR) analysis. Our study aimed to develop risk prediction models using machine learning (ML) and LR techniques and compare their performances. This was a multinational, multi-institutional, retrospective study. Clinical variables including age, sex, main duct diameter, cyst size, mural nodule, and tumour location were factors considered for model development (MD). After the division into a MD set and a test set (2:1), the best ML and LR models were developed by training with the MD set using a tenfold cross validation. The test area under the receiver operating curves (AUCs) of the two models were calculated using an independent test set. A total of 3,708 patients were included. The stacked ensemble algorithm in the ML model and variable combinations containing all variables in the LR model were the most chosen during 200 repetitions. After 200 repetitions, the mean AUCs of the ML and LR models were comparable (0.725 vs. 0.725). The performances of the ML and LR models were comparable. The LR model was more practical than ML counterpart, because of its convenience in clinical use and simple interpretability.

Machine learning to predict transplant outcomes: helpful or hype? A national cohort study

An increasing number of studies claim machine learning (ML) predicts transplant outcomes more accurately. However, these claims were possibly confounded by other factors, namely, supplying new variables to ML models. To better understand the prospects of ML in transplantation, we compared ML to conventional regression in a "common" analytic task: predicting kidney transplant outcomes using national registry data. We studied 133 431 adult deceased-donor kidney transplant recipients between 2005 and 2017. Transplant centers were randomly divided into 70% training set (190 centers/97 787 recipients) and 30% validation set (82 centers/35 644 recipients). Using the training set, we performed regression and ML procedures [gradient boosting (GB) and random forests (RF)] to predict delayed graft function, one-year acute rejection, death-censored graft failure C, all-cause graft failure, and death. Their performances were compared on the validation set using -statistics. In predicting rejection, regression (C = 0.601 0.6110.621 ) actually outperformed GB (C = 0.581 0.5910.601 ) and RF (C = 0.569 0.5790.589 ). For all other outcomes, the C-statistics were nearly identical across methods (delayed graft function, 0.717-0.723; death-censored graft failure, 0.637-0.642; all-cause graft failure, 0.633-0.635; and death, 0.705-0.708). Given its shortcomings in model interpretability and hypothesis testing, ML is advantageous only when it clearly outperforms conventional regression; in the case of transplant outcomes prediction, ML seems more hype than helpful.

Personalized prediction of delayed graft function for recipients of deceased donor kidney transplants with machine learning

Machine learning (ML) has shown its potential to improve patient care over the last decade. In organ transplantation, delayed graft function (DGF) remains a major concern in deceased donor kidney transplantation (DDKT). To this end, we harnessed ML to build personalized prognostic models to predict DGF. Registry data were obtained on adult DDKT recipients for model development (n = 55,044) and validation (n = 6176). Incidence rates of DGF were 25.1% and 26.3% for the development and validation sets, respectively. Twenty-six predictors were identified via recursive feature elimination with random forest. Five widely-used ML algorithms-logistic regression (LR), elastic net, random forest, artificial neural network (ANN), and extreme gradient boosting (XGB) were trained and compared with a baseline LR model fitted with previously identified risk factors. The new ML models, particularly ANN with the area under the receiver operating characteristic curve (ROC-AUC) of 0.732 and XGB with ROC-AUC of 0.735, exhibited superior performance to the baseline model (ROC-AUC = 0.705). This study demonstrates the use of ML as a viable strategy to enable personalized risk quantification for medical applications. If successfully implemented, our models may aid in both risk quantification for DGF prevention clinical trials and personalized clinical decision making.

Machine learning is a valid method for predicting prehospital delay after acute ischemic stroke

OBJECTIVES

This study aimed to identify the influencing factors associated with long onset-to-door time and establish predictive models that could help to assess the probability of prehospital delay in populations with a high risk for stroke.

MATERIALS AND METHODS

Patients who were diagnosed with acute ischemic stroke (AIS) and hospitalized between 1 November 2018 and 31 July 2019 were interviewed, and their medical records were extracted for data analysis. Two machine learning algorithms (support vector machine and Bayesian network) were applied in this study, and their predictive performance was compared with that of the classical logistic regression models after using several variable selection methods. Timely admission (onset-to-door time < 3 hr) and prehospital delay (onset-to-door time ≥ 3 hr) were the outcome variables. We computed the area under curve (AUC) and the difference in the mean AUC values between the models.

