Shapley Value Analysis Research Articles

Understanding Polymers Through Transfer Learning and Explainable AI

In this work we study the use of artificial intelligence models, particularly focusing on transfer learning and interpretability, to predict polymer properties. Given the challenges imposed by data scarcity in polymer science, transfer learning offers a promising solution by using learnt features of models pre-trained on other datasets. We conducted a comparative analysis of direct modelling and transfer learning-based approaches using a polyacrylates’ glass transitions dataset as a proof-of-concept study. The AI models utilized tokenized SMILES strings to represent polymer structures, with convolutional neural networks processing these representations to predict Tg. To enhance model interpretability, Shapley value analysis was employed to assess the contribution of specific chemical groups to the predictions. The results indicate that while transfer learning provides robust predictive capabilities, direct modelling on polymer-specific data offers superior performance, particularly in capturing the complex interactions influencing Tg. This work highlights the importance of model interpretability and the limitations of applying molecular-level models to polymer systems.

Accurate Estimation of Gross Primary Production of Paddy Rice Cropland with UAV Imagery-Driven Leaf Biochemical Model

Accurate estimation of gross primary production (GPP) of paddy rice fields is essential for understanding cropland carbon cycles, yet remains challenging due to spatial heterogeneity. In this study, we integrated high-resolution unmanned aerial vehicle (UAV) imagery into a leaf biochemical properties-based model for improving GPP estimation. The key parameter, maximum carboxylation rate at the top of the canopy (Vcmax,025), was quantified using various spatial information representation methods, including mean (μref) and standard deviation (σref) of reflectance, gray-level co-occurrence matrix (GLCM)-based features, local binary pattern histogram (LBPH), and convolutional neural networks (CNNs). Our models were evaluated using a two-year eddy covariance (EC) system and UAV measurements. The result shows that incorporating spatial information can vastly improve the accuracy of Vcmax,025 and GPP estimation. CNN methods achieved the best Vcmax,025 estimation, with an R of 0.94, an RMSE of 19.44 μmol m−2 s−1, and an MdAPE of 11%, and further produced highly accurate GPP estimates, with an R of 0.92, an RMSE of 6.5 μmol m−2 s−1, and an MdAPE of 23%. The μref-GLCM texture features and μref-LBPH joint-driven models also gave promising results. However, σref contributed less to Vcmax,025 estimation. The Shapley value analysis revealed that the contribution of input features varied considerably across different models. The CNN model focused on nir and red-edge bands and paid much attention to the subregion with high spatial heterogeneity. The μref-LBPH joint-driven model mainly prioritized reflectance information. The μref-GLCM-based features joint-driven model emphasized the role of GLCM texture indices. As the first study to leverage the spatial information from high-resolution UAV imagery for GPP estimation, our work underscores the critical role of spatial information and provides new insight into monitoring the carbon cycle.

Identification of optimal waste-to-energy strategies for sustainable development in Iran: SWOT analysis, hybrid MCDM methods, and game theory

The growing global focus on environmental sustainability and the pressing need for efficient waste management and energy crisis mitigation have prompted extensive investigations into sustainable waste management strategies. Given Iran's economic development and population growth, the country faces challenges in energy supply and waste management, making waste-to-energy (WtE) applications a promising solution for sustainable waste management. This study employs a combination of SWOT (Strengths, Weaknesses, Opportunities, and Threats) analysis, Multi-Criteria Decision-Making (MCDM) approaches, and game theory to explore WtE technologies and processes, evaluate their applicability within Iran's context, and identify the most effective WtE strategies for the country. Furthermore, game theory and fuzzy Shapley value analysis were utilized to determine the optimal combination of WtE development strategies. The strategy player “SO” held the highest position of power, with a fuzzy Shapley value of (0.391, 1.744, 3.264), resulting in the identification of the optimal strategy combination of SO2ST2WO1WT2, with a value of (0.429, 0.679, 0.893). The recommended combination emphasizes landfill with gas capture, development of incineration power plants, and integrated resource recovery through residue-derived fuel (RDF) implementation. Achieving these objectives necessitates energy policy reforms, pricing adjustments, improvements in the energy subsidy system, and infrastructure and market development.

