Learning from Imbalanced Data

A Predictive Model for Perinatal Brain Injury Using Machine Learning Based on Early Birth Data

BACKGROUND/OBJECTIVE

It is difficult to predict perinatal brain injury, and performing brain magnetic resonance imaging (MRI) based on suspected injury remains a clinical challenge. Therefore, we aimed to develop a reliable method for predicting perinatal brain injury using a machine learning model with early birth data.

METHODS

Neonates admitted to our institution from January 2017 to June 2024 with a gestational age of ≥36 weeks, a birth weight of ≥1800 g, admission within 6 h of birth, and who underwent brain MRI to confirm perinatal brain injury were included. Various machine learning models, including gradient boosting, were trained using early birth data to predict perinatal brain injury. Synthetic minority over-sampling and adaptive synthetic sampling (ADASYN) were applied to address class imbalance. Model performance was evaluated using accuracy, F1 score, and ROC curves. Feature importance scores and Shapley additive explanations (SHAP) values were also calculated.

RESULTS

Among 179 neonates, 39 had perinatal brain injury. There were significant differences between the injury and non-injury groups in mode of delivery, Apgar scores, capillary pH, lactate dehydrogenase (LDH) levels, and whether therapeutic hypothermia was performed. The gradient boosting model with the ADASYN method achieved the best performance. In terms of feature importance scores, the 1 min Apgar score was the most influential predictor. Additionally, SHAP analysis showed that LDH levels had the highest SHAP values.

CONCLUSION

the gradient boosting model with ADASYN oversampling effectively predicts perinatal brain injury, potentially improving early detection for predicting long-term outcomes, reducing unnecessary MRI scans, and lowering healthcare costs.

Methods to optimize optical sensing of biotic plant stress - combined effects of hyperspectral imaging at night and spatial binning

In spatio-temporal plant monitoring, optical sensing (including hyperspectral imaging), is being deployed to, non-invasively, detect and diagnose plant responses to abiotic and biotic stressors. Early and accurate detection and diagnosis of stressors are key objectives. Level of radiometric repeatability of optical sensing data and ability to accurately detect and diagnose biotic stress are inversely correlated. Accordingly, it may be argued that one of the most significant frontiers and challenges regarding widespread adoption of optical sensing in plant research and crop production hinges on methods to maximize radiometric repeatability. In this study, we acquired hyperspectral optical sensing data at noon and midnight from soybean (Glycine max) and coleus wizard velvet red (Solenostemon scutellarioides) plants with/without experimentally infestation of two-spotted spider mites (Tetranychus urticae). We addressed three questions related to optimization of radiometric repeatability: (1) are reflectance-based plant responses affected by time of optical sensing? (2) if so, are plant responses to two-spotted spider mite infestations (biotic stressor) more pronounced at midnight versus at noon? (3) Is detection of biotic stress enhanced by spatial binning (smoothing) of hyperspectral imaging data? Results from this study provide insight into calculations of radiometric repeatability. Results strongly support claims that acquisition of optical sensing data to detect and characterize stress responses by plants to detect biotic stressors should be performed at night. Moreover, the combination of midnight imaging and spatial binning increased classification accuracies with 29% and 31% for soybean and coleus, respectively. Practical implications of these findings are discussed. Study results are relevant to virtually all applications of optical sensing to detect and diagnose abiotic and biotic stress responses by plants in both controlled environments and in outdoor crop production systems.

Machine learning model for predicting the cold-heat pattern in Kampo medicine: a multicenter prospective observational study

Objective

The purpose of this study was to predict the four cold-heat patterns in patients who have the subjective symptoms of the cold-heat pattern described in the International Classification of Diseases Traditional Medicine Conditions - Module 1 by applying a machine learning algorithm.

Methods

Subjects were first-visit Kampo outpatients at six institutions who agreed to participate in this multicenter prospective observational study. The cold pattern model and the heat pattern model were created separately with 148 symptoms, body mass index, blood pressure (systolic and diastolic), age, and sex. Along with a single cold or heat pattern, the tangled heat/cold pattern is defined as being predicted by both cold and heat patterns, while the moderate (heat/cold) pattern is defined as being predicted by neither the cold pattern nor the heat pattern.

