Predicting the efficiency of arsenic immobilization in soils by biochar using machine learning.

Arsenic (As) pollution in soils is a pervasive environmental issue. Biochar immobilization offers a promising solution for addressing soil As contamination. The efficiency of biochar in immobilizing As in soils primarily hinges on the characteristics of both the soil and the biochar. However, the influence of a specific property on As immobilization varies among different studies, and the development and application of arsenic passivation materials based on biochar often rely on empirical knowledge. To enhance immobilization efficiency and reduce labor and time costs, a machine learning (ML) model was employed to predict As immobilization efficiency before biochar application. In this study, we collected a dataset comprising 182 data points on As immobilization efficiency from 17 publications to construct three ML models. The results demonstrated that the random forest (RF) model outperformed gradient boost regression tree and support vector regression models in predictive performance. Relative importance analysis and partial dependence plots based on the RF model were conducted to identify the most crucial factors influencing As immobilization. These findings highlighted the significant roles of biochar application time and biochar pH in As immobilization efficiency in soils. Furthermore, the study revealed that Fe-modified biochar exhibited a substantial improvement in As immobilization. These insights can facilitate targeted biochar property design and optimization of biochar application conditions to enhance As immobilization efficiency.

Rapid detection of colored and colorless macro- and micro-plastics in complex environment via near-infrared spectroscopy and machine learning.

To better understand the migration behavior of plastic fragments in the environment, development of rapid non-destructive methods for in-situ identification and characterization of plastic fragments is necessary. However, most of the studies had focused only on colored plastic fragments, ignoring colorless plastic fragments and the effects of different environmental media (backgrounds), thus underestimating their abundance. To address this issue, the present study used near-infrared spectroscopy to compare the identification of colored and colorless plastic fragments based on partial least squares-discriminant analysis (PLS-DA), extreme gradient boost, support vector machine and random forest classifier. The effects of polymer color, type, thickness, and background on the plastic fragments classification were evaluated. PLS-DA presented the best and most stable outcome, with higher robustness and lower misclassification rate. All models frequently misinterpreted colorless plastic fragments and its background when the fragment thickness was less than 0.1mm. A two-stage modeling method, which first distinguishes the plastic types and then identifies colorless plastic fragments that had been misclassified as background, was proposed. The method presented an accuracy higher than 99% in different backgrounds. In summary, this study developed a novel method for rapid and synchronous identification of colored and colorless plastic fragments under complex environmental backgrounds.

Characteristics of pain empathic networks in healthy and primary dysmenorrhea women: an fMRI study.

Pain empathy enables us to understand and share how others feel pain. Few studies have investigated pain empathy-related functional interactions at the whole-brain level across all networks. Additionally, women with primary dysmenorrhea (PDM) have abnormal pain empathy, and the association among the whole-brain functional network, pain, and pain empathy remain unclear. Using resting-state functional magnetic resonance imaging (fMRI) and machine learning analysis, we identified the brain functional network connectivity (FNC)-based features that are associated with pain empathy in two studies. Specifically, Study 1 examined 41 healthy controls (HCs), while Study 2 investigated 45 women with PDM. Additionally, in Study 3, a classification analysis was performed to examine the differences in FNC between HCs and women with PDM. Pain empathy was evaluated using a visual stimuli experiment, and trait and state of menstrual pain were recorded. In Study 1, the results showed that pain empathy in HCs relied on dynamic interactions across whole-brain networks and was not concentrated in a single or two brain networks, suggesting the dynamic cooperation of networks for pain empathy in HCs. In Study 2, PDM exhibited a distinctive network for pain empathy. The features associated with pain empathy were concentrated in the sensorimotor network (SMN). In Study 3, the SMN-related dynamic FNC could accurately distinguish women with PDM from HCs and exhibited a significant association with trait menstrual pain. This study may deepen our understanding of the neural mechanisms underpinning pain empathy and suggest that menstrual pain may affect pain empathy through maladaptive dynamic interaction between brain networks.