Prediction of pre-eclampsia with machine learning approaches: Leveraging important information from routinely collected data

BackgroundGlobally, pre-eclampsia (PE) is a leading cause of maternal and perinatal morbidity and mortality. PE prediction using routinely collected data has the advantage of being widely applicable, particularly in low-resource settings. Early intervention for high-risk women might reduce PE incidence and related complications. We aimed to replicate our machine learning (ML) published work predicting another maternal condition (gestational diabetes) to (1) predict PE using routine health data, (2) identify the optimal ML model, and (3) compare it with logistic regression approach. MethodsData were from a large health service network with 48,250 singleton pregnancies between January 2016 and June 2021. Supervised ML models were employed. Maternal clinical and medical characteristics were the feature variables (predictors), and a 70/30 data split was used for training and testing the model. Predictive performance was assessed using area under the curve (AUC) and calibration plots. Shapley value analysis assessed the contribution of feature variables. ResultsThe random forest approach provided excellent discrimination with an AUC of 0.84 (95% CI: 0.82–0.86) and highest prediction accuracy (0.79); however, the calibration curve (slope of 1.21, 95% CI 1.13–1.30) was acceptable only for a threshold of 0.3 or less. The next best approach was extreme gradient boosting, which provided an AUC of 0.77 (95% CI: 0.76–0.79) and well-calibrated (slope of 0.93, 95% CI 0.85–1.01). Logistic regression provided good discrimination performance with an AUC of 0.75 (95% CI: 0.74–0.76) and perfect calibration. Nulliparous, pre-pregnancy body mass index, previous pregnancy with prior PE, maternal age, family history of hypertension, and pre-existing hypertension and diabetes were the top-ranked features in Shapley value analysis. ConclusionTwo ML models created the highest-performing prediction using routinely collected data to identify women at high risk of PE, with acceptable discrimination. However, to confirm this result and also examine model generalisability, external validation studies are needed in other settings, utilising standardised prognostic factors.

Explainable artificial intelligence for the interpretation of ensemble learning performance in algal bloom estimation.

Chlorophyll-a (Chl-a) concentrations, a key indicator of algal blooms, were estimated using the XGBoost machine learning model with 23 variables, including water quality and meteorological factors. The model performance was evaluated using three indices: root mean square error (RMSE), RMSE-observation standard deviation ratio (RSR), and Nash-Sutcliffe efficiency. Nine datasets were created by averaging 1hour data to cover time frequencies ranging from 1hour to 1month. The dataset with relatively high observation frequencies (1-24 h) maintained stability, with an RSR ranging between 0.61 and 0.65. However, the model's performance declined significantly for datasets with weekly and monthly intervals. The Shapley value (SHAP) analysis, an explainable artificial intelligence method, was further applied to provide a quantitative understanding of how environmental factors in the watershed impact the model's performance and is also utilized to enhance the practical applicability of the model in the field. The number of input variables for model construction increased sequentially from 1 to 23, starting from the variable with the highest SHAP value to that with the lowest. The model's performance plateaued after considering five or more variables, demonstrating that stable performance could be achieved using only a small number of variables, including relatively easily measured data collected by real-time sensors, such as pH, dissolved oxygen, and turbidity. This result highlights the practicality of employing machine learning models and real-time sensor-based measurements for effective on-site water quality management. PRACTITIONER POINTS: XAI quantifies the effects of environmental factors on algal bloom prediction models The effects of input variable frequency and seasonality were analyzed using XAI XAI analysis on key variables ensures cost-effective model development.

Diabetes prediction model based on GA-XGBoost and stacking ensemble algorithm.

Diabetes, as an incurable lifelong chronic disease, has profound and far-reaching effects on patients. Given this, early intervention is particularly crucial, as it can not only significantly improve the prognosis of patients but also provide valuable reference information for clinical treatment. This study selected the BRFSS (Behavioral Risk Factor Surveillance System) dataset, which is publicly available on the Kaggle platform, as the research object, aiming to provide a scientific basis for the early diagnosis and treatment of diabetes through advanced machine learning techniques. Firstly, the dataset was balanced using various sampling methods; secondly, a Stacking model based on GA-XGBoost (XGBoost model optimized by genetic algorithm) was constructed for the risk prediction of diabetes; finally, the interpretability of the model was deeply analyzed using Shapley values. The results show: (1) Random oversampling, ADASYN, SMOTE, and SMOTEENN were used for data balance processing, among which SMOTEENN showed better efficiency and effect in dealing with data imbalance. (2) The GA-XGBoost model optimized the hyperparameters of the XGBoost model through a genetic algorithm to improve the model's predictive accuracy. Combined with the better-performing LightGBM model and random forest model, a two-layer Stacking model was constructed. This model not only outperforms single machine learning models in predictive effect but also provides a new idea and method in the field of model integration. (3) Shapley value analysis identified features that have a significant impact on the prediction of diabetes, such as age and body mass index. This analysis not only enhances the transparency of the model but also provides more precise treatment decision support for doctors and patients. In summary, this study has not only improved the accuracy of predicting the risk of diabetes by adopting advanced machine learning techniques and model integration strategies but also provided a powerful tool for the early diagnosis and personalized treatment of diabetes.

