Explainable Machine Learning Models Research Articles

Deciphering the effects of 2D/3D urban morphology on diurnal cooling efficiency of urban green space

Urban green spaces (UGS) have emerged as a main nature-based solution proven to effectively cool urban areas, and cooling efficiency (CE) serves as a prominent indicator for assessing the performance of UGS. Yet few studies have explored the diurnal dynamics of CE and the comprehensive effects of both 2D and 3D urban morphology on CE over a 24-h period. In this study, we employed diurnal land surface temperature (LST) data from Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) to delve into the diurnal CE of UGS within the urban core area of megacity Paris, France. Meanwhile, using high-resolution land cover data, digital height model, and explainable machine learning models, we elucidated the intricate relationship between various 2D and 3D urban morphology metrics and CE during a 24-h cycle. In summary, city-scale continuous monitoring revealed diurnal CE variations, with values ranging from 0.008 °C to 0.13 °C and a standard deviation of 0.04 °C. On an average of a day, the relative importance of the 2D and 3D urban form indicators was 77.25 % and 22.75 %, respectively. Enhancing the dispersion of trees or improving the integrity of non-tree vegetation patches proves to be an effective strategy for augmenting CE. The results can be further used in urban planning to improve urban thermal resilience and adaptability.

Associations of personal PM2.5-bound heavy metals and heavy metal mixture with lung function: Results from a panel study in Chinese urban residents

BackgroundThere are a few reports on the associations between fine particulate matter (PM2.5)-bound heavy metals and lung function. ObjectivesTo evaluate the associations of single and mixed PM2.5-bound heavy metals with lung function. MethodsThis study included 316 observations of 224 Chinese adults from the Wuhan-Zhuhai cohort over two study periods, and measured participants' personal PM2.5-bound heavy metals and lung function. Three linear mixed models, including the single constituent model, the PM2.5-adjusted constituent model, and the constituent residual model were used to evaluate the association between single metal and lung function. Mixed exposure models including Bayesian kernel machine regression (BKMR) model, weighted quantile sum (WQS) model, and Explainable Machine Learning model were used to assess the relationship between PM2.5-bound heavy metal mixtures and lung function. ResultsIn the single exposure analyses, significant negative associations of PM2.5-bound lead, antimony, and cadmium with peak expiratory flow (PEF) were observed. In the mixed exposure analyses, significant decreases in forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC), maximal mid-expiratory flow (MMF), and forced expiratory flow at 75% of the pulmonary volume (FEF75) were associated with the increased PM2.5-bound heavy metal mixture. The BKMR models suggested negative associations of PM2.5-bound lead and antimony with lung function. In addition, PM2.5-bound copper was positively associated with FEV1/FVC, MMF, and FEF75. The Explainable Machine Learning models suggested that FEV1/FVC, MMF, and FEF75 decreased with the elevated PM2.5-bound lead, manganese, and vanadium, and increased with the elevated PM2.5-bound copper. ConclusionsThe negative relationships were detected between PM2.5-bound heavy metal mixture and FEV1/FVC, MMF, as well as FEF75. Among the PM2.5-bound heavy metal mixture, PM2.5-bound lead, antimony, manganese, and vanadium were negatively associated with FEV1/FVC, MMF, and FEF75, while PM2.5-bound copper was positively associated with FEV1/FVC, MMF, and FEF75.

Understanding predictions of drug profiles using explainable machine learning models

PurposeThe analysis of absorption, distribution, metabolism, and excretion (ADME) molecular properties is of relevance to drug design, as they directly influence the drug’s effectiveness at its target location. This study concerns their prediction, using explainable Machine Learning (ML) models. The aim of the study is to find which molecular features are relevant to the prediction of the different ADME properties and measure their impact on the predictive model.MethodsThe relative relevance of individual features for ADME activity is gauged by estimating feature importance in ML models’ predictions. Feature importance is calculated using feature permutation and the individual impact of features is measured by SHAP additive explanations.ResultsThe study reveals the relevance of specific molecular descriptors for each ADME property and quantifies their impact on the ADME property prediction.ConclusionThe reported research illustrates how explainable ML models can provide detailed insights about the individual contributions of molecular features to the final prediction of an ADME property, as an effort to support experts in the process of drug candidate selection through a better understanding of the impact of molecular features.