Continuous reach-to-grasp motion recognition based on an extreme learning machine algorithm using sEMG signals.

Recognizing user intention in reach-to-grasp motions is a critical challenge in rehabilitation engineering. To address this, a Machine Learning (ML) algorithm based on the Extreme Learning Machine (ELM) was developed for identifying motor actions using surface Electromyography (sEMG) during continuous reach-to-grasp movements, involving multiple Degrees of Freedom (DoFs). This study explores feature extraction methods based on time domain and autoregressive models to evaluate ELM performance under different conditions. The experimental setup encompassed variations in neuron size, time windows, validation with each muscle, increase in the number of features, comparison with five conventional ML-based classifiers, inter-subjects variability, and temporal dynamic response. To evaluate the efficacy of the proposed ELM-based method, an openly available sEMG dataset containing data from 12 participants was used. Results highlight the method's performance, achieving Accuracy above 85%, F-score above 90%, Recall above 85%, Area Under the Curve of approximately 84% and compilation times (computational cost) of less than 1 ms. These metrics significantly outperform standard methods (p < 0.05). Additionally, specific trends were found in increasing and decreasing performance in identifying specific tasks, as well as variations in the continuous transitions in the temporal dynamics response. Thus, the ELM-based method effectively identifies continuous reach-to-grasp motions through myoelectric data. These findings hold promise for practical applications. The method's success prompts future research into implementing it for more reliable and effective Human-Machine Interface (HMI) control. This can revolutionize real-time upper limb rehabilitation, enabling natural and complex Activities of Daily Living (ADLs) like object manipulation. The robust results encourages further research and innovative solutions to improve people's quality of life through more effective interventions.

Socio-economic differences in remote schoolwork during the COVID-19 pandemic: a trend analysis of the 2020 and 2021 school-closure periods using the UK Understanding Society data.

In the United Kingdom, the COVID-19 pandemic in 2020 and 2021 led to two extended periods of school closures. Research on inequality of learning opportunity as a result of these closures used a single indicator of socio-economic status, neglecting important determinants of remote learning. Using data from the Understanding Society (USoc) COVID-19 surveys we analysed the levels and differentials in the uptake of remote schoolwork using parental social class, information technology (IT) availability in the home and parental working patterns to capture the distinct resources that families needed to complete remote schoolwork. This is also the first study to assess the extent to which the differentials between socio-economic groups changed between the first and second school-closure periods caused by the pandemic. We found that each of the three factors showed an independent association with the volume of remote schoolwork and that their effect was magnified by their combination. Children in families where the main parent was in an upper-class occupation, where both parents worked from home and where the children had their own IT spent more time doing remote schoolwork than other groups, particularly compared to children of single parents who work from home, children in families where the main parent was in a working-class occupation, where the child had to share IT, and where the parents did not work regularly from home. The differentials between socio-economic groups in the uptake of schoolwork were found to be stable between the two school-closure periods.

Nursing students' engagement in online learning.

Reliance on digital technology may have implications for our social and economic wellbeing, including factors such as health, environmental quality, social interaction, and educational levels. Although there may be concerns, it is important to acknowledge that digital technology also offers immediate, cost-effective and accessible solutions that are transforming various services. The COVID-19 pandemic, through the disruption of educational systems worldwide, has accelerated the transformation of higher education, leading to changes in the way it is perceived. However, there is a lack of understanding regarding the relationship between digital poverty, digital literacy, and students' online experiences. This article aims to explore the engagement of nursing students in online learning post COVID.

In Vivo Exposure Pathways of Ambient Magnetite Nanoparticles Revealed by Machine Learning-Aided Single-Particle Mass Spectrometry.