Optimal Service Strategies of Online Platform Based on Purchase Behavior

In the rapidly evolving platform economy, online platforms have emerged as pivotal providers of digital services to sellers. The paper investigates how online platforms optimize service strategies based on consumers’ purchase behavior, influencing sellers’ pricing and social welfare. Using a two-period Hotelling model and a cooperative game framework, we discover that the optimal service strategies of a platform with data collecting capabilities are collaborating with two sellers to offer to extend services to new consumers in the second period, maximizing profits for all sellers and platform. Applying Shapley value analysis, we determine the platform’s equitable service charge strategies. When sellers adopt behavior-based pricing (BBP), the pricing escalates in the first period, and the platform’s optimal service strategies also enhance the pricing of sellers. However, in the second period, BBP intensifies competition, leading to generally lower pricing. Our findings suggest that optimal pricing in the second period for new consumers should increase with enhanced quality perception, which is provided by the platform’s digital services and heightened by consumers’ privacy concerns, while decreasing for regular consumers. Lastly, we offer policy recommendations, exploring optimal regulatory scenarios—limiting or not limiting data collection—to maximize social welfare or consumer surplus, and the Mathematica software is used to identify distinct optimal policy intervals.

Combination of integrated machine learning model frameworks and infrared spectroscopy towards fast and interpretable characterization of model pyrolysis oil

Model pyrolysis oil (bio-oil) is a promising bio-fuel with complex and unstable contents, making its fast characterization an essential demand in industrial production. This study proposed an integrated machine learning framework to predict elemental composition, low heating value, and unsaturated concentration from infrared spectra, achieving fast and interpretable characterization of bio-oil. In the integrated framework, a peak loading-based strategy was used to dimensionally reduce the spectral data. Bayesian optimized random forest (RF) and extreme gradient boosting (XGBoost) models were used to predict bio-oil properties from dimensionally reduced spectral data. Ensemble learning was used to combine RF and XGBoost models together for better predicting performance. Results showed that the proposed characterization method achieved an average accuracy of 99.53 %, a low RMSE value of 0.726, and an R2 of 0.98. The Shapley value analysis revealed that the vibration of NH2 stretch (1594 cm−1), C-H stretch (2868 cm−1), and C-N stretch in the aromatic ring (1229 cm−1) have a significant contribution to the characterization results. The working mechanism of the proposed characterization method was interpreted by the internal relationship among spectral peak location/height, functional group species/amount, and the predicted characteristics. The results are hoped to serve quality control in production of bio-oil.

Integrating supervised and unsupervised learning approaches to unveil critical process inputs

This study introduces a machine learning framework tailored to large-scale industrial processes characterized by a plethora of numerical and categorical inputs. The framework aims to (i) discern critical parameters that influence the output and (ii) generate accurate out-of-sample qualitative and quantitative predictions of production outcomes. Specifically, we address the pivotal question of the significance of each input in shaping the process outcome, using an industrial Chemical Vapor Deposition (CVD) process as an example. The initial objective involves merging subject matter expertise and clustering techniques exclusively on the process output, here, coating thickness measurements at various positions in the reactor. This approach identifies groups of production runs that share similar qualitative characteristics, such as film mean thickness and standard deviation. In particular, the differences of the outcomes represented by the different clusters can be attributed to differences in specific inputs, indicating that these inputs are potentially critical to the production outcome. Shapley value analysis corroborates the formed hypotheses. Leveraging this insight, we subsequently implement supervised classification and regression methods using the identified critical process inputs. The proposed methodology proves to be valuable in scenarios with a multitude of inputs and insufficient data for the direct application of deep learning techniques, providing meaningful insights into the underlying processes.