Explainable machine learning models for early gastric cancer diagnosis

Gastric cancer remains a significant global health concern, with a notably high incidence in East Asia. This paper explores the potential of explainable machine learning models in enhancing the early diagnosis of gastric cancer. Through comprehensive evaluations, various machine learning models, including WeightedEnsemble, CatBoost, and RandomForest, demonstrated high potential in accurately diagnosing early gastric cancer. The study emphasizes the importance of model explainability in medical diagnostics, showing how transparent, explainable models can increase trust and clinical acceptance, thereby improving diagnostic accuracy and patient outcomes. This research not only highlights key biomarkers and clinical features critical for early detection but also presents a versatile approach that could be applied to other medical diagnostics, promoting broader adoption of machine learning in clinical settings.

Chemistry and Local Environment Adaptive Representation graphs as material descriptors

Accelerating materials design and property prediction via supervised learning involves manually generating feature vectors or performing complicated atom coordinates transformation, which restricts the model to a limited set of crystal structures or makes it challenging to provide chemical insights. In this work, we devised a unique low-dimensional featurization technique known as "Chemistry and Local Environment Adaptive Representation" (CLEAR) graphs to automatically learn the properties of materials through the chemistry of the atom connections. CLEAR is an adaptive featurization method that integrates the chemistry of atoms with their atomic environment using Voronoi nearest neighbours (NN). In addition, integrating the CLEAR descriptors with explainable machine learning models unravels its potential to obtain numerous scientific insights. We applied this approach to study the stability of compositional and configurational diverse high entropy alloys (HEAs). The proposed framework provides a universal low-dimensional representation with chemistry-informed local environment descriptors, which outperforms the prediction of phases and formation energies in HEAs.

Explainable AI Method for Tinnitus Diagnosis via Neighbor-Augmented Knowledge Graph and Traditional Chinese Medicine: Development and Validation Study.

Tinnitus diagnosis poses a challenge in otolaryngology owing to an extremely complex pathogenesis, lack of effective objectification methods, and factor-affected diagnosis. There is currently a lack of explainable auxiliary diagnostic tools for tinnitus in clinical practice. This study aims to develop a diagnostic model using an explainable artificial intelligence (AI) method to address the issue of low accuracy in tinnitus diagnosis. In this study, a knowledge graph-based tinnitus diagnostic method was developed by combining clinical medical knowledge with electronic medical records. Electronic medical record data from 1267 patients were integrated with traditional Chinese clinical medical knowledge to construct a tinnitus knowledge graph. Subsequently, weights were introduced, which measured patient similarity in the knowledge graph based on mutual information values. Finally, a collaborative neighbor algorithm was proposed, which scored patient similarity to obtain the recommended diagnosis. We conducted 2 group experiments and 1 case derivation to explore the effectiveness of our models and compared the models with state-of-the-art graph algorithms and other explainable machine learning models. The experimental results indicate that the method achieved 99.4% accuracy, 98.5% sensitivity, 99.6% specificity, 98.7% precision, 98.6% F1-score, and 99% area under the receiver operating characteristic curve for the inference of 5 tinnitus subtypes among 253 test patients. Additionally, it demonstrated good interpretability. The topological structure of knowledge graphs provides transparency that can explain the reasons for the similarity between patients. This method provides doctors with a reliable and explainable diagnostic tool that is expected to improve tinnitus diagnosis accuracy.