RESULTS

A total of 450 patients with AIS were enrolled; 57 (12.7%) with timely admission and 393 (87.3%) patients with prehospital delay. All models, both those constructed by logistic regression and those by machine learning, performed well in predicting prehospital delay (range mean AUC: 0.800-0.846). The difference in the mean AUC values between the best performing machine learning model and the best performing logistic regression model was negligible (0.014; 95% CI: 0.013-0.015).

CONCLUSIONS

Machine learning algorithms were not inferior to logistic regression models for prediction of prehospital delay after stroke. All models provided good discrimination, thereby creating valuable diagnostic programs for prehospital delay prediction.

Using Artificial Intelligence Resources in Dialysis and Kidney Transplant Patients: A Literature Review

BACKGROUND

The purpose of this review is to depict current research and impact of artificial intelligence/machine learning (AI/ML) algorithms on dialysis and kidney transplantation. Published studies were presented from two points of view: What medical aspects were covered? What AI/ML algorithms have been used?

METHODS

We searched four electronic databases or studies that used AI/ML in hemodialysis (HD), peritoneal dialysis (PD), and kidney transplantation (KT). Sixty-nine studies were split into three categories: AI/ML and HD, PD, and KT, respectively. We identified 43 trials in the first group, 8 in the second, and 18 in the third. Then, studies were classified according to the type of algorithm.

RESULTS

AI and HD trials covered: (a) dialysis service management, (b) dialysis procedure, (c) anemia management, (d) hormonal/dietary issues, and (e) arteriovenous fistula assessment. PD studies were divided into (a) peritoneal technique issues, (b) infections, and (c) cardiovascular event prediction. AI in transplantation studies were allocated into (a) management systems (ML used as pretransplant organ-matching tools), (b) predicting graft rejection, (c) tacrolimus therapy modulation, and (d) dietary issues.

CONCLUSIONS

Although guidelines are reluctant to recommend AI implementation in daily practice, there is plenty of evidence that AI/ML algorithms can predict better than nephrologists: volumes, Kt/V, and hypotension or cardiovascular events during dialysis. Altogether, these trials report a robust impact of AI/ML on quality of life and survival in G5D/T patients. In the coming years, one would probably witness the emergence of AI/ML devices that facilitate the management of dialysis patients, thus increasing the quality of life and survival.

Predictive validity of radiographic signs of complete discoid lateral meniscus in children using machine learning techniques

The diagnostic utility of radiographic signs of complete discoid lateral meniscus remains controversial. This study aimed to investigate the diagnostic accuracy and determine which sign is most reliably detects the presence of a complete discoid lateral meniscus in children. A total of 141 knees (age 7-16) with complete discoid lateral meniscus and 141 age- and sex-matched knees with normal meniscus were included. The following radiographic signs were evaluated: lateral joint (LJ) space, fibular head (FH) height, lateral tibial spine (LTS) height, lateral tibial plateau (LTP) obliquity, lateral femoral condyle (LFC) squaring, LTP cupping, LFC notching, and prominence ratio of the femoral condyle. Prediction models were constructed using logistic regressions, decision trees, and random forest analyses. Receiver operating characteristic curves and area under the curve (AUC) were estimated to compare the diagnostic accuracy of the radiographic signs and model fit. The random forest model yielded the best diagnostic accuracy (AUC: 0.909), with 86.5% sensitivity and 82.2% specificity. LJ space height, FH height, and prominence ratio showed statistically large AUC compared with LTS height and LTP obliquity (P < .05 in all). The cut-off values for diagnosing discoid meniscus to be <12.55 mm for FH height, <0.804 for prominence ratio, and >6.6 mm for LJ space height when using the random forest model. On the basis of the results of this study, in clinical practice, LJ space height, FH height and prominence ratio could be easily used as supplementary tools for complete discoid lateral meniscus in children.