Nanosized ultrafine particles (UFPs) from natural and anthropogenic sources are widespread and pose serious health risks when inhaled by humans. However, tracing the inhaled UFPs in vivo is extremely difficult, and the distribution, translocation, and metabolism of UFPs remain unclear. Here, we report a label-free, machine learning-aided single-particle inductively coupled plasma mass spectrometry (spICP-MS) approach for tracing the exposure pathways of airborne magnetite nanoparticles (MNPs), including external emission sources, and distribution and translocation in vivo using a mouse model. Our results provide quantitative analysis of different metabolic pathways in mice exposed to MNPs, revealing that the spleen serves as the primary site for MNP metabolism (84.4%), followed by the liver (11.4%). The translocation of inhaled UFPs across different organs alters their particle size. This work provides novel insights into the in vivo fate of UFPs as well as a versatile and powerful platform for nanotoxicology and risk assessment.

Leveraging technology to improve access to medical education in anatomy and physiology in resource-limited settings: A mixed methods study.

This study assesses the acceptability, appropriateness, feasibility, and efficacy of a novel asynchronous video-based intervention for teaching respiratory physiology and anatomy to medical students in resource-limited settings. A series of short video lectures on pleural anatomy, pulmonary physiology, and pathophysiology was created using Lightboard and screen capture technology. These were uploaded to YouTube and Google Drive and made available to 1st-3rd year medical students at two Latin American universities for 1 week. Employing a parallel-convergent mixed methods design, we conducted surveys, focus groups, interviews, and pre/post testing for qualitative and quantitative data. Thematic Analysis was used to analyze qualitative data and McNemar's test for quantitative analysis. Seventy-six students participated. The videos' short format, interactivity, and Lightboard style were highly valued for their flexibility, time efficiency, and educational impact. Students recognized their clinical relevance and trusted their content, suggesting potential applicability in similar settings. Despite infrastructure and connectivity challenges, the use of flexible streaming and downloadable options facilitated learning. Survey results indicated high levels of feasibility (99%), appropriateness (95%), and acceptability (95%), with significant knowledge gains observed (37% correct pre-test answers vs. 56% post-test, p < 0.0001). Our findings demonstrate high acceptability, appropriateness, feasibility, and efficacy of a targeted asynchronous education centered on short-format videos in resource-limited settings, enabling robust learning despite local barriers. Flexible access is key for overcoming localized barriers. Taking an adaptive, learner-centered approach to content creation and delivery to address constraints was pivotal to success. Our modular videos could serve as versatile models for flexible education in resource-constrained settings.

Enhancing Spatial Resolution in Tandem Mass Spectrometry Ion/Ion Reaction Imaging Experiments through Image Fusion.

We have recently developed a charge inversion ion/ion reaction to selectively derivatize phosphatidylserine lipids via gas-phase Schiff base formation. This tandem mass spectrometry (MS/MS) workflow enables the separation and detection of isobaric lipids in imaging mass spectrometry, but the images acquired using this workflow are limited to relatively poor spatial resolutions due to the current time and limit of detection requirements for these ion/ion reaction imaging mass spectrometry experiments. This trade-off between chemical specificity and spatial resolution can be overcome by using computational image fusion, which combines complementary information from multiple images. Herein, we demonstrate a proof-of-concept workflow that fuses a low spatial resolution (i.e., 125 µm) ion/ion reaction product ion image with higher spatial resolution (i.e., 25 µm) ion images from a full scan experiment performed using the same tissue section, which results in a predicted ion/ion reaction product ion image with a 5-fold improvement in spatial resolution. Linear regression, random forest regression, and two-dimensional convolutional neural network (2-D CNN) predictive models were tested for this workflow. Linear regression and 2D CNN models proved optimal for predicted ion/ion images of PS 40:6 and SHexCer d38:1, respectively.

A unified framework to control estimation error in reinforcement learning.

