Interpretable machine learning algorithms for calculating the heat conduction coefficient of crop-derived renewable fuels

Developing Machine Learning (ML) algorithms for heterogeneous/mixed data is a longstanding problem. Many ML algorithms are not applicable to mixed data, which include numeric and non-numeric data, text, graphs and so on to generate interpretable models. Another longstanding problem is developing algorithms for lossless visualization of multidimensional mixed data. The further progress in ML heavily depends on success interpretable ML algorithms for mixed data and lossless interpretable visualization of multidimensional data. The later allows developing interpretable ML models using visual knowledge discovery by end-users, who can bring valuable domain knowledge which is absent in the training data. The challenges for mixed data include: (1) generating numeric coding schemes for non-numeric attributes for numeric ML algorithms to provide accurate and interpretable ML models, (2) generating methods for lossless visualization of n-D non-numeric data and visual rule discovery in these visualizations. This paper presents a classification of mixed data types, analyzes their importance for ML and present the developed experimental toolkit to deal with mixed data. It combines the Data Types Editor, VisCanvas data visualization and rule discovery system which is available on GitHub.

Introduction: Alzheimer’s disease is one of the most common neurodegenerative diseases in adults. The progressive nature of Alzheimer’s disease causes widespread damage to the brain, and early diagnosis can manage the disease and slow down its progression effectively. Method: In this study, a dataset related to the early prediction of Alzheimer’s was used. Spark framework was used for data management and three machine learning algorithms including Naïve Bayes, Decision Tree, and Artificial Neural Networks with the best hyperparameters were implemented and compared. To prevent overfitting and measure the efficiency of the models, a 5 -fold cross -validation method was utilized. Furthermore, a method was used for interpreting machine learning black box models. Results: The decision tree and artificial neural network models obtained 98.61% accuracy and 98.60% F1 -Score in the Spark framework including one or three computers. Important features in the decision -making process of the artificial neural network were identified using the interpretability technique. In addition, the computational time required for training the proposed models was calculated through different approaches, and the use of multiple computers was 35.95% faster than a single computer. Conclusion: With increasing the number of Alzheimer’s disease patients around the world, the need for a decision support system using machine learning algorithms, which can predict the disease early in a huge amount of data, is felt more. Therefore, the machine learning models proposed in this study for early prediction of Alzheimer’s disease as an interpretable auxiliary tool in the decision -making process can help clinicians .

BACKGROUND Acute kidney injury (AKI) is a common adverse outcome following nephrectomy. The progression from AKI to acute kidney disease (AKD) and subsequently to chronic kidney disease (CKD) remains a concern; yet, the predictive mechanisms for these transitions are not fully understood. Interpretable machine learning (ML) models offer insights into how clinical features influence long-term renal function outcomes after nephrectomy, providing a more precise framework for identifying patients at risk and supporting improved clinical decision-making processes. OBJECTIVE This study aimed to (1) evaluate postnephrectomy rates of AKI, AKD, and CKD, analyzing long-term renal outcomes along different trajectories; (2) interpret AKD and CKD models using Shapley Additive Explanations values and Local Interpretable Model-Agnostic Explanations algorithm; and (3) develop a web-based tool for estimating AKD or CKD risk after nephrectomy. METHODS We conducted a retrospective cohort study involving patients who underwent nephrectomy between July 2012 and June 2019. Patient data were randomly split into training, validation, and test sets, maintaining a ratio of 76.5:8.5:15. Eight ML algorithms were used to construct predictive models for postoperative AKD and CKD. The performance of the best-performing models was assessed using various metrics. We used various Shapley Additive Explanations plots and Local Interpretable Model-Agnostic Explanations bar plots to interpret the model and generated directed acyclic graphs to explore the potential causal relationships between features. Additionally, we developed a web-based prediction tool using the top 10 features for AKD prediction and the top 5 features for CKD prediction. RESULTS The study cohort comprised 1559 patients. Incidence rates for AKI, AKD, and CKD were 21.7% (n=330), 15.3% (n=238), and 10.6% (n=165), respectively. Among the evaluated ML models, the Light Gradient-Boosting Machine (LightGBM) model demonstrated superior performance, with an area under the receiver operating characteristic curve of 0.97 for AKD prediction and 0.96 for CKD prediction. Performance metrics and plots highlighted the model’s competence in discrimination, calibration, and clinical applicability. Operative duration, hemoglobin, blood loss, urine protein, and hematocrit were identified as the top 5 features associated with predicted AKD. Baseline estimated glomerular filtration rate, pathology, trajectories of renal function, age, and total bilirubin were the top 5 features associated with predicted CKD. Additionally, we developed a web application using the LightGBM model to estimate AKD and CKD risks. CONCLUSIONS An interpretable ML model effectively elucidated its decision-making process in identifying patients at risk of AKD and CKD following nephrectomy by enumerating critical features. The web-based calculator, found on the LightGBM model, can assist in formulating more personalized and evidence-based clinical strategies.