A 12-Lead ECG-Based System With Physiological Parameters and Machine Learning to Identify Right Ventricular Hypertrophy in Young Adults

OBJECTIVE

The presence of right ventricular hypertrophy (RVH) accounts for approximately 5-10% in young adults. The sensitivity estimated by commonly used 12-lead electrocardiographic (ECG) criteria for identifying the presence of RVH is under 20% in the general population. The aim of this study is to develop a 12-lead ECG system with the related information of age, body height and body weight via machine learning to increase the sensitivity and the precision for detecting RVH.

METHOD

In a sample of 1,701 males, aged 17-45 years, support vector machine is used for the training of 31 parameters including age, body height and body weight in addition to 28 ECG data such as axes, intervals and wave voltages as the inputs to link the output RVH. The RVH is defined on the echocardiographic finding for young males as right ventricular anterior wall thickness > 5.5 mm.

RESULTS

On the system goal for increasing sensitivity, the specificity is controlled around 70-75% and all data tested in the proposed method show competent sensitivity up to 70.3%. The values of area under curve of receiver operating characteristic curve and precision-recall curve using the proposed method are 0.780 and 0.285, respectively, which are better than 0.518 and 0.112 using the Sokolow-Lyon voltage criterion, respectively, for detecting unspecific RVH.

CONCLUSION

We present a method using simple physiological parameters with ECG data to effectively identify more than 70% of the RVH among young adults. Clinical Impact: This system provides a fast, precise and feasible diagnosis tool to screen RVH.

Prediction of the Mortality Risk in Peritoneal Dialysis Patients using Machine Learning Models: A Nation-wide Prospective Cohort in Korea

Herein, we aim to assess mortality risk prediction in peritoneal dialysis patients using machine-learning algorithms for proper prognosis prediction. A total of 1,730 peritoneal dialysis patients in the CRC for ESRD prospective cohort from 2008 to 2014 were enrolled in this study. Classification algorithms were used for prediction of N-year mortality including neural network. The survival hazard ratio was presented by machine-learning algorithms using survival statistics and was compared to conventional algorithms. A survival-tree algorithm presented the most accurate prediction model and outperformed a conventional method such as Cox regression (concordance index 0.769 vs 0.745). Among various survival decision-tree models, the modified Charlson Comorbidity index (mCCI) was selected as the best predictor of mortality. If peritoneal dialysis patients with high mCCI (>4) were aged ≥70.5 years old, the survival hazard ratio was predicted as 4.61 compared to the overall study population. Among the various algorithm using longitudinal data, the AUC value of logistic regression was augmented at 0.804. In addition, the deep neural network significantly improved performance to 0.841. We propose machine learning-based final model, mCCI and age were interrelated as notable risk factors for mortality in Korean peritoneal dialysis patients.

Machine learning algorithms performed no better than regression models for prognostication in traumatic brain injury

OBJECTIVE

We aimed to explore the added value of common machine learning (ML) algorithms for prediction of outcome for moderate and severe traumatic brain injury.

STUDY DESIGN AND SETTING

We performed logistic regression (LR), lasso regression, and ridge regression with key baseline predictors in the IMPACT-II database (15 studies, n = 11,022). ML algorithms included support vector machines, random forests, gradient boosting machines, and artificial neural networks and were trained using the same predictors. To assess generalizability of predictions, we performed internal, internal-external, and external validation on the recent CENTER-TBI study (patients with Glasgow Coma Scale <13, n = 1,554). Both calibration (calibration slope/intercept) and discrimination (area under the curve) was quantified.

RESULTS

In the IMPACT-II database, 3,332/11,022 (30%) died and 5,233(48%) had unfavorable outcome (Glasgow Outcome Scale less than 4). In the CENTER-TBI study, 348/1,554(29%) died and 651(54%) had unfavorable outcome. Discrimination and calibration varied widely between the studies and less so between the studied algorithms. The mean area under the curve was 0.82 for mortality and 0.77 for unfavorable outcomes in the CENTER-TBI study.

CONCLUSION

ML algorithms may not outperform traditional regression approaches in a low-dimensional setting for outcome prediction after moderate or severe traumatic brain injury. Similar to regression-based prediction models, ML algorithms should be rigorously validated to ensure applicability to new populations.