In reinforcement learning, accurate estimation of the Q-value is crucial for acquiring an optimal policy. However, current successful Actor-Critic methods still suffer from underestimation bias. Additionally, there exists a significant estimation bias, regardless of the method used in the critic initialization phase. To address these challenges and reduce estimation errors, we propose CEILING, a simple and compatible framework that can be applied to any model-free Actor-Critic methods. The core idea of CEILING is to evaluate the superiority of different estimation methods by incorporating the true Q-value, calculated using Monte Carlo, during the training process. CEILING consists of two implementations: the Direct Picking Operation and the Exponential Softmax Weighting Operation. The first implementation selects the optimal method at each fixed step and applies it in subsequent interactions until the next selection. The other implementation utilizes a nonlinear weighting function that dynamically assigns larger weights to more accurate methods. Theoretically, we demonstrate that our methods provide a more accurate and stable Q-value estimation. Additionally, we analyze the upper bound of the estimation bias. Based on two implementations, we propose specific algorithms and their variants, and our methods achieve superior performance on several benchmark tasks.

Graph Aggregating-Repelling Network: Do Not Trust All Neighbors in Heterophilic Graphs.

Graph neural networks (GNNs) have demonstrated exceptional performance in processing various types of graph data, such as citation networks and social networks, etc. Although many of these GNNs prove their superiority in handling homophilic graphs, they often overlook the other kind of widespread heterophilic graphs, in which adjacent nodes tend to have different classes or dissimilar features. Recent methods attempt to address heterophilic graphs from the graph spatial domain, which try to aggregate more similar nodes or prevent dissimilar nodes with negative weights. However, they may neglect valuable heterophilic information or extract heterophilic information ineffectively, which could cause poor performance of downstream tasks on heterophilic graphs, including node classification and graph classification, etc. Hence, a novel framework named GARN is proposed to effectively extract both homophilic and heterophilic information. First, we analyze the shortcomings of most GNNs in tackling heterophilic graphs from the perspective of graph spectral and spatial theory. Then, motivated by these analyses, a Graph Aggregating-Repelling Convolution (GARC) mechanism is designed with the objective of fusing both low-pass and high-pass graph filters. Technically, it learns positive attention weights as a low-pass filter to aggregate similar adjacent nodes, and learns negative attention weights as a high-pass filter to repel dissimilar adjacent nodes. A learnable integration weight is used to adaptively fuse these two filters and balance the proportion of the learned positive and negative weights, which could control our GARC to evolve into different types of graph filters and prevent it from over-relying on high intra-class similarity. Finally, a framework named GARN is established by simply stacking several layers of GARC to evaluate its graph representation learning ability on both the node classification and image-converted graph classification tasks. Extensive experiments conducted on multiple homophilic and heterophilic graphs and complex real-world image-converted graphs indicate the effectiveness of our proposed framework and mechanism over several representative GNN baselines.

MIST: Multi-instance selective transformer for histopathological subtype prediction.

Accurate histopathological subtype prediction is clinically significant for cancer diagnosis and tumor microenvironment analysis. However, achieving accurate histopathological subtype prediction is a challenging task due to (1) instance-level discrimination of histopathological images, (2) low inter-class and large intra-class variances among histopathological images in their shape and chromatin texture, and (3) heterogeneous feature distribution over different images. In this paper, we formulate subtype prediction as fine-grained representation learning and propose a novel multi-instance selective transformer (MIST) framework, effectively achieving accurate histopathological subtype prediction. The proposed MIST designs an effective selective self-attention mechanism with multi-instance learning (MIL) and vision transformer (ViT) to adaptive identify informative instances for fine-grained representation. Innovatively, the MIST entrusts each instance with different contributions to the bag representation based on its interactions with instances and bags. Specifically, a SiT module with selective multi-head self-attention (S-MSA) is well-designed to identify the representative instances by modeling the instance-to-instance interactions. On the contrary, a MIFD module with the information bottleneck is proposed to learn the discriminative fine-grained representation for histopathological images by modeling instance-to-bag interactions with the selected instances. Substantial experiments on five clinical benchmarks demonstrate that the MIST achieves accurate histopathological subtype prediction and obtains state-of-the-art performance with an accuracy of 0.936. The MIST shows great potential to handle fine-grained medical image analysis, such as histopathological subtype prediction in clinical applications.