Ground-truth validation of uni- and multivariate lesion inference approaches.

Lesion analysis aims to reveal the causal contributions of brain regions to brain functions. Various strategies have been used for such lesion inferences. These approaches can be broadly categorized as univariate or multivariate methods. Here we analysed data from 581 patients with acute ischaemic injury, parcellated into 41 Brodmann areas, and systematically investigated the inferences made by two univariate and two multivariate lesion analysis methods via ground-truth simulations, in which we defined a priori contributions of brain areas to assumed brain function. Particularly, we analysed single-region models, with only single areas presumed to contribute functionally, and multiple-region models, with two contributing regions that interacted in a synergistic, redundant or mutually inhibitory mode. The functional contributions could vary in proportion to the lesion damage or in a binary way. The analyses showed a considerably better performance of the tested multivariate than univariate methods in terms of accuracy and mis-inference error. Specifically, the univariate approaches of Lesion Symptom Mapping as well as Lesion Symptom Correlation mis-inferred substantial contributions from several areas even in the single-region models, and also after accounting for lesion size. By contrast, the multivariate approaches of Multi-Area Pattern Prediction, which is based on machine learning, and Multi-perturbation Shapley value Analysis, based on coalitional game theory, delivered consistently higher accuracy and specificity. Our findings suggest that the tested multivariate approaches produce largely reliable lesion inferences, without requiring lesion size consideration, while the application of the univariate methods may yield substantial mis-localizations that limit the reliability of functional attributions.

Microgrid F36ault Detection Method Based on Lightweight Gradient Boosting Machine–Neural Network Combined Modeling

The intelligent architecture based on the microgrid (MG) system enhances distributed energy access through an effective line network. However, the increased paths between power sources and loads complicate the system’s topology. This complexity leads to multidirectional line currents, heightening the risk of current loops, imbalances, and potential short-circuit faults. To address these challenges, this study proposes a new approach to accurately locate and identify faults based on MG lines. Initially, characteristic indices such as fault voltage, voltage fundamentals at each MG measurement point, and extracted features like peak voltage values in specific frequency bands, phase-to-phase voltage differences, and the sixth harmonic components are utilized as model inputs. Subsequently, these features are classified using the Lightweight Gradient Boosting Machine (LightGBM), complemented by the bagging (Bootstrap Aggregating) ensemble learning algorithm to consolidate multiple strong LightGBM classifiers in parallel. The output classification results of the integrated model are then fed into a neural network (NN) for further training and learning for fault-type identification and localization. In addition, a Shapley value analysis is introduced to quantify the contribution of each feature and visualize the fault diagnosis decision-making process. A comparative analysis with existing methodologies demonstrates that the LightGBM-NN model not only improves fault detection accuracy but also exhibits greater resilience against noise interference. The introduction of the bagging method, by training multiple base models on the initial classification subset of LightGBM and aggregating their prediction results, can reduce the model variance and prevent overfitting, thus improving the stability and accuracy of fault detection in the combined model and making the interpretation of the Shapley value more stable and reliable. The introduction of the Shapley value analysis helps to quantify the contribution of each feature to improve the transparency and understanding of the combined model’s troubleshooting decision-making process, reduces the model’s subsequent collection of data from different line operations, further optimizes the collection of line feature samples, and ensures the model’s effectiveness and adaptability.

Community medical service construction: identifying factors that influence medical choice for patients with non-communicable chronic diseases in the Southwest China

BackgroundCommunity medical institutions play a vital role in China’s healthcare system. While the number of these institutions has increased in recent years, their construction contents remain insufficient. The potential of community medical institutions in preventing, screening, diagnosing, and treating non-communicable chronic diseases (NCDs) has not been fully utilized. This study aims to assess the status of construction contents in community medical institutions in Southwest China and examine how these contents influence the medical choices of NCD patients.MethodsDescriptive statistics were used to evaluate the construction content of community medical institutions. Multiple-sets of multinomial logistic regression were employed to analyze the associations and marginal impacts between construction content and medical choices. Shapley value analysis was applied to determine the contribution and ranking of these impacts.ResultsDescriptive statistics revealed satisfactory construction contents in community medical institutions. Notably, factors such as service attitude, nursing services, expert consultations, charging standards, medical equipment, medical examinations, privacy protection, and referrals significantly influenced medical choices. Among these, service attitude, charging standards, and privacy protection had the most significant marginal improvement effects on NCD patients’ choices, with improvements of 12.7%, 10.2%, and 5.9%, respectively. The combined contribution of privacy protection, medical examinations, service attitude, charging standards, and nursing services to medical choices exceeded 80%.ConclusionOptimizing the service contents of community institutions can encourage NCD patients to seek medical care at grassroots hospitals. This study addresses crucial gaps in existing literature and offers practical insights for implementing new medical reform policies, particularly in underdeveloped regions of Southwest China focusing on hierarchical diagnosis and treatment.