Results

We included 622 participants (mean age ±standard deviation, 54.4 ± 16.9; with female 501). The accuracy, macro-recall, precision, and F1-score of a combination of the two prediction models were 96.7%, 93.2%, 85.6%, and 88.5% respectively. The important items were compatible with the definitions of the cold-heat pattern.

Conclusion

We developed a prediction model on cold-heat patterns with data from patients whose subjective cold/heat-related symptoms matched the cold-heat pattern diagnosis by the physician.

A model for predicting academic performance on standardised tests for lagging regions based on machine learning and Shapley additive explanations

Data are becoming more important in education since they allow for the analysis and prediction of future behaviour to improve academic performance and quality at educational institutions. However, academic performance is affected by regions' conditions, such as demographic, psychographic, socioeconomic and behavioural variables, especially in lagging regions. This paper presents a methodology based on applying nine classification algorithms and Shapley values to identify the variables that influence the performance of the Colombian standardised test: the Saber 11 exam. This study is innovative because, unlike others, it applies to lagging regions and combines the use of EDM and Shapley values to predict students' academic performance and analyse the influence of each variable on academic performance. The results show that the algorithms with the best accuracy are Extreme Gradient Boosting Machine, Light Gradient Boosting Machine, and Gradient Boosting Machine. According to the Shapley values, the most influential variables are the socioeconomic level index, gender, region, location of the educational institution, and age. For Colombia, the results showed that male students from urban educational institutions over 18 years have the best academic performance. Moreover, there are differences in educational quality among the lagging regions. Students from Nariño have advantages over ones from other departments. The proposed methodology allows for generating public policies better aligned with the reality of lagging regions and achieving equity in access to education.

Handling the Imbalanced Problem in Agri-Food Data Analysis

Imbalanced data situations exist in most fields of endeavor. The problem has been identified as a major bottleneck in machine learning/data mining and is becoming a serious issue of concern in food processing applications. Inappropriate analysis of agricultural and food processing data was identified as limiting the robustness of predictive models built from agri-food applications. As a result of rare cases occurring infrequently, classification rules that detect small groups are scarce, so samples belonging to small classes are largely misclassified. Most existing machine learning algorithms including the K-means, decision trees, and support vector machines (SVMs) are not optimal in handling imbalanced data. Consequently, models developed from the analysis of such data are very prone to rejection and non-adoptability in real industrial and commercial settings. This paper showcases the reality of the imbalanced data problem in agri-food applications and therefore proposes some state-of-the-art artificial intelligence algorithm approaches for handling the problem using methods including data resampling, one-class learning, ensemble methods, feature selection, and deep learning techniques. This paper further evaluates existing and newer metrics that are well suited for handling imbalanced data. Rightly analyzing imbalanced data from food processing application research works will improve the accuracy of results and model developments. This will consequently enhance the acceptability and adoptability of innovations/inventions.

Fixing imbalanced binary classification: An asymmetric Bayesian learning approach

Most statistical and machine learning models used for binary data modeling and classification assume that the data are balanced. However, this assumption can lead to poor predictive performance and bias in parameter estimation when there is an imbalance in the data due to the threshold election for the binary classification. To address this challenge, several authors suggest using asymmetric link functions in binary regression, instead of the traditional symmetric functions such as logit or probit, aiming to highlight characteristics that would help the classification task. Therefore, this study aims to introduce new classification functions based on the Lomax distribution (and its variations; including power and reverse versions). The proposed Bayesian functions have proven asymmetry and were implemented in a Stan program into the R workflow. Additionally, these functions showed promising results in real-world data applications, outperforming classical link functions in terms of metrics. For instance, in the first example, comparing the reverse power double Lomax (RPDLomax) with the logit link showed that, regardless of the data imbalance, the RPDLomax model assigns effectively lower mean posterior predictive probabilities to failure and higher probabilities to success (21.4% and 63.7%, respectively), unlike Logistic regression, which does not clearly distinguish between the mean posterior predictive probabilities for these two classes (36.0% and 39.5% for failure and success, respectively). That is, the proposed asymmetric Lomax approach is a competitive model for differentiating binary data classification in imbalanced tasks against the Logistic approach.