Explainable machine learning predicts survival of retroperitoneal liposarcoma: A study based on the SEER database and external validation in China

AbstractObjectiveWe have developed explainable machine learning models to predict the overall survival (OS) of retroperitoneal liposarcoma (RLPS) patients. This approach aims to enhance the explainability and transparency of our modeling results.MethodsWe collected clinicopathological information of RLPS patients from The Surveillance, Epidemiology, and End Results (SEER) database and allocated them into training and validation sets with a 7:3 ratio. Simultaneously, we obtained an external validation cohort from The First Affiliated Hospital of Naval Medical University (Shanghai, China). We performed LASSO regression and multivariate Cox proportional hazards analysis to identify relevant risk factors, which were then combined to develop six machine learning (ML) models: Cox proportional hazards model (Coxph), random survival forest (RSF), ranger, gradient boosting with component‐wise linear models (GBM), decision trees, and boosting trees. The predictive performance of these ML models was evaluated using the concordance index (C‐index), the integrated cumulative/dynamic area under the curve (AUC), and the integrated Brier score, as well as the Cox–Snell residual plot. We also used time‐dependent variable importance, analysis of partial dependence survival plots, and the generation of aggregated survival SHapley Additive exPlanations (SurvSHAP) plots to provide a global explanation of the optimal model. Additionally, SurvSHAP (t) and survival local interpretable model‐agnostic explanations (SurvLIME) plots were used to provide a local explanation of the optimal model.ResultsThe final ML models are consisted of six factors: patient's age, gender, marital status, surgical history, as well as tumor's histopathological classification, histological grade, and SEER stage. Our prognostic model exhibits significant discriminative ability, particularly with the ranger model performing optimally. In the training set, validation set, and external validation set, the AUC for 1, 3, and 5 year OS are all above 0.83, and the integrated Brier scores are consistently below 0.15. The explainability analysis of the ranger model also indicates that histological grade, histopathological classification, and age are the most influential factors in predicting OS.ConclusionsThe ranger ML prognostic model exhibits optimal performance and can be utilized to predict the OS of RLPS patients, offering valuable and crucial references for clinical physicians to make informed decisions in advance.

Evaluating Explainable Machine Learning Models for Clinicians

Gaining clinicians’ trust will unleash the full potential of artificial intelligence (AI) in medicine, and explaining AI decisions is seen as the way to build trustworthy systems. However, explainable artificial intelligence (XAI) methods in medicine often lack a proper evaluation. In this paper, we present our evaluation methodology for XAI methods using forward simulatability. We define the Forward Simulatability Score (FSS) and analyze its limitations in the context of clinical predictors. Then, we applied FSS to our XAI approach defined over an ML-RO, a machine learning clinical predictor based on random optimization over a multiple kernel support vector machine (SVM) algorithm. To Compare FSS values before and after the explanation phase, we test our evaluation methodology for XAI methods on three clinical datasets, namely breast cancer, VTE, and migraine. The ML-RO system is a good model on which to test our XAI evaluation strategy based on the FSS. Indeed, ML-RO outperforms two other base models—a decision tree (DT) and a plain SVM—in the three datasets and gives the possibility of defining different XAI models: TOPK, MIGF, and F4G. The FSS evaluation score suggests that the explanation method F4G for the ML-RO is the most effective in two datasets out of the three tested, and it shows the limits of the learned model for one dataset. Our study aims to introduce a standard practice for evaluating XAI methods in medicine. By establishing a rigorous evaluation framework, we seek to provide healthcare professionals with reliable tools for assessing the performance of XAI methods to enhance the adoption of AI systems in clinical practice.

Use of explainable machine learning models in blast load prediction

The effects of blast waves and their consequent damage to structures have been an increasingly popular research topic in the past decade. Various methods are used in blast load prediction on structures, including experimental, semi-empirical and numerical approaches. However, there is a demand for developing time-efficient predictive methods that help various professionals, such as engineering practitioners and first responders, in their regular activities. Machine learning (ML), as a subset of artificial intelligence (AI), has gained significant attention in various engineering applications over the past decades for its ability to aid decision-making. However, a critical gap exists in the comprehensive analysis of these ML models' transparency and interpretability levels. This paper presents a novel approach to employing explainable machine learning (XML) to predict the blast loads generated by high explosives. The primary objective is to assess the validity of utilising a machine learning (ML) framework for blast load predictions, examining the scientific consistency of the model's prediction process. This research utilises XML to interpret the ML model-based blast load prediction framework, marking the first documented use of XML in published literature for blast load prediction. This research study utilised three ML models, namely a) Decision Tree, b) Random Forest, and c) Extreme Gradient Boost (XGB) models, to predict blast loads on rigid structures. Validated numerical model results were used to generate the dataset to train the ML models, and the respective dataset consists of approximately 600 independent blast events. The models were evaluated for their predictability using error metrics: adjusted coefficient of determination (adjusted R2), root mean square error (RMSE), and mean absolute error (MAE). The overall ML model performance evaluation showed that the best-performing model, XGB, could make predictions with 98 % accuracy compared to validated numerical predictions. The XML explanations for the XGB models showcased that the blast load prediction procedure of the models aligns with the blast physics principles, ensuring the credibility of the developed ML models. In summary, these XML predictive models can be considered accurate and reliable for blast load prediction on structures.