The impact of deceased donor maintenance on delayed kidney allograft function: A machine learning analysis

Background

This study evaluated the risk factors for delayed graft function (DGF) in a country where its incidence is high, detailing donor maintenance-related (DMR) variables and using machine learning (ML) methods beyond the traditional regression-based models.

Methods

A total of 443 brain dead deceased donor kidney transplants (KT) from two Brazilian centers were retrospectively analyzed and the following DMR were evaluated using predictive modeling: arterial blood gas pH, serum sodium, blood glucose, urine output, mean arterial pressure, vasopressors use, and reversed cardiac arrest.

Results

Most patients (95.7%) received kidneys from standard criteria donors. The incidence of DGF was 53%. In multivariable logistic regression analysis, DMR variables did not impact on DGF occurrence. In post-hoc analysis including only KT with cold ischemia time<21h (n = 220), urine output in 24h prior to recovery surgery (OR = 0.639, 95%CI 0.444–0.919) and serum sodium (OR = 1.030, 95%CI 1.052–1.379) were risk factors for DGF. Using elastic net regularized regression model and ML analysis (decision tree, neural network and support vector machine), urine output and other DMR variables emerged as DGF predictors: mean arterial pressure, ≥ 1 or high dose vasopressors and blood glucose.

Conclusions

Some DMR variables were associated with DGF, suggesting a potential impact of variables reflecting poor clinical and hemodynamic status on the incidence of DGF.

Using machine learning techniques to develop risk prediction models to predict graft failure following kidney transplantation: protocol for a retrospective cohort study

Background: A mechanism to predict graft failure before the actual kidney transplantation occurs is crucial to clinical management of chronic kidney disease patients.  Several kidney graft outcome prediction models, developed using machine learning methods, are available in the literature.  However, most of those models used small datasets and none of the machine learning-based prediction models available in the medical literature modelled time-to-event (survival) information, but instead used the binary outcome of failure or not. The objective of this study is to develop two separate machine learning-based predictive models to predict graft failure following live and deceased donor kidney transplant, using time-to-event data in a large national dataset from Australia.   Methods: The dataset provided by the Australia and New Zealand Dialysis and Transplant Registry will be used for the analysis. This retrospective dataset contains the cohort of patients who underwent a kidney transplant in Australia from January 1 st, 2007, to December 31 st, 2017.  This included 3,758 live donor transplants and 7,365 deceased donor transplants.  Three machine learning methods (survival tree, random survival forest and survival support vector machine) and one traditional regression method, Cox proportional regression, will be used to develop the two predictive models.  The best predictive model will be selected based on the model's performance. Discussion: This protocol describes the development of two separate machine learning-based predictive models to predict graft failure following live and deceased donor kidney transplant, using a large national dataset from Australia.   Furthermore, these two models will be the most comprehensive kidney graft failure predictive models that have used survival data to model using machine learning techniques.  Thus, these models are expected to provide valuable insight into the complex interactions between graft failure and donor and recipient characteristics.

Using machine learning techniques to develop risk prediction models to predict graft failure following kidney transplantation: protocol for a retrospective cohort study

Background: A mechanism to predict graft failure before the actual kidney transplantation occurs is crucial to clinical management of chronic kidney disease patients.  Several kidney graft outcome prediction models, developed using machine learning methods, are available in the literature.  However, most of those models used small datasets and none of the machine learning-based prediction models available in the medical literature modelled time-to-event (survival) information, but instead used the binary outcome of failure or not. The objective of this study is to develop two separate machine learning-based predictive models to predict graft failure following live and deceased donor kidney transplant, using time-to-event data in a large national dataset from Australia.   Methods: The dataset provided by the Australia and New Zealand Dialysis and Transplant Registry will be used for the analysis. This retrospective dataset contains the cohort of patients who underwent a kidney transplant in Australia from January 1 st, 2007, to December 31 st, 2017. This included 3,758 live donor transplants and 7,365 deceased donor transplants. Three machine learning methods (survival tree, random survival forest and survival support vector machine) and one traditional regression method, Cox proportional regression, will be used to develop the two predictive models (for live donor and deceased donor transplants). The best predictive model will be selected based on the model's performance. Discussion: This protocol describes the development of two separate machine learning-based predictive models to predict graft failure following live and deceased donor kidney transplant, using a large national dataset from Australia. Furthermore, these two models will be the most comprehensive kidney graft failure predictive models that have used survival data to model using machine learning techniques. Thus, these models are expected to provide valuable insight into the complex interactions between graft failure and donor and recipient characteristics.