Feature Selection Enhances Peptide Binding Predictions for TCR-Specific Interactions

T-cell receptors (TCRs) play a critical role in the immune response by recognizing specific ligand peptides presented by major histocompatibility complex (MHC) molecules. Accurate prediction of peptide binding to TCRs is essential for advancing immunotherapy, vaccine design, and understanding mechanisms of autoimmune disorders. This study presents a novel theoretical method that explores the impact of feature selection techniques on enhancing the predictive accuracy of peptide binding models tailored for specific TCRs. To evaluate the universality of our approach across different TCR systems, we utilized a dataset that includes peptide libraries tested against three distinct murine TCRs. A broad range of physicochemical properties, including amino acid composition, dipeptide composition, and tripeptide features, were integrated into the machine learning-based feature selection framework to identify key features contributing to binding affinity. Our analysis reveals that leveraging optimized feature subsets not only simplifies the model complexity but also enhances predictive performance, enabling more precise identification of TCR-peptide interactions. The results of our feature selection method are consistent with findings from hybrid approaches that utilize both sequence and structural data as input as well as experimental data. Our theoretical approach highlights the role of feature selection in peptide-TCR interactions, providing a powerful tool for uncovering the molecular mechanisms of the T-cell response and assisting in the design of more advanced targeted therapeutics.

Risk Factors for Musculoskeletal Disorders in Korean Farmers: Survey on Occupational Diseases in 2020 and 2022

BACKGROUND/OBJECTIVES

This study investigated factors influencing the prevalence of musculoskeletal disorders (MSDs) resulting from agricultural work, utilizing the 2020 and 2022 occupational disease survey data collected by the Rural Development Administration. The combined data from these years indicated a 6.02% prevalence of MSDs, reflecting a significant class imbalance in the binary response variables. This imbalance could lead to classifiers overlooking rare events, potentially inflating accuracy assessments.

METHODS

We evaluated five distinct models to compare their performance using both original and synthetic data and assessing the models' performance based on synthetic data generation. In the multivariate logistic model, we focused on the main effects of the covariates as there were no statistically significant second-order interactions.

RESULTS

Focusing on the random over-sampling examples (ROSE) method, gender, age, and pesticide use were particularly impactful. The odds of experiencing MSDs were 1.29 times higher for females than males. The odds increased with age: 2.66 times higher for those aged 50-59, 4.60 times higher for those aged 60-69, and 7.16 times higher for those aged 70 or older, compared to those under 50. Pesticide use was associated with 1.26 times higher odds of developing MSDs. Among body part usage variables, all except wrists and knees were significant. Farmers who frequently used their necks, arms, and waist showed 1.27, 1.11, and 1.23 times higher odds of developing MSDs, respectively.

CONCLUSIONS

The accuracy of the raw method was high, but the ROSE method outperformed it for precision and F1 score, and both methods showed similar AUC.

Equilibrium Optimization-Based Ensemble CNN Framework for Breast Cancer Multiclass Classification Using Histopathological Image

Background: Breast cancer is one of the most lethal cancers among women. Early detection and proper treatment reduce mortality rates. Histopathological images provide detailed information for diagnosing and staging breast cancer disease. Methods: The BreakHis dataset, which includes histopathological images, is used in this study. Medical images are prone to problems such as different textural backgrounds and overlapping cell structures, unbalanced class distribution, and insufficiently labeled data. In addition to these, the limitations of deep learning models in overfitting and insufficient feature extraction make it extremely difficult to obtain a high-performance model in this dataset. In this study, 20 state-of-the-art models are trained to diagnose eight types of breast cancer using the fine-tuning method. In addition, a comprehensive experimental study was conducted to determine the most successful new model, with 20 different custom models reported. As a result, we propose a novel model called MultiHisNet. Results: The most effective new model, which included a pointwise convolution layer, residual link, channel, and spatial attention module, achieved 94.69% accuracy in multi-class breast cancer classification. An ensemble model was created with the best-performing transfer learning and custom models obtained in the study, and model weights were determined with an Equilibrium Optimizer. The proposed ensemble model achieved 96.71% accuracy in eight-class breast cancer detection. Conclusions: The results show that the proposed model will support pathologists in successfully diagnosing breast cancer.