Machine learning models on a web application to predict short-term postoperative outcomes following anterior cervical discectomy and fusion

BackgroundThe frequency of anterior cervical discectomy and fusion (ACDF) has increased up to 400% since 2011, underscoring the need to preoperatively anticipate adverse postoperative outcomes given the procedure’s expanding use. Our study aims to accomplish two goals: firstly, to develop a suite of explainable machine learning (ML) models capable of predicting adverse postoperative outcomes following ACDF surgery, and secondly, to embed these models in a user-friendly web application, demonstrating their potential utility.MethodsWe utilized data from the National Surgical Quality Improvement Program database to identify patients who underwent ACDF surgery. The outcomes of interest were four short-term postoperative adverse outcomes: prolonged length of stay (LOS), non-home discharges, 30-day readmissions, and major complications. We utilized five ML algorithms - TabPFN, TabNET, XGBoost, LightGBM, and Random Forest - coupled with the Optuna optimization library for hyperparameter tuning. To bolster the interpretability of our models, we employed SHapley Additive exPlanations (SHAP) for evaluating predictor variables’ relative importance and used partial dependence plots to illustrate the impact of individual variables on the predictions generated by our top-performing models. We visualized model performance using receiver operating characteristic (ROC) curves and precision-recall curves (PRC). Quantitative metrics calculated were the area under the ROC curve (AUROC), balanced accuracy, weighted area under the PRC (AUPRC), weighted precision, and weighted recall. Models with the highest AUROC values were selected for inclusion in a web application.ResultsThe analysis included 57,760 patients for prolonged LOS [11.1% with prolonged LOS], 57,780 for non-home discharges [3.3% non-home discharges], 57,790 for 30-day readmissions [2.9% readmitted], and 57,800 for major complications [1.4% with major complications]. The top-performing models, which were the ones built with the Random Forest algorithm, yielded mean AUROCs of 0.776, 0.846, 0.775, and 0.747 for predicting prolonged LOS, non-home discharges, readmissions, and complications, respectively.ConclusionsOur study employs advanced ML methodologies to enhance the prediction of adverse postoperative outcomes following ACDF. We designed an accessible web application to integrate these models into clinical practice. Our findings affirm that ML tools serve as vital supplements in risk stratification, facilitating the prediction of diverse outcomes and enhancing patient counseling for ACDF.

A large-scale machine learning analysis of inorganic nanoparticles in preclinical cancer research.

Owing to their distinct physical and chemical properties, inorganic nanoparticles (NPs) have shown promising results in preclinical cancer therapy, but designing and engineering them for effective therapeutic purposes remains a challenge. Although a comprehensive database of inorganic NP research is not currently available, it is crucial for developing effective cancer therapies. In this context, machine learning (ML) has emerged as a transformative tool, but its adaptation to nanomedicine is hindered by inexistent or small datasets. Here we assembled a large database of inorganic NPs, comprising experimental datasets from 745 preclinical studies in cancer nanomedicine. Using descriptive statistics and explainable ML models we mined this database to gain knowledge of inorganic NP design patterns and inform future NP research for cancer treatment. Our analyses suggest that NP shape and therapy type are prominent features in determining in vivo efficacy, measured as a percentage of tumour reduction. Moreover, our database provides a large-scale open-access resource for discriminative ML that the broader nanotechnology community can utilize. Our work blueprints data mining for translational cancer research and offers evidence for standardizing NP reporting to accelerate and de-risk inorganic NP-based drug delivery, which may help to improve patient outcomes in clinical settings.