Machine learning in predicting graft failure following kidney transplantation: A systematic review of published predictive models

INTRODUCTION

Machine learning has been increasingly used to develop predictive models to diagnose different disease conditions. The heterogeneity of the kidney transplant population makes predicting graft outcomes extremely challenging. Several kidney graft outcome prediction models have been developed using machine learning, and are available in the literature. However, a systematic review of machine learning based prediction methods applied to kidney transplant has not been done to date. The main aim of our study was to perform an in-depth systematic analysis of different machine learning methods used to predict graft outcomes among kidney transplant patients, and assess their usefulness as an aid to decision-making.

METHODS

A systemic review of machine learning methods used to predict graft outcomes among kidney transplant patients was carried out using a search of the Medline, the Cumulative Index to Nursing and Allied Health Literature, EMBASE, PsycINFO and Cochrane databases.

RESULTS

A total of 295 articles were identified and extracted. Of these, 18 met the inclusion criteria. Most of the studies were published in the United States after 2010. The population size used to develop the models varied from 80 to 92,844, and the number of features in the models ranged from 6 to 71. The most common machine learning methods used were artificial neural networks, decision trees and Bayesian belief networks. Most of the machine learning based predictive models predicted graft failure with high sensitivity and specificity. Only one machine learning based prediction model had modelled time-to-event (survival) information. Seven studies compared the predictive performance of machine learning models with traditional regression methods and the performance of machine learning methods was found to be mixed, when compared with traditional regression methods.

CONCLUSION

There was a wide variation in the size of the study population and the input variables used. However, the prediction accuracy provided mixed results when machine learning and traditional predictive methods are compared. Based on reported gains in predictive performance, machine learning has the potential to improve kidney transplant outcome prediction and aid medical decision making.

Comparison between logistic regression and machine learning algorithms on survival prediction of traumatic brain injuries

PURPOSE

To compare twenty-two machine learning (ML) models against logistic regression on survival prediction in severe traumatic brain injury (STBI) patients in a single center study.

MATERIALS AND METHODS

Data was collected from STBI patients admitted to the Sichuan Provincial People's Hospital between December 2009 and November 2011. Twenty-two machine learning (ML) models were tested, and their predictive performance compared with logistic regression (LR) model. Receiver operating characteristics (ROC), area under curve (AUC), accuracy, F-score, precision, recall and Decision Curve Analysis (DCA) were used as performance metrics.

RESULTS

A total of 117 patients were enrolled. AUC of all ML models ranged from 86.3% to 94%. AUC of LR was 83%, and accuracy was 88%. The AUC of Cubic SVM, Quadratic SVM and Linear SVM were higher than that of LR. The precision ratio of LR was 95% and recall ratio was 91%, both were lower than most ML models. The F-Score of LR was 0.93, which was only slightly better than that of Linear Discriminant and Quadratic Discriminant.

CONCLUSIONS

The twenty-two ML models selected have capabilities comparable to classical LR model for outcome prediction in STBI patients. Of these, Cubic SVM, Quadratic SVM, Linear SVM performed significantly better than LR.

Development and future deployment of a 5 years allograft survival model for kidney transplantation

AIM

Identifying kidney transplant patients at highest risk for graft loss prior to loss may allow for effective interventions to improve 5 years survival.

METHODS

We performed a 10 years retrospective cohort study of adult kidney transplant recipients (n = 1747). We acquired data from electronic health records, United Network of Organ Sharing, social determinants of health, natural language processing data extraction, and real-time capture of dynamically evolving clinical data obtained within 1 year of transplant; from which we developed a 5 years graft survival model.