When an extra rejection class meets out-of-distribution detection in long-tailed image classification

Detecting Out-of-Distribution (OOD) inputs is essential for reliable deep learning in the open world. However, most existing OOD detection methods have been developed based on training sets that exhibit balanced class distributions, making them susceptible when confronted with training sets following a long-tailed distribution. To alleviate this problem, we propose an effective three-branch training framework, which demonstrates the efficacy of incorporating an extra rejection class along with auxiliary outlier training data for effective OOD detection in long-tailed image classification. In our proposed framework, all outlier training samples are assigned the label of the rejection class. We employ an inlier loss, an outlier loss, and a Tail-class prototype induced Supervised Contrastive Loss (TSCL) to train both the in-distribution classifier and OOD detector within one network. During inference, the OOD detector is constructed using the rejection class. Extensive experimental results demonstrate that the superior OOD detection performance of our proposed method in long-tailed image classification. For example, in the more challenging case where CIFAR100-LT is used as in-distribution, our method improves the average AUROC by 1.23% and reduces the average FPR95 by 3.18% compared to the baseline method utilizing Outlier Exposure (OE). Code is available at github.

Comparative analysis of chronic progressive nephropathy (CPN) diagnosis in rat kidneys using an artificial intelligence deep learning model

With the development of artificial intelligence (AI), technologies based on machines and deep learning are being used in many academic fields. In toxicopathology, research is actively underway to analyze whole slide image (WSI)-level images using AI deep-learning models. However, few studies have been conducted on models for diagnosing complex lesions comprising multiple lesions. Therefore, this study used deep learning segmentation models (YOLOv8, Mask R-CNN, and SOLOv2) to identify three representative lesions (tubular basophilia with atrophy, mononuclear cell infiltration, and hyaline casts) of chronic progressive nephropathy of the kidney, a complex lesion observed in a non-clinical test using rats and selected an initial model appropriate for diagnosing complex lesions by analyzing the characteristics of each algorithm. Approximately 2000 images containing three lesions were extracted using 33 WSI of rat kidneys with chronic progressive nephropathy. Among them, 1701 images were divided into first and second rounds of learning. The loss and mAP50 values were measured twice to confirm the performances of the three algorithms. Loss measurements were stopped at an appropriate epoch to prevent overfitting, and the loss value decreased in the second round based on the data learned in the first round. After measuring the accuracy twice, detection using Mask R-CNN showed the highest mAP50 in all lesions among the three models and was considered sufficient as an initial model for diagnosing complex lesions. By contrast, the YOLOv8 and SOLOv2 models showed low accuracy for all three lesions and had difficulty with segmentation tasks. Therefore, this paper proposes a Mask R-CNN as the initial model for segmenting complex lesions. Precise diagnosis is possible if the model can be trained by increasing the input data, thereby providing greater accuracy in diagnosing pathological images.

Reinforcement Learning-Based Adaptive Classification for Medication State Monitoring in Parkinson's Disease

Parkinson's Disease (PD) patients frequently transition between the 'ON' state, where medication is effective, and the 'OFF' state, affecting their quality of life. Monitoring these transitions is vital for personalized therapy. We introduced a framework based on Reinforcement Learning (RL) to detect transitions between medication states by learning from continuous movement data. Unlike traditional approaches that typically identify each state based on static data patterns, our approach focuses on understanding the dynamic patterns of change throughout the transitions, providing a more generalizable medication state monitoring method. We integrated a deep Long Short-Term Memory (LSTM) neural network and three one-class unsupervised classifiers to implement an RL-based adaptive classifier. We tested on two PD datasets: Dataset PD1 with 12 subjects (14-minute average recording) and Dataset PD2 with seven subjects (120-minute average recording). Data from wrist and ankle wearables captured transitions during 2 to 4-hour daily activities. The algorithm demonstrated its effectiveness in detecting medication states, achieving an average weighted F1-score of 82.94% when trained and tested on Dataset PD1. It performed well when trained on Dataset PD1 and tested on Dataset PD2, with a weighted F1-score of 76.67%. It surpassed other models, was resilient to severe PD symptoms, and performed well with imbalanced data. Notably, prior work has not addressed the generalizability from one dataset to another, essential for real-world applications with varied sensors. Our innovative framework revolutionizes PD monitoring, setting the stage for advanced therapeutic methods and greatly enhancing the life quality of PD patients.