Internet traffic classification is a fundamental task for network services and management. There are good machine learning models to identify the class of traffic. However, finding the most discriminating features to have efficient models remains essential. In this paper, we use interpretable machine learning algorithms such as decision tree, random forest and eXtreme gradient boosting (XGBoost) to find the most discriminating features for internet traffic classification. The dataset used contains 377,526 traffics. Each traffic is described by 248 features. From these features, we propose a 12-feature model with an accuracy of up to 99.76%. We tested it on another dataset with 19626 flows and obtained 98.40% of accuracy. This shows the efficiency and stability of our model. Also, we identify a set of 14 important features for internet traffic classification, including two that are crucial: port number (server) and minimum segment size (client to server).

Amyloid-β(Aβ) PET positivity in patients with suspected cerebral amyloid angiopathy (CAA) MRI markers is predictive of a worse cognitive trajectory, and it provides insights into the underlying vascular pathology (CAA vs. hypertensive angiopathy) to facilitate prognostic prediction and appropriate treatment decisions. In this study, we applied two interpretable machine learning algorithms, gradient boosting machine (GBM) and random forest (RF), to predict Aβ PET positivity in patients with CAA MRI markers. In the GBM algorithm, the number of lobar cerebral microbleeds (CMBs), deep CMBs, lacunes, CMBs in dentate nuclei, and age were ranked as the most influential to predict Aβ positivity. In the RF algorithm, the absence of diabetes was additionally chosen. Cut-off values of the above variables predictive of Aβ positivity were as follows: (1) the number of lobar CMBs > 16.4(GBM)/14.3(RF), (2) no deep CMBs(GBM/RF), (3) the number of lacunes > 7.4(GBM/RF), (4) age > 74.3(GBM)/64(RF), (5) no CMBs in dentate nucleus(GBM/RF). The classification performances based on the area under the receiver operating characteristic curve were 0.83 in GBM and 0.80 in RF. Our study demonstrates the utility of interpretable machine learning in the clinical setting by quantifying the relative importance and cutoff values of predictive variables for Aβ positivity in patients with suspected CAA markers.

Cancer is already the leading cause of death in most Brazilian cities and in the world. The understanding of its internal mechanisms and the design of computational models capable of improving its diagnosis will have strong benefits for humanity. New technologies have made available a wealth of data, which can be used to improve the diagnosis of cancer. As a manual analysis of this data is impracticable, many black-box machine learning algorithms have been employed successfully for cancer diagnosis. Despite their high accuracy prediction abilities, black-box models sacrifice transparency and accountability. In contrast, interpretable machine learning algorithms are powerful tools for understanding the underlying mechanism present within a large corpus of data. In this work, Linear Projections and Radviz were used as visualization techniques for data exploration and feature selection. Further, Decision Tree induction algorithms were used to create models that are able to differentiate between Malignant and Benign breast tumors from breast mass images. These models can be considered white-box models which means their inner workings are easier to explain and interpret. The result shows Classification and Regression Trees achieved an accuracy of 96% in predicting breast cancer.