Abstract 4939: Computational pathology approach for prognostic advancements and therapeutic benefits in gastrointestinal cancer: A multi-centric retrospective study

Abstract Background: The current cancer staging methods cannot accurately predict survival and therapeutic benefits in cancer patients. Digital pathomics is an emerging field with potential to revolutionize disease evaluation. The present study aims to develop and validate a deep learning pathomics signature from digital hematoxylin-eosin (H&E) staining slides to predict the survival of pan-gastrointestinal cancer, and further investigated its associations with chemotherapy and immunotherapy response. P atients and Methods: This international multicenter study included 2,463 patients with pan-gastrointestinal cancer from twelve cohorts of seven cancer centers, among which 1653 patients were diagnosed with GC. We proposed a deep learning pathomics signature (DLPS) by integrating information on three scales from the whole slide images (WSIs) of H&E, including pathomics nucleus, pathomics microenvironment, and single-cell spatial distribution features. Next, we evaluated the predictive accuracy of DLPS for prognosis, chemotherapy response, and immunotherapy response. To interpret the predictive ability of the DLPS, we analyzed multi-omics data and employed the Shapley value strategy to provide its biological insights and importance ranking. Results: The DLPS was significantly associated with overall survival in patients with GC (hazard ratio range: 1.203-2.717, all p < 0.001), which was further validated in other gastrointestinal cancers (all p < 0.02). The DLPS remained an independent predictor of prognosis in multivariable analysis (all p < 0.001). Furthermore, a nomogram incorporating the DLPS and TNM stage shows significantly improved accuracy in predicting cancer prognosis compared to that with TNM stage alone (all p < 0.05). Shapley value analysis highlighted DLPS as the strongest predictor for prognosis. Importantly, GC patients with a low-DLPS (but not those with a high-DLPS) exhibited substantial benefits from adjuvant chemotherapy (all p < 0.05). Furthermore, we found the objective responses of anti-PD-1 immunotherapy is significantly higher in the low-DLPS group (29.6%) than in the high-DLPS group (8.3%, p < 0.05). Upon analyzing multi-omics data, we found that a higher DLPS was positively correlated with tumor promoting, chemotherapy resistance, immune tumor microenvironment and metabolic signaling. Conclusion: The DLPS enabled improved assessment on prognosis, and has the potential to identify patients who will benefit from adjuvant chemotherapy and immunotherapy, which can further be extended to many gastrointestinal cancers or other solid tumors. Citation Format: Taojun Zhang, Zepang Sun, Zhe Li, M. Usman Ahmad, Md Tauhidul Islam, Fan Yang, Zhenhui Li, Yuming Jiang. Computational pathology approach for prognostic advancements and therapeutic benefits in gastrointestinal cancer: A multi-centric retrospective study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4939.

Single-cell multi-ome regression models identify functional and disease-associated enhancers and enable chromatin potential analysis

We present a gene-level regulatory model, single-cell ATAC + RNA linking (SCARlink), which predicts single-cell gene expression and links enhancers to target genes using multi-ome (scRNA-seq and scATAC–seq co-assay) sequencing data. The approach uses regularized Poisson regression on tile-level accessibility data to jointly model all regulatory effects at a gene locus, avoiding the limitations of pairwise gene–peak correlations and dependence on peak calling. SCARlink outperformed existing gene scoring methods for imputing gene expression from chromatin accessibility across high-coverage multi-ome datasets while giving comparable to improved performance on low-coverage datasets. Shapley value analysis on trained models identified cell-type-specific gene enhancers that are validated by promoter capture Hi-C and are 11× to 15× and 5× to 12× enriched in fine-mapped eQTLs and fine-mapped genome-wide association study (GWAS) variants, respectively. We further show that SCARlink-predicted and observed gene expression vectors provide a robust way to compute a chromatin potential vector field to enable developmental trajectory analysis.