Learning using privileged information with logistic regression on acute respiratory distress syndrome detection

The advanced learning paradigm, learning using privileged information (LUPI), leverages information in training that is not present at the time of prediction. In this study, we developed privileged logistic regression (PLR) models under the LUPI paradigm to detect acute respiratory distress syndrome (ARDS), with mechanical ventilation variables or chest x-ray image features employed in the privileged domain and electronic health records in the base domain. In model training, the objective of privileged logistic regression was designed to incorporate data from the privileged domain and encourage knowledge transfer across the privileged and base domains. An asymptotic analysis was also performed, yielding sufficient conditions under which the addition of privileged information increases the rate of convergence in the proposed model. Results for ARDS detection show that PLR models achieve better classification performances than logistic regression models trained solely on the base domain, even when privileged information is partially available. Furthermore, PLR models demonstrate performance on par with or superior to state-of-the-art models under the LUPI paradigm. As the proposed models are effective, easy to interpret, and highly explainable, they are ideal for other clinical applications where privileged information is at least partially available.

Predicting the resolution of hypertension following adrenalectomy in primary aldosteronism: Controversies and unresolved issues a narrative review

BACKGROUND

Hypertension resolution following adrenalectomy in patients with primary aldosteronism (PA) remains a critical clinical challenge. Identifying preoperatively which patients will become normotensive is both a priority and a point of contention. In this narrative review, we explore the controversies and unresolved issues surrounding the prediction of hypertension resolution after adrenalectomy in PA.

METHODS

A comprehensive literature review was conducted, focusing on studies published between 1954 and 2024 that evaluated all studies that discussed predictive models for hypertension resolution post-adrenalectomy in PA patients. Databases searched included MEDLINE®, Ovid Embase, and Web of Science databases.

RESULTS

The review identified several predictors and predictive models of hypertension resolution, including female sex, duration of hypertension, antihypertensive medication, and BMI. However, inconsistencies in study designs and patient populations led to varied conclusions.

CONCLUSIONS

Although certain predictors and predictive models of hypertension resolution post-adrenalectomy in PA patients are supported by evidence, significant controversies and unresolved issues remain. While the current predictive models provide valuable insights, there is a clear need for further research in this area. Future studies should focus on validating and refining these models.

Unmasking biases and navigating pitfalls in the ophthalmic artificial intelligence lifecycle: A narrative review

Over the past 2 decades, exponential growth in data availability, computational power, and newly available modeling techniques has led to an expansion in interest, investment, and research in Artificial Intelligence (AI) applications. Ophthalmology is one of many fields that seek to benefit from AI given the advent of telemedicine screening programs and the use of ancillary imaging. However, before AI can be widely deployed, further work must be done to avoid the pitfalls within the AI lifecycle. This review article breaks down the AI lifecycle into seven steps-data collection; defining the model task; data preprocessing and labeling; model development; model evaluation and validation; deployment; and finally, post-deployment evaluation, monitoring, and system recalibration-and delves into the risks for harm at each step and strategies for mitigating them.

Hybrid Machine Learning Approach to Zero-Inflated Data Improves Accuracy of Dengue Prediction

BACKGROUND

Spatiotemporal dengue forecasting using machine learning (ML) can contribute to the development of prevention and control strategies for impending dengue outbreaks. However, training data for dengue incidence may be inflated with frequent zero values because of the rarity of cases, which lowers the prediction accuracy. This study aimed to understand the influence of spatiotemporal resolutions of data on the accuracy of dengue incidence prediction using ML models, to understand how the influence of spatiotemporal resolution differs between quantitative and qualitative predictions of dengue incidence, and to improve the accuracy of dengue incidence prediction with zero-inflated data.

METHODOLOGY

We predicted dengue incidence at six spatiotemporal resolutions and compared their prediction accuracy. Six ML algorithms were compared: generalized additive models, random forests, conditional inference forest, artificial neural networks, support vector machines and regression, and extreme gradient boosting. Data from 2009 to 2012 were used for training, and data from 2013 were used for model validation with quantitative and qualitative dengue variables. To address the inaccuracy in the quantitative prediction of dengue incidence due to zero-inflated data at fine spatiotemporal scales, we developed a hybrid approach in which the second-stage quantitative prediction is performed only when/where the first-stage qualitative model predicts the occurrence of dengue cases.