Explainable machine learning models for defects detection in industrial processes

Machine learning algorithms in non-linear pattern recognition for defect detection in manufacturing processes are increasingly prevalent in the context of Industry 4.0. This scenario enables factories to develop methodologies for monitoring and controlling the quality of their products, obtaining better operational efficiency indicators, and offering increasingly competitive products in the market. Explainable Artificial Intelligence can facilitate understanding how the models make decisions and assist in tracking anomaly points. This paper proposes an explainability framework for machine learning models in defect detection. This work applied the multinational industry’s approach to the tire manufacturing process. Models such as random forest, gradient boosting decision tree, light gradient boosting machine, logistic regression, support vector machine, and multi-layer perceptron were considered to evaluate the performance of tires in compliance with production standards. Statistical methodologies had to be adopted to deal with the small number of samples training the machine learning models. The random forest and logistic regression models achieved the best performances with an accuracy of 92% and 96%, respectively. The Local Interpretable Model-Agnostic Explanations and SHapley Additive exPlanations methods were used to identify the relevant process variables. The approach presents the global and local interpretations of the results, allowing for a deeper understanding of how process variables discriminate the fault. The proposed framework determines reference values for each selected process variable for tire manufacturing in compliance with quality standards. This approach provides a relevant tool for quality control management in tire manufacturing industries.

How Cloud Droplet Number Concentration Impacts Liquid Water Path and Precipitation in Marine Stratocumulus Clouds—A Satellite-Based Analysis Using Explainable Machine Learning

Aerosol–cloud–precipitation interactions (ACI) are a known major cause of uncertainties in simulations of the future climate. An improved understanding of the in-cloud processes accompanying ACI could help in advancing their implementation in global climate models. This is especially the case for marine stratocumulus clouds, which constitute the most common cloud type globally. In this work, a dataset composed of satellite observations and reanalysis data is used in explainable machine learning models to analyze the relationship between the cloud droplet number concentration (Nd), cloud liquid water path (LWP), and the fraction of precipitating clouds (PF) in five distinct marine stratocumulus regions. This framework makes use of Shapley additive explanation (SHAP) values, allowing to isolate the impact of Nd from other confounding factors, which proved to be very difficult in previous satellite-based studies. All regions display a decrease of PF and an increase in LWP with increasing Nd, despite marked inter-regional differences in the distribution of Nd. Polluted (high Nd) conditions are characterized by an increase of 12 gm−2 in LWP and a decrease of 0.13 in PF on average when compared to pristine (low Nd) conditions. The negative Nd–PF relationship is stronger in high LWP conditions, while the positive Nd–LWP relationship is amplified in precipitating clouds. These findings indicate that precipitation suppression plays an important role in MSC adjusting to aerosol-driven perturbations in Nd.

Performance investigation of osmotically assisted reverse osmosis using explainable machine learning models: A comparative study

Osmotically assisted reverse osmosis (OARO) can treat highly concentrated water and achieve higher recovery rates than reverse osmosis (RO). However, existing mathematical models cannot satisfactorily explain OARO systems due to certain limitations. To explore OARO more dynamically, we aimed to analyze OARO performance based on the laboratory-scale data acquired under different operating conditions, by implementing six machine learning (ML) models with the OARO data, evaluating their predictive capabilities, and investigating the effects of input variables on output variables using Shapley Additive Explanation (SHAP) analysis. In predicting the OARO outputs, the ensemble ML models outperformed the conventional ML models. Incorporating FO/RO membrane datasets significantly enhanced the ML model accuracy and mitigated overfitting issues. Regarding the prediction of water flux, the adjusted determination coefficient values increased by 0.478 for Random Forest and 0.569 for Extreme Gradient Boost in the test step. The SHAP analysis consistently ranked the importance of the input variables for OARO outputs. It revealed the presence of potentially influential but previously unrecognized input variables, such as temperature, for the draw concentration. The results of this study are expected to provide a better understanding of OARO.

Exploration and demonstration of explainable machine learning models in prosthetic rehabilitation-based gait analysis.