RESULTS

Total of 1439 met eligibility; 265 (18.4%) of them experienced graft loss by 5 years. Graft loss patients were characterized by: older age, being African-American, diabetic, unemployed, smokers, having marginal donor kidneys and cardiovascular comorbidities. Predictive dynamic variables included: low mean blood pressure, higher pulse pressures, higher heart rate, anaemia, lower estimated glomerular filtration rate peak, increased tacrolimus variability, rejection and readmissions. This Big Data analysis generated a 5 years graft loss model with an 82% predictive capacity, versus 66% using baseline United Network of Organ Sharing data alone.

CONCLUSION

Our analysis yielded a 5 years graft loss model demonstrating superior predictive capacity compared with United Network of Organ Sharing data alone, allowing post-transplant individualized risk-assessed care prior to transitioning back to community care.

Predicting factors for survival of breast cancer patients using machine learning techniques

BACKGROUND

Breast cancer is one of the most common diseases in women worldwide. Many studies have been conducted to predict the survival indicators, however most of these analyses were predominantly performed using basic statistical methods. As an alternative, this study used machine learning techniques to build models for detecting and visualising significant prognostic indicators of breast cancer survival rate.

METHODS

A large hospital-based breast cancer dataset retrieved from the University Malaya Medical Centre, Kuala Lumpur, Malaysia (n = 8066) with diagnosis information between 1993 and 2016 was used in this study. The dataset contained 23 predictor variables and one dependent variable, which referred to the survival status of the patients (alive or dead). In determining the significant prognostic factors of breast cancer survival rate, prediction models were built using decision tree, random forest, neural networks, extreme boost, logistic regression, and support vector machine. Next, the dataset was clustered based on the receptor status of breast cancer patients identified via immunohistochemistry to perform advanced modelling using random forest. Subsequently, the important variables were ranked via variable selection methods in random forest. Finally, decision trees were built and validation was performed using survival analysis.

RESULTS

In terms of both model accuracy and calibration measure, all algorithms produced close outcomes, with the lowest obtained from decision tree (accuracy = 79.8%) and the highest from random forest (accuracy = 82.7%). The important variables identified in this study were cancer stage classification, tumour size, number of total axillary lymph nodes removed, number of positive lymph nodes, types of primary treatment, and methods of diagnosis.

CONCLUSION

Interestingly the various machine learning algorithms used in this study yielded close accuracy hence these methods could be used as alternative predictive tools in the breast cancer survival studies, particularly in the Asian region. The important prognostic factors influencing survival rate of breast cancer identified in this study, which were validated by survival curves, are useful and could be translated into decision support tools in the medical domain.

Predicting outcomes of chronic kidney disease from EMR data based on Random Forest Regression

Chronic kidney disease (CKD) is prevalent across the world, and kidney function is well defined by an estimated glomerular filtration rate (eGFR). The progression of kidney disease can be predicted if the future eGFR can be accurately estimated using predictive analytics. In this study, we developed and validated a prediction model of eGFR by data extracted from a regional health system. This dataset includes demographic, clinical and laboratory information from primary care clinics. The model was built using Random Forest regression and evaluated using Goodness-of-fit statistics and discrimination metrics. After data preprocessing, the patient cohort for model development and validation contained 61,740 patients. The final model included eGFR, age, gender, body mass index (BMI), obesity, hypertension, and diabetes, which achieved a mean coefficient of determination of 0.95. The estimated eGFRs were used to classify patients into CKD stages with high macro-averaged and micro-averaged metrics. In conclusion, a model using real-world electronic medical records (EMR) data can accurately predict future kidney functions and provide clinical decision support.

A Study of Machine-Learning Classifiers for Hypertension Based on Radial Pulse Wave

Objective

In this study, machine learning was utilized to classify and predict pulse wave of hypertensive group and healthy group and assess the risk of hypertension by observing the dynamic change of the pulse wave and provide an objective reference for clinical application of pulse diagnosis in traditional Chinese medicine (TCM).

Method

The basic information from 450 hypertensive cases and 479 healthy cases was collected by self-developed H20 questionnaires and pulse wave information was acquired by self-developed pulse diagnostic instrument (PDA-1). H20 questionnaires and pulse wave information were used as input variables to obtain different machine learning classification models of hypertension. This method was aimed at analyzing the influence of pulse wave on the accuracy and stability of machine learning model, as well as the feature contribution of hypertension model after removing noise by K-means.