ObjectiveAcute ischemic stroke (AIS) brings an increasingly heavier economic burden nowadays. Prolonged length of stay (LOS) is a vital factor in healthcare expenditures. The aim of this study was to predict prolonged LOS in AIS patients based on an interpretable machine learning algorithm.MethodsWe enrolled AIS patients in our hospital from August 2017 to July 2019, and divided them into the “prolonged LOS” group and the “no prolonged LOS” group. Prolonged LOS was defined as hospitalization for more than 7 days. The least absolute shrinkage and selection operator (LASSO) regression was applied to reduce the dimensionality of the data. We compared the predictive capacity of extended LOS in eight different machine learning algorithms. SHapley Additive exPlanations (SHAP) values were used to interpret the outcome, and the most optimal model was assessed by discrimination, calibration, and clinical utility.ResultsProlonged LOS developed in 149 (22.0%) of the 677 eligible patients. In eight machine learning algorithms, prolonged LOS was best predicted by the Gaussian naive Bayes (GNB) model, which had a striking area under the curve (AUC) of 0.878 ± 0.007 in the training set and 0.857 ± 0.039 in the validation set. The variables sorted by the gap values showed that the strongest predictors were pneumonia, dysphagia, thrombectomy, and stroke severity. High net benefits were observed at 0%–76% threshold probabilities, while good agreement was found between the observed and predicted probabilities.ConclusionsThe model using the GNB algorithm proved excellent for predicting prolonged LOS in AIS patients. This simple model of prolonged hospitalization could help adjust policies and better utilize resources.

Septic shock is a severe complication of acute pancreatitis (AP), often associated with poor prognosis. This study aims to analyze the clinical characteristics of patients with acute pancreatitis and develop an interpretable early prediction model for septic shock in these patients using machine learning (ML). The model is intended to assist emergency physicians in resource allocation and medical decision making. Data were collected from the MIMIC-IV 3.0 database. The dataset was divided into a training set and a test set in a 7:3 ratio. Feature selection was performed using LASSO (Least Absolute Shrinkage and Selection Operator) regression. Subsequently, 10 ML models were developed: Random Forest, Logistic Regression, Gradient Boosting Machine, Neural Network, Extreme Gradient Boosting (XGBoost), K-Nearest Neighbor, Adaptive Boosting, Light Gradient Boosting Machine, Category Boosting, and Support Vector Machine. To enhance and optimize model interpretability, Shapley Additive Explanations (SHAP) were employed. A total of 1032 patients with AP were included in this study, from which 31 variables were selected for model development. By comparing the area under the receiver operating characteristic curve and decision curve analysis results between the training and test sets, the XGBoost model demonstrated a significant advantage over other models. SHAP analysis revealed that white blood cell count, total bilirubin (bilirubin total), and bicarbonate (HCO3 -) levels were the three most critical risk factors for the development of septic shock in patients with AP. ML approaches exhibited promising performance in predicting septic shock in patients with AP. These models may aid in guiding treatment decisions for patients with AP in the emergency department.

The maximum diversified demand is an important factor to consider when utilities design new distribution systems. To estimate the maximum diversified demand, engineers need to make an estimate of the diversity factor (DF). In practice, electricity utility companies usually estimate the DF using DF tables, in which the DF changes with the number of customers. However, besides the number of customers, DF also depends on many other factors, such as customer type, weather, demographics, and socioeconomic conditions. Ignoring these factors, DF tables have limited accuracy. In addition, engineers cannot interpret or understand how various factors affect the DF. In this paper, by leveraging supervised machine learning algorithms, we build comprehensive DF prediction models that take a variety of factors into account. These models show high prediction accuracy and interpretabilty when applied to real-world distribution feeders. Using the interpretation method called SHapley Additive exPlanations, we quantity the importance of different features in determining DFs. Finally, we offer more insights into how various factors affect DFs.