Analysis of Artificial Neural Networks on Predicting Dementia

Dementia, a chronic neurological disorder characterized by memory loss and impaired cognitive ability, has affected tens of millions of people globally and projections estimate this number to rise in the coming decade. Despite its widespread impact, dementia remains one of many diseases with no true cure or definitive method of diagnosis. This study aims to explore the question of how artificial neural networks can be used to predict dementia. Analyzing current research on neural networks for disease diagnosis highlights a notable gap, known as the "black box conundrum," indicating the challenge of tracing steps leading to a given output. Another gap in the literature is the absence of comparative data on factors influencing successful neural network algorithms. Conducting a comprehensive cross-continental comparative analysis across North America, Asia, and Europe, and utilizing an experimental design with the UCI ML Repository’s Parkinson’s Database, this study specifically aims to fill the identified gaps. The findings reveal key elements for successful neural networks: a data sample exceeding 100, clinically intelligible data values, and a backpropagation model. In the experimental phase, a comparison between a backpropagation model and logistic regression, paired with Shapley value analysis, elucidated a 94.87% accuracy for the former, surpassing the latter's 92.3%. These findings extend beyond dementia, providing valuable insights for neural network applications in all healthcare settings. Future research endeavors should focus on refining and optimizing neural networks with the ultimate goal of achieving 100% accuracy and maximized efficiency.

A novel hybrid modelling framework for GPP estimation: Integrating a multispectral surface reflectance based [formula omitted] simulator into the process-based model

Terrestrial gross primary productivity (GPP) is the key element in the carbon cycle process. Accurate GPP estimation hinges on the maximum carboxylation rate (Vcmax,025). The high uncertainty in deriving ecosystem-level Vcmax,025 has long hampered efforts toward the performance of the GPP model. Recently studies suggest the strong relationship between spectral reflectance and Vcmax,025. We proposed the multispectral surface reflectance-driven Vcmax,025 simulator using the fully connected deep neural network and built the hybrid modelling framework for GPP estimation by integrating the data-driven Vcmax,025 simulator in the process-based model. The performance of hybrid GPP model was evaluated at 95 flux sites. The result shows that the multispectral surface reflectance-driven Vcmax,025 simulator acquires the satisfactory estimation, with correlation coefficient (R), root mean square error (RMSE) and median absolute percentage error (MdAPE) ranging from 0.34 to 0.80, 14 to 43 μmol m−2 s−1 and 21 % to 66 % across different land cover types, respectively. The hybrid framework generates good GPP estimates with R, RMSE and MdAPE varying from 0.76 to 0.89, 1.79 to 6.16 μmol m−2 s−1 and 27 % to 90 %, respectively. Compared with EVI-driven method, the multispectral surface reflectance significantly improves the Vcmax,025 and GPP estimates, with MdAPE declining by 0.6 %–18 % and 1 % to 21 %, respectively. The Shapley value analysis reveals that red (620–670 nm), near-infrared (841–876 nm) and shortwave infrared (1628–1652 nm and 2105–2155 nm) are the key bands for Vcmax,025 estimation. This study highlights the potential of multispectral surface reflectance for quantifying ecosystem-level Vcmax,025. The new hybrid framework fully extracts the information of all available spectral bands using deep learning to reduce parameter uncertainty while maintains the description of photosynthetic process to ensure its physical reasonability. It can serve as a powerful tool for accurate global GPP estimation.

Machine learning predicted inelasticity in defective two-dimensional transition metal dichalcogenides using SHAP analysis.

The manipulation of crystallographic defects in 2H-transition metal dichalcogenides (2H-TMDCs), whether pre- or post-synthesis, has garnered significant interest recently, as it holds the promise of tuning the thermal, chemical, and electronic properties of these materials. However, such desirable improvements often come at the cost of deteriorated elastic and inelastic properties, which may lead to serious concerns considering mechanical reliability issues. Therefore, persistent efforts are needed to explore the effects of energetically favorable vacancies on the mechanical properties of 2D TMDCs for an effective tuning of material properties for versatile applications. In this context, machine learning models trained on data based on molecular models can not only provide fast, efficient material models but also unearth crucial structure-property relations. However, such efforts are at an early stage of development. In this study, machine learning and deep learning techniques are used to analyze the mechanical properties of 2D transition metal dichalcogenides (TMDCs) in both pristine and defect forms. The goal is to predict failure stress, strain to failure, and strength based on chirality and strain. Various crystallographic defects were considered, and extensive molecular dynamics simulations were performed. XGBoost and densely connected neural network (DenseNet) algorithms were used to make accurate state-of-the-art predictions, and comparative evaluation and Shapley value analysis of both models are presented to improve interpretability.