Result

Compared with the classification results before removing noise, the accuracy and the area under the curve (AUC) had been improved. The accuracy rates of AdaBoost, Gradient Boosting, and Random Forest (RF) were 86.41%, 86.41%, and 85.33%, respectively. AUC were 0.86, 0.86, and 0.85, respectively. The maximum accuracy of SVM increased from 79.57% to 83.15%, and the AUC stability increased from 0.79 to 0.83. In addition, the features of importance on traditional statistics and machine learning were consistent. After removing noise, the features with large changes were h1/t1, w1/t, t, w2, h2, t1, and t5 in AdaBoost and Gradient Boosting (top10). The common variables for machine learning and traditional statistics were h1/t1, h5, t, Ad, BMI, and t2.

Conclusion

Pulse wave-based diagnostic method of hypertension has significant value in reference. In view of the feasibility of digital-pulse-wave diagnosis and dynamically evaluating hypertension, it provides the research direction and foundation for Chinese medicine in the dynamic evaluation of modern disease diagnosis and curative effect.

Predicting hospital and emergency department utilization among community-dwelling older adults: Statistical and machine learning approaches

OBJECTIVE

The objective of this study was to compare the performance of several commonly used machine learning methods to traditional statistical methods for predicting emergency department and hospital utilization among patients receiving publicly-funded home care services.

STUDY DESIGN AND SETTING

We conducted a population-based retrospective cohort study of publicly-funded home care recipients in the Hamilton-Niagara-Haldimand-Brant region of southern Ontario, Canada between 2014 and 2016. Gradient boosted trees, neural networks, and random forests were tested against two variations of logistic regression for predicting three outcomes related to emergency department and hospital utilization within six months of a comprehensive home care clinical assessment. Models were trained on data from years 2014 and 2015 and tested on data from 2016. Performance was compared using logarithmic score, Brier score, AUC, and diagnostic accuracy measures.

RESULTS

Gradient boosted trees achieved the best performance on all three outcomes. Gradient boosted trees provided small but statistically significant performance gains over both traditional methods on all three outcomes, while neural networks significantly outperformed logistic regression on two of three outcomes. However, sensitivity and specificity gains from using gradient boosted trees over logistic regression were only in the range of 1%-2% at several classification thresholds.

CONCLUSION

Gradient boosted trees and simple neural networks yielded small performance benefits over logistic regression for predicting emergency department and hospital utilization among patients receiving publicly-funded home care. However, the performance benefits were of negligible clinical importance.

Predicting Outcome of Endovascular Treatment for Acute Ischemic Stroke: Potential Value of Machine Learning Algorithms

Background: Endovascular treatment (EVT) is effective for stroke patients with a large vessel occlusion (LVO) of the anterior circulation. To further improve personalized stroke care, it is essential to accurately predict outcome after EVT. Machine learning might outperform classical prediction methods as it is capable of addressing complex interactions and non-linear relations between variables.

Methods: We included patients from the Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN) Registry, an observational cohort of LVO patients treated with EVT. We applied the following machine learning algorithms: Random Forests, Support Vector Machine, Neural Network, and Super Learner and compared their predictive value with classic logistic regression models using various variable selection methodologies. Outcome variables were good reperfusion (post-mTICI ≥ 2b) and functional independence (modified Rankin Scale ≤2) at 3 months using (1) only baseline variables and (2) baseline and treatment variables. Area under the ROC-curves (AUC) and difference of mean AUC between the models were assessed.

Results: We included 1,383 EVT patients, with good reperfusion in 531 (38%) and functional independence in 525 (38%) patients. Machine learning and logistic regression models all performed poorly in predicting good reperfusion (range mean AUC: 0.53–0.57), and moderately in predicting 3-months functional independence (range mean AUC: 0.77–0.79) using only baseline variables. All models performed well in predicting 3-months functional independence using both baseline and treatment variables (range mean AUC: 0.88–0.91) with a negligible difference of mean AUC (0.01; 95%CI: 0.00–0.01) between best performing machine learning algorithm (Random Forests) and best performing logistic regression model (based on prior knowledge).