Quantitative gait analysis is important for understanding the non-typical walking patterns associated with mobility impairments. Conventional linear statistical methods and machine learning (ML) models are commonly used to assess gait performance and related changes in the gait parameters. Nonetheless, explainable machine learning provides an alternative technique for distinguishing the significant and influential gait changes stemming from a given intervention. The goal of this work was to demonstrate the use of explainable ML models in gait analysis for prosthetic rehabilitation in both population- and sample-based interpretability analyses. Models were developed to classify amputee gait with two types of prosthetic knee joints. Sagittal plane gait patterns of 21 individuals with unilateral transfemoral amputations were video-recorded and 19 spatiotemporal and kinematic gait parameters were extracted and included in the models. Four ML models-logistic regression, support vector machine, random forest, and LightGBM-were assessed and tested for accuracy and precision. The Shapley Additive exPlanations (SHAP) framework was applied to examine global and local interpretability. Random Forest yielded the highest classification accuracy (98.3%). The SHAP framework quantified the level of influence of each gait parameter in the models where knee flexion-related parameters were found the most influential factors in yielding the outcomes of the models. The sample-based explainable ML provided additional insights over the population-based analyses, including an understanding of the effect of the knee type on the walking style of a specific sample, and whether or not it agreed with global interpretations. It was concluded that explainable ML models can be powerful tools for the assessment of gait-related clinical interventions, revealing important parameters that may be overlooked using conventional statistical methods.

Predicting early Alzheimer’s with blood biomarkers and clinical features

Alzheimer’s disease (AD) is an incurable neurodegenerative disorder that leads to dementia. This study employs explainable machine learning models to detect dementia cases using blood gene expression, single nucleotide polymorphisms (SNPs), and clinical data from Alzheimer’s Disease Neuroimaging Initiative (ADNI). Analyzing 623 ADNI participants, we found that the Support Vector Machine classifier with Mutual Information (MI) feature selection, trained on all three data modalities, achieved exceptional performance (accuracy = 0.95, AUC = 0.94). When using gene expression and SNP data separately, we achieved very good performance (AUC = 0.65, AUC = 0.63, respectively). Using SHapley Additive exPlanations (SHAP), we identified significant features, potentially serving as AD biomarkers. Notably, genetic-based biomarkers linked to axon myelination and synaptic vesicle membrane formation could aid early AD detection. In summary, this genetic-based biomarker approach, integrating machine learning and SHAP, shows promise for precise AD diagnosis, biomarker discovery, and offers novel insights for understanding and treating the disease. This approach addresses the challenges of accurate AD diagnosis, which is crucial given the complexities associated with the disease and the need for non-invasive diagnostic methods.

Conventional and unconventional T-cell responses contribute to the prediction of clinical outcome and causative bacterial pathogen in sepsis patients.

Sepsis is characterized by a dysfunctional host response to infection culminating in life-threatening organ failure that requires complex patient management and rapid intervention. Timely diagnosis of the underlying cause of sepsis is crucial, and identifying those at risk of complications and death is imperative for triaging treatment and resource allocation. Here, we explored the potential of explainable machine learning models to predict mortality and causative pathogen in sepsis patients. By using a modelling pipeline employing multiple feature selection algorithms, we demonstrate the feasibility of identifying integrative patterns from clinical parameters, plasma biomarkers, and extensive phenotyping of blood immune cells. While no single variable had sufficient predictive power, models that combined five and more features showed a macro area under the curve (AUC) of 0.85 to predict 90-day mortality after sepsis diagnosis, and a macro AUC of 0.86 to discriminate between Gram-positive and Gram-negative bacterial infections. Parameters associated with the cellular immune response contributed the most to models predictive of 90-day mortality, most notably, the proportion of T cells among PBMCs, together with expression of CXCR3 by CD4+ T cells and CD25 by mucosal-associated invariant T (MAIT) cells. Frequencies of Vδ2+ γδ T cells had the most profound impact on the prediction of Gram-negative infections, alongside other T-cell-related variables and total neutrophil count. Overall, our findings highlight the added value of measuring the proportion and activation patterns of conventional and unconventional T cells in the blood of sepsis patients in combination with other immunological, biochemical, and clinical parameters.