Validating the AI-assisted second language (L2) learning attitude scale for Chinese college students and its correlation with L2 proficiency.

The positive impact of Artificial Intelligence (AI) on second language (L2) learning is well-documented. An individual's attitude toward AI significantly influences its adoption. Despite this, no specific scale has been designed to measure this attitude, particularly in the Chinese context. To address this gap, our study aims to construct the AI-Assisted L2 Learning Attitude Scale for Chinese College Students (AL2AS-CCS) and evaluate its reliability, validity, and relationship with L2 proficiency. Our research comprises two phases, each involving separate samples. In Phase One (Sample 1: n = 379), we conducted exploratory factor analysis (EFA) to determine the factor structure of the AL2AS-CCS. The resulting two-factor structure consists of 12 items, categorized into cognitive and behavioral components. In Phase Two (Sample 2: n = 429), we performed confirmatory factor analysis (CFA) to validate the factor structure and assess model fit. CFA in Sample 2 confirmed the factor structure and demonstrated a good model fit. Additionally, the AL2AS-CCS exhibited high criterion validity, internal consistency, and cross-gender invariance. Our findings suggest that the AL2AS-CCS is a valid measurement tool for assessing Chinese college students' attitude toward AI-assisted L2 learning. Moreover, Chinese college students were discovered to maintain a moderately positive attitude toward AI-assisted L2 learning. Additionally, a positive correlation was identified between this attitude and their L2 proficiency.

Pharmacists and pharmacy students' perceptions on how a new teaching model supports their clinical decision-making.

BACKGROUND AND PURPOSE: Clinical decision-making (CDM) is crucial in pharmacy practice, necessitating effective teaching in undergraduate and postgraduate pharmacy education. This study aims to explore undergraduates and postgraduates' perceptions of how a new teaching model supports their CDM when addressing patient cases. EDUCATIONAL ACTIVITY AND SETTING: Implemented in a full-day CDM course for pharmacy students and a half-day course for pharmacists in the Netherlands, the model, accompanied by a learning guide, facilitated CDM in patient cases. Eight courses were conducted between September 2022 to June 2023, followed by an online survey measuring participants' agreement on how the model supported their CDM, using a 5-point Likert scale. Additionally, three open-ended questions were included to elicit learning outcomes and self-development opportunities. FINDINGS: Of 175 invited participants, 159 (91%) completed the survey. Most agreed the teaching model supported their CDM, particularly in considering the patient's healthcare needs and context (96%), and exploring all available options (96%). Participants found the model provided a clear structure (97%), and fostered critical thinking (93%). The most frequently mentioned learning outcomes and self-development opportunities included collecting sufficient relevant information, maintaining a broad perspective, and decelerating the process to avoid premature closure. SUMMARY: Participants agreed that the teaching model helped them to make clinical decisions. Both undergraduate and postgraduate pharmacy education could possibly benefit from the teaching model's implementation in supporting pharmacy students and pharmacists conducting CDM in pharmacy practice.

Machine learning assisted designing of conjugated organic chromophores, light absorption, and emission behavior prediction.

Organic materials have several important characteristics that make them suitable for use in optoelectronics and optical signal processing applications. For absorption and emission maxima, the stabilities and photoactivities of conjugated organic chromophores can be tailored by selecting a suitable parent structure and incorporating substituents that predictably change the optical characteristics. However, a high-throughput design of efficient conjugated organic chromophores without using trial-and-error experimental approaches is required. In this study, machine learning (ML) is used to design and test the conjugated organic chromophores and predict light absorption and emission behavior. Many machine learning models are tried to select the best models for the prediction of absorption and emission maxima. Extreme gradient boosting regressor has appeared as the best model for the prediction of absorption maxima. Random forest regressor stands out as the best model for the prediction of emission maxima. Breaking Retrosynthetically Interesting Chemical Substructures (BRICS) is used to generate 10,000 organic chromophores. Chemical similarity analysis is performed to obtain a deeper understanding of the characteristics and actions of compounds. Furthermore, clustering and heatmap approaches are utilized.