A quantitative investigation on pyrolysis behaviors of metal ion-exchanged coal macerals by interpretable machine learning algorithms

Time-dependent interpretable survival prediction model for second primary NSCLC patients.

Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Mosti TED1 grant Background No study has used interpretative Machine Learning (ML) algorithms to predict in-hospital mortality for the Asian elderly (65+). TIMI predicts mortality for STEMI and NSTEMI using two different scores. It was based on the Western cohort with limited Asian data. Purpose Develop a single mortality risk scoring system for STEMI and NSTEMI patients and use interpretative ML to identify and analyse risk factors in elderly Asian patients with ACS. Methods The National Cardiovascular Disease Database registry identified 4305 elderly. 70% of the data was used to develop algorithms and 30% for validation. Fifty-four parameters were considered, including demographics, cardiovascular risk, medications, and clinical variables. A sequential backward elimination (SBE) algorithm was used to identify variables associated to elderly mortality. XGBoost classification algorithm and SHapley Additive exPlanation (SHAP) were used to understand mortality impact. The SHAP value of each variable represents its impact on model output (mortality). The main performance metric was area under the curve (AUC). The model was validated using a validation dataset and compared to STEMI and NSTEMI for TIMI. Results XGBoost's validation dataset performance using the top 12 predictors from SBE for; STEMI (AUC = 0.822, 95% CI: 0.775-0.868, Accuracy: 0.875, Sensitivity: 0.164, Specificity: 0.966) and NSTEMI (AUC = 0.853, 95% CI: 0.802-0.904, Accuracy: 0.950, Sensitivity: 0.154, Specificity: 0.997). XGBoost's validation dataset performance using the eight emergency predictors selected from the top twelve predictors for; STEMI (AUC = 0.813, 95% CI: 0.766-0.861, Accuracy: 0.868, Sensitivity: 0.194, Specificity 0.954) and NSTEMI (AUC = 0.867, 95% CI: 0.812-0.921, Accuracy: 0.941, Sensitivity: 0.333, Specificity: 0.978). Both models outperformed TIMI score (STEMI AUC = 0.702, NSTEMI AUC = 0.524). The predictors were chosen and ranked in ascending order using the SHAP values (Figure 1). On the y-axis, the variable names are displayed in ascending order of importance and the colour represents the feature's value, ranging from low to high, allowing comprehension of the distribution of SHAP values for each feature. The x-axis displays the SHAP values. Eight out of the twelve predictors were identified to be emergency variables and was ranked according to SHAP values (Figure 2). When compared to TIMI, cardiac catheterization, percutaneous coronary intervention, and pharmacotherapy drugs are chosen as predictors that improve mortality prediction in STEMI and NSTEMI elderly patients. High killip class and age are linked to a poorer ACS patient survival rate, but cardiac catheterization and use of pharmacotherapy drugs improve patient mortality. Conclusions A single algorithm can better classify elderly ASIAN patients with ACS than TIMI, which requires two scores. The use of interpretative algorithms aids in the understanding of ACS elderly hospital mortality factors.