Conclusion: In patients with LVO machine learning algorithms did not outperform logistic regression models in predicting reperfusion and 3-months functional independence after endovascular treatment. For all models at time of admission radiological outcome was more difficult to predict than clinical outcome.

Development of a prediction model for long-term quality of life in critically ill patients

PURPOSE

We developed a prediction model for quality of life (QOL) 1 year after intensive care unit (ICU) discharge based upon data available at the first ICU day to improve decision-making.

METHODS

The database of a 1-year prospective study concerning long-term outcome and QOL (assessed by EuroQol-5D) in critically ill adult patients consecutively admitted to the ICU of a university hospital was used. Cases with missing data were excluded. Utility indices at baseline (UIb) and at 1 year (UI1y) were surrogates for QOL. For 1-year non-survivors UI1y was set at zero. The grouped lasso technique selected the most important variables in the prediction model. R² and adjusted R² were calculated.

RESULTS

1831 of 1953 cases (93.8%) were complete. UI1y depended significantly on: UIb (P<0.001); solid tumor (P<0.001); age (P<0.001); activity of daily living (P<0.001); imaging (P<0.001); APACHE II-score (P=0.001); ≥80 years (P=0.001); mechanical ventilation (P=0.006); hematological patient (P=0.007); SOFA-score (P=0.008); tracheotomy (P=0.018); admission diagnosis surgical P<0.001 (versus medical); and comorbidity (P=0.049). Only baseline health status and surgical patients were positively associated with UI1y. R² was 0.3875 and adjusted R² 0.3807.

CONCLUSION

Although only 40% of variability in long-term QOL could be explained, this prediction model can be helpful in decision-making.

A Machine Learning Approach Using Survival Statistics to Predict Graft Survival in Kidney Transplant Recipients: A Multicenter Cohort Study

Accurate prediction of graft survival after kidney transplant is limited by the complexity and heterogeneity of risk factors influencing allograft survival. In this study, we applied machine learning methods, in combination with survival statistics, to build new prediction models of graft survival that included immunological factors, as well as known recipient and donor variables. Graft survival was estimated from a retrospective analysis of the data from a multicenter cohort of 3,117 kidney transplant recipients. We evaluated the predictive power of ensemble learning algorithms (survival decision tree, bagging, random forest, and ridge and lasso) and compared outcomes to those of conventional models (decision tree and Cox regression). Using a conventional decision tree model, the 3-month serum creatinine level post-transplant (cut-off, 1.65 mg/dl) predicted a graft failure rate of 77.8% (index of concordance, 0.71). Using a survival decision tree model increased the index of concordance to 0.80, with the episode of acute rejection during the first year post-transplant being associated with a 4.27-fold increase in the risk of graft failure. Our study revealed that early acute rejection in the first year is associated with a substantially increased risk of graft failure. Machine learning methods may provide versatile and feasible tools for forecasting graft survival.

In Silico Prediction of Gamma-Aminobutyric Acid Type-A Receptors Using Novel Machine-Learning-Based SVM and GBDT Approaches

Gamma-aminobutyric acid type-A receptors (GABA_ARs) belong to multisubunit membrane spanning ligand-gated ion channels (LGICs) which act as the principal mediators of rapid inhibitory synaptic transmission in the human brain. Therefore, the category prediction of GABA_ARs just from the protein amino acid sequence would be very helpful for the recognition and research of novel receptors. Based on the proteins' physicochemical properties, amino acids composition and position, a GABA_AR classifier was first constructed using a 188-dimensional (188D) algorithm at 90% cd-hit identity and compared with pseudo-amino acid composition (PseAAC) and ProtrWeb web-based algorithms for human GABA_AR proteins. Then, four classifiers including gradient boosting decision tree (GBDT), random forest (RF), a library for support vector machine (libSVM), and k-nearest neighbor (k-NN) were compared on the dataset at cd-hit 40% low identity. This work obtained the highest correctly classified rate at 96.8% and the highest specificity at 99.29%. But the values of sensitivity, accuracy, and Matthew's correlation coefficient were a little lower than those of PseAAC and ProtrWeb; GBDT and libSVM can make a little better performance than RF and k-NN at the second dataset. In conclusion, a GABA_AR classifier was successfully constructed using only the protein sequence information.