Explainable machine learning models with privacy

AbstractThe importance of explainable machine learning models is increasing because users want to understand the reasons behind decisions in data-driven models. Interpretability and explainability emerge from this need to design comprehensible systems. This paper focuses on privacy-preserving explainable machine learning. We study two data masking techniques: maximum distance to average vector (MDAV) and additive noise. The former is for achieving k-anonymity, and the second uses Laplacian noise to avoid record leakage and provide a level of differential privacy. We are interested in the process of developing data-driven models that, at the same time, make explainable decisions and are privacy-preserving. That is, we want to avoid the decision-making process leading to disclosure. To that end, we propose building models from anonymized data. More particularly, data that are k-anonymous or that have been anonymized add an appropriate level of noise to satisfy some differential privacy requirements. In this paper, we study how explainability has been affected by these data protection procedures. We use TreeSHAP as our technique for explainability. The experiments show that we can keep up to a certain degree both accuracy and explainability. So, our results show that some trade-off between privacy and explainability is possible for data protection using k-anonymity and noise addition.

Explainable ensemble learning predictive model for thermal conductivity of cement-based foam

Cement-based foam has emerged as a strong contender in sustainable construction owing to its superior thermal and sound insulation properties, fire resistance, and cost-effectiveness. To effectively use cement-based foam as a thermal insulation material, it is important to accurately predict its thermal conductivity. The current study aims at coining an accurate methodology for predicting the thermal conductivity of cement-based foam using state-of-the-art machine learning techniques. A comprehensive experimental dataset of 504 data points was developed and used for training ensemble learning models including XGBoost, CatBoost, LightGBM and Random Forest. The independent variables of this dataset affecting the thermal conductivity are the cast density, percentage of pozzolan, porosity, percentage of moisture, and duration of hydration in days. Using the Isolation Forest algorithm proved effective in detecting and eliminating outliers in the dataset. All the ensemble learning techniques explored in this study achieved superior predictive accuracy with a coefficient of determination greater than 0.98 on the test dataset. The influence of the input features on the thermal conductivity was visualized using the SHapley Additive exPlanations (SHAP) approach and individual conditional expectation (ICE) plots. The cast density had the greatest effect on thermal conductivity. The explainable machine learning models demonstrated superior accuracy, efficiency, and reliability in estimating the thermal insulation of cement-based foam, opening the door for wider acceptance of this material in sustainable energy efficient construction.

Investigation on explainable machine learning models to predict chronic kidney diseases

Chronic kidney disease (CKD) is a major worldwide health problem, affecting a large proportion of the world’s population and leading to higher morbidity and death rates. The early stages of CKD sometimes present without visible symptoms, causing patients to be unaware. Early detection and treatments are critical in reducing complications and improving the overall quality of life for people afflicted. In this work, we investigate the use of an explainable artificial intelligence (XAI)-based strategy, leveraging clinical characteristics, to predict CKD. This study collected clinical data from 491 patients, comprising 56 with CKD and 435 without CKD, encompassing clinical, laboratory, and demographic variables. To develop the predictive model, five machine learning (ML) methods, namely logistic regression (LR), random forest (RF), decision tree (DT), Naïve Bayes (NB), and extreme gradient boosting (XGBoost), were employed. The optimal model was selected based on accuracy and area under the curve (AUC). Additionally, the SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) algorithms were utilized to demonstrate the influence of the features on the optimal model. Among the five models developed, the XGBoost model achieved the best performance with an AUC of 0.9689 and an accuracy of 93.29%. The analysis of feature importance revealed that creatinine, glycosylated hemoglobin type A1C (HgbA1C), and age were the three most influential features in the XGBoost model. The SHAP force analysis further illustrated the model’s visualization of individualized CKD predictions. For further insights into individual predictions, we also utilized the LIME algorithm. This study presents an interpretable ML-based approach for the early prediction of CKD. The SHAP and LIME methods enhance the interpretability of ML models and help clinicians better understand the rationale behind the predicted outcomes more effectively.