The American College of Surgeons (ACS) NSQIP risk calculator helps guide operative decision making. In patients with significant surgical risk, it may be unclear whether to proceed with "Hail Mary"-type interventions. To refine predictions, a local interpretable model-agnostic explanations machine (LIME) learning algorithm was explored to determine weighted patient-specific factors' contribution to mortality. The ACS-NSQIP database was queried for all surgical patients with mortality probability greater than 50% between 2012 and 2019. Preoperative factors (n = 38) were evaluated using stepwise logistic regression; 26 significant factors were used in gradient boosted machine (GBM) modeling. Data were divided into training and testing sets, and model performance was substantiated with 10-fold cross validation. LIME provided individual subject mortality. The GBM-trained model was interpolated to LIME, and predictions were made using the test dataset. There were 6,483 deaths (53%) among 12,248 admissions. GBM modeling displayed good performance (area under the curve = 0.65, 95% CI 0.636-0.671). The top 5 factors (% contribution) to mortality included: septic shock (27%), elevated International Normalized Ratio (22%), ventilator-dependence (14%), thrombocytopenia (14%), and elevated serum creatinine (5%). LIME modeling subset personalized patients by factors and weights on survival. In the entire cohort, mortality positive predictive value with 2 factor combinations was 53.5% (specificity 0.713), 3 combinations 64.2% (specificity 0.835), 4 combinations 72.1% (specificity 0.943), and all 5 combinations 77.9% (specificity 0.993). Conversely, mortality positive predictive value fell to 34% in the absence of 4 factors. Through the application of machine learning algorithms (GBM and LIME), our model individualized predicted mortality and contributing factors with substantial ACS-NSQIP predicted mortality. USE of machine learning techniques may better inform operative decisions and family conversations in cases of significant surgical risk.

BackgroundOvarian clear cell carcinoma (OCCC) represents a subtype of ovarian epithelial carcinoma (OEC) known for its limited responsiveness to chemotherapy, and the onset of distant metastasis significantly impacts patient prognoses. This study aimed to identify potential risk factors contributing to the occurrence of distant metastasis in OCCC.MethodsUtilizing the Surveillance, Epidemiology, and End Results (SEER) database, we identified patients diagnosed with OCCC between 2004 and 2015. The most influential factors were selected through the application of Gaussian Naive Bayes (GNB) and Adaboost machine learning algorithms, employing a Venn test for further refinement. Subsequently, six machine learning (ML) techniques, namely XGBoost, LightGBM, Random Forest (RF), Adaptive Boosting (Adaboost), Support Vector Machine (SVM), and Multilayer Perceptron (MLP), were employed to construct predictive models for distant metastasis. Shapley Additive Interpretation (SHAP) analysis facilitated a visual interpretation for individual patient. Model validity was assessed using accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and the area under the receiver operating characteristic curve (AUC).ResultsIn the realm of predicting distant metastasis, the Random Forest (RF) model outperformed the other five machine learning algorithms. The RF model demonstrated accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and AUC (95% CI) values of 0.792 (0.762–0.823), 0.904 (0.835–0.973), 0.759 (0.731–0.787), 0.221 (0.186–0.256), 0.974 (0.967–0.982), 0.353 (0.306–0.399), and 0.834 (0.696–0.967), respectively, surpassing the performance of other models. Additionally, the calibration curve's Brier Score (95%) for the RF model reached the minimum value of 0.06256 (0.05753–0.06759). SHAP analysis provided independent explanations, reaffirming the critical clinical factors associated with the risk of metastasis in OCCC patients.ConclusionsThis study successfully established a precise predictive model for OCCC patient metastasis using machine learning techniques, offering valuable support to clinicians in making informed clinical decisions.

Chemical toxicity evaluations for drugs, consumer products, and environmental chemicals have a critical impact on human health. Traditional animal models to evaluate chemical toxicity are expensive, time-consuming, and often fail to detect toxicants in humans. Computational toxicology is a promising alternative approach that utilizes machine learning (ML) and deep learning (DL) techniques to predict the toxicity potentials of chemicals. Although the applications of ML- and DL-based computational models in chemical toxicity predictions are attractive, many toxicity models are "black boxes" in nature and difficult to interpret by toxicologists, which hampers the chemical risk assessments using these models. The recent progress of interpretable ML (IML) in the computer science field meets this urgent need to unveil the underlying toxicity mechanisms and elucidate the domain knowledge of toxicity models. In this review, we focused on the applications of IML in computational toxicology, including toxicity feature data, model interpretation methods, use of knowledge base frameworks in IML development, and recent applications. The challenges and future directions of IML modeling in toxicology are also discussed. We hope this review can encourage efforts in developing interpretable models with new IML algorithms that can assist new chemical assessments by illustrating toxicity mechanisms in humans.