A novel experiential work-based learning model in paediatric secondary care using entrustable professional activities to develop clinical knowledge and communication skills.

BACKGROUND: Initial education and training standards for pharmacists in Great Britain require early clinical exposure to patients using experiential work-based learning. However, there is poor evidence of this approach in some settings, such as paediatric care. The aim of this study was therefore to explore a novel model of experiential work-based learning for student pharmacists in a paediatric setting. METHODS: Fourth-year student pharmacists enrolled on a Master of Pharmacy programme were allocated five three-hour placement sessions at a paediatric hospital. Sessions consisted of a briefing, ward activities, scaffolded consultations with children and their carers, followed by a debriefing session with a clinical supervisor. Data were collected relating to the ward, patient details, student reported activities, learning outcomes and if follow up was required by a member of the clinical team. Data were cleaned, quality checked, then descriptive statistical analysis and inductive content analysis were conducted. MAIN FINDINGS: Seventy-four students took part in 28 individual sessions and 233 consultations were recorded. Consultations included a best-possible medical history (76%, n = 177), a satisfactory drug history (45%, n = 104), or discussed hospital discharge (11%, n = 26). Students were exposed to patients with diagnosed acute conditions (41%, n = 96) and chronic conditions (33%, n = 76), as well as children awaiting diagnosis (13%, n = 30). Students reported learning about the pathology, diagnosis and symptoms of paediatric conditions (48%, n = 81), medicines used in children (24%, n = 41), patient experiences of recieving care (15%, n = 25), carer experiences (2%, n = 3), the hospital environment (2%, n = 4), career progression (2%, n = 4), and experiences of social care (11%, n = 18). Findings were synthesised with existing entrustable professional activities from the literature to generate novel EPAs specific to paediatric settings. CONCLUSIONS: A paediatric setting offers a suitable environment to host experiential work-based learning in pharmacy education. Standards of initial education and training which require pharmacists to prescribe in Great Britain must recognise the importance of exposure to the health needs and experiences of children, young people's and carers prior to graduation.

Revisiting Perdeck's massive avian migration experiments debunks alternative social interpretations.

Whether avian migrants can adapt to their changing world depends on the relative importance of genetic and environmental variation for the timing and direction of migration. In the classic series of field experiments on avian migration, A. C. Perdeck discovered that translocated juveniles failed to reach goal areas, whereas translocated adults performed 'true-goal navigation'. His translocations of > 14 000 common starlings (Sturnus vulgaris) suggested that genetic mechanisms guide juveniles into a population-specific direction, i.e. 'vector navigation'. However, alternative explanations involving social learning after release in juveniles could not be excluded. By adding historical data from translocation sites, data that was unavailable in Perdeck's days, and by integrated analyses including the original data, we could not explain juvenile migrations from possible social information upon release. Despite their highly social behaviour, our findings are consistent with the idea that juvenile starlings follow inherited information and independently reach their winter quarters. Similar to more solitarily migrating songbirds, starlings would require genetic change to adjust the migration route in response to global change.

Ovarian cancer identification technology based on deep learning and second harmonic generation imaging.

Ovarian cancer is among the most common gynecological cancers and the eighth leading cause of cancer-related deaths among women worldwide. Surgery is among the most important options for cancer treatment. During surgery, a biopsy is generally required to screen for lesions; however, traditional case examinations are time consuming and laborious and require extensive experience and knowledge from pathologists. Therefore, this study proposes a simple, fast, and label-free ovarian cancer diagnosis method that combines second harmonic generation (SHG) imaging and deep learning. Unstained fresh human ovarian tissues were subjected to SHG imaging and accurately characterized using the Pyramid Vision Transformer V2 (PVTv2) model. The results showed that the SHG imaged collagen fibers could quantify ovarian cancer. In addition, the PVTv2 model could accurately differentiate the 3240 SHG images obtained from our imaging collection into benign, normal, and malignant images, with a final accuracy of 98.4%. These results demonstrate the great potential of SHG imaging techniques combined with deep learning models for diagnosing the diseased ovarian tissues.