Alternative splicing events as peripheral biomarkers for motor learning deficit caused by adverse prenatal environments.

Severity of neurobehavioral deficits in children born from adverse pregnancies, such as maternal alcohol consumption and diabetes, does not always correlate with the adversity's duration and intensity. Therefore, biological signatures for accurate prediction of the severity of neurobehavioral deficits, and robust tools for reliable identification of such biomarkers, have an urgent clinical need. Here, we demonstrate that significant changes in the alternative splicing (AS) pattern of offspring lymphocyte RNA can function as accurate peripheral biomarkers for motor learning deficits in mouse models of prenatal alcohol exposure (PAE) and offspring of mother with diabetes (OMD). An aptly trained deep-learning model identified 29 AS events common to PAE and OMD as superior predictors of motor learning deficits than AS events specific to PAE or OMD. Shapley-value analysis, a game-theory algorithm, deciphered the trained deep-learning model's learnt associations between its input, AS events, and output, motor learning performance. Shapley values of the deep-learning model's input identified the relative contribution of the 29 common AS events to the motor learning deficit. Gene ontology and predictive structure-function analyses, using Alphafold2 algorithm, supported existing evidence on the critical roles of these molecules in early brain development and function. The direction of most AS events was opposite in PAE and OMD, potentially from differential expression of RNA binding proteins in PAE and OMD. Altogether, this study posits that AS of lymphocyte RNA is a rich resource, and deep-learning is an effective tool, for discovery of peripheral biomarkers of neurobehavioral deficits in children of diverse adverse pregnancies.

Using interpretable gradient‐boosted decision‐tree ensembles to uncover novel dynamical relationships governing monsoon low‐pressure systems

AbstractLow‐pressure systems (LPSs) are the primary rainbringers of the South Asian monsoon. Yet, their interactions with the large‐scale monsoon circulation, as well as the highly variable land and sea surfaces they pass over, are complex and generally not well understood. In this article, we present a novel, top‐down approach to investigate these relationships and quantify their importance in describing LPS behaviour. We also show that, if the approach is sufficiently well posed, it is productive at hypothesis generation. For each of five predictands (i.e., LPS intensification rate, propagation speed/direction, post‐landfall survival, peak intensity, and precipitation rate) we train an additive decision‐tree ensemble model using the XGBoost algorithm. Shapley value analysis is then applied to the models to determine which variables are important predictors and to establish their relationship with the predictand, with additional analysis following cases of interest. Novel relationships established using this technique include that LPS vorticity intensifies preferentially in the early morning at the same time as the peak in the diurnal cycle of their convection occurs, that vertical wind shear suppresses continued growth of strong LPSs, that large‐scale barotropic instability plays an important role in both the inland penetration and peak intensity of LPSs, and that LPS propagation depends on the depth of its vortex with shallower LPSs advected by low‐level winds and taller LPSs advected by mid‐level winds. We also use this framework to identify and discuss potential new avenues of research for monsoon LPSs.

CIRCE: Web-Based Platform for the Prediction of Cannabinoid Receptor Ligands Using Explainable Machine Learning.

The endocannabinoid system, which includes cannabinoid receptor 1 and 2 subtypes (CB1R and CB2R, respectively), is responsible for the onset of various pathologies including neurodegeneration, cancer, neuropathic and inflammatory pain, obesity, and inflammatory bowel disease. Given the high similarity of CB1R and CB2R, generating subtype-selective ligands is still an open challenge. In this work, the Cannabinoid Iterative Revaluation for Classification and Explanation (CIRCE) compound prediction platform has been generated based on explainable machine learning to support the design of selective CB1R and CB2R ligands. Multilayer classifiers were combined with Shapley value analysis to facilitate explainable predictions. In test calculations, CIRCE predictions reached ∼80% accuracy and structural features determining ligand predictions were rationalized. CIRCE was designed as a web-based prediction platform that is made freely available as a part of our study.