DeepSAP: A Novel Brain Image-Based Deep Learning Model for Predicting Stroke-Associated Pneumonia From Spontaneous Intracerebral Hemorrhage.

RATIONALE AND OBJECTIVE: Stroke-associated pneumonia (SAP) often appears as a complication following intracerebral hemorrhage (ICH), leading to poor prognosis and increased mortality rates. Previous studies have typically developed prediction models based on clinical data alone, without considering that ICH patients often undergo CT scans immediately upon admission. As a result, these models are subjective and lack real-time applicability, with low accuracy that does not meet clinical needs. Therefore, there is an urgent need for a quick and reliable model to timely predict SAP. METHODS: In this retrospective study, we developed an image-based model (DeepSAP) using brain CT scans from 244 ICH patients to classify the presence and severity of SAP. First, DeepSAP employs MRI-template-based image registration technology to eliminate structural differences between samples, achieving statistical quantification and spatial standardization of cerebral hemorrhage. Subsequently, the processed images and filtered clinical data were simultaneously input into a deep-learning neural network for training and analysis. The model was tested on a test set to evaluate diagnostic performance, including accuracy, specificity, and sensitivity. RESULTS: Brain CT scans from 244 ICH patients (mean age, 60.24; 66 female) were divided into a training set (n = 170) and a test set (n = 74). The cohort included 143 SAP patients, accounting for 58.6% of the total, with 66 cases classified as moderate or above, representing 27% of the total. Experimental results showed an AUC of 0.93, an accuracy of 0.84, a sensitivity of 0.79, and a precision of 0.95 for classifying the presence of SAP. In comparison, the model relying solely on clinical data showed an AUC of only 0.76, while the radiomics method had an AUC of 0.74. Additionally, DeepSAP achieved an optimal AUC of 0.84 for the SAP grading task. CONCLUSION: DeepSAP's accuracy in predicting SAP stems from its spatial normalization and statistical quantification of the ICH region. DeepSAP is expected to be an effective tool for predicting and grading SAP in clinic.

Prediction of Anastomotic Leakage in Esophageal Cancer Surgery: A Multimodal Machine Learning Model Integrating Imaging and Clinical Data.

RATIONALE AND OBJECTIVES: Surgery in combination with chemo/radiotherapy is the standard treatment for locally advanced esophageal cancer. Even after the introduction of minimally invasive techniques, esophagectomy carries significant morbidity and mortality. One of the most common and feared complications of esophagectomy is anastomotic leakage (AL). Our work aimed to develop a multimodal machine-learning model combining CT-derived and clinical data for predicting AL following esophagectomy for esophageal cancer. MATERIAL AND METHODS: A total of 471 patients were prospectively included (Jan 2010-Dec 2022). Preoperative computed tomography (CT) was used to evaluate celia trunk stenosis and vessel calcification. Clinical variables, including demographics, disease stage, operation details, postoperative CRP, and stage, were combined with CT data to build a model for AL prediction. Data was split into 80%:20% for training and testing, and an XGBoost model was developed with 10-fold cross-validation and early stopping. ROC curves and respective areas under the curve (AUC), sensitivity, specificity, PPV, NPV, and F1-scores were calculated. RESULTS: A total of 117 patients (24.8%) exhibited post-operative AL. The XGboost model achieved an AUC of 79.2% (95%CI 69%-89.4%) with a specificity of 77.46%, a sensitivity of 65.22%, PPV of 48.39%, NPV of 87.3%, and F1-score of 56%. Shapley Additive exPlanation analysis showed the effect of individual variables on the result of the model. Decision curve analysis showed that the model was particularly beneficial for threshold probabilities between 15% and 48%. CONCLUSION: A clinically relevant multimodal model can predict AL, which is especially valuable in cases with low clinical probability of AL.