PRINCIPAL FINDINGS

At higher resolutions, the dengue incidence data were zero-inflated, which was insufficient for quantitative pattern extraction of relationships between dengue incidence and environmental variables by ML. Qualitative models, used as binary variables, eased the effect of data distribution. Our novel hybrid approach of combining qualitative and quantitative predictions demonstrated high potential for predicting zero-inflated or rare phenomena, such as dengue.

SIGNIFICANCE

Our research contributes valuable insights to the field of spatiotemporal dengue prediction and provides an alternative solution to enhance prediction accuracy in zero-inflated data where hurdle or zero-inflated models cannot be applied.

Enhancing severe hypoglycemia prediction in type 2 diabetes mellitus through multi-view co-training machine learning model for imbalanced dataset

Patients with type 2 diabetes mellitus (T2DM) who have severe hypoglycemia (SH) poses a considerable risk of long-term death, especially among the elderly, demanding urgent medical attention. Accurate prediction of SH remains challenging due to its multifaced nature, contributed from factors such as medications, lifestyle choices, and metabolic measurements. In this study, we propose a systematic approach to improve the robustness and accuracy of SH predictions using machine learning models, guided by clinical feature selection. Our focus is on developing long-term SH prediction models using both semi-supervised learning and supervised learning algorithms. Using the action to control cardiovascular risk in diabetes trial, which includes electronic health records for over 10,000 individuals, we focus on studying adults with T2DM. Our results indicate that the application of a multi-view co-training method, incorporating the random forest algorithm, improves the specificity of SH prediction, while the same setup with Naive Bayes replacing random forest demonstrates better sensitivity. Our framework also provides interpretability of machine learning models by identifying key predictors for hypoglycemia, including fasting plasma glucose, hemoglobin A1c, general diabetes education, and NPH or L insulins. The integration of data routinely available in electronic health records significantly enhances our model's capability to predict SH events, showcasing its potential to transform clinical practice by facilitating early interventions and optimizing patient management. By enhancing prediction accuracy and identifying crucial predictive features, our study contributes to advancing the understanding and management of hypoglycemia in this population.

Predicting diabetes in adults: identifying important features in unbalanced data over a 5-year cohort study using machine learning algorithm

BACKGROUND

Imbalanced datasets pose significant challenges in predictive modeling, leading to biased outcomes and reduced model reliability. This study addresses data imbalance in diabetes prediction using machine learning techniques. Utilizing data from the Fasa Adult Cohort Study (FACS) with a 5-year follow-up of 10,000 participants, we developed predictive models for Type 2 diabetes.

METHODS

We employed various data-level and algorithm-level interventions, including SMOTE, ADASYN, SMOTEENN, Random Over Sampling and KMeansSMOTE, paired with Random Forest, Gradient Boosting, Decision Tree and Multi-Layer Perceptron (MLP) classifier. We evaluated model performance using F1 score, AUC, and G-means-metrics chosen to provide a comprehensive assessment of model accuracy, discrimination ability, and overall balance in performance, particularly in the context of imbalanced datasets.

RESULTS

our study uncovered key factors influencing diabetes risk and evaluated the performance of various machine learning models. Feature importance analysis revealed that the most influential predictors of diabetes differ between males and females. For females, the most important factors are triglyceride (TG), basal metabolic rate (BMR), and total cholesterol (CHOL), whereas for males, the key predictors are body Mass Index (BMI), serum glutamate Oxaloacetate Transaminase (SGOT), and Gamma-Glutamyl (GGT). Across the entire dataset, BMI remains the most important variable, followed by SGOT, BMR, and energy intake. These insights suggest that gender-specific risk profiles should be considered in diabetes prevention and management strategies. In terms of model performance, our results show that ADASYN with MLP classifier achieved an F1 score of 82.17 ± 3.38, AUC of 89.61 ± 2.09, and G-means of 89.15 ± 2.31. SMOTE with MLP followed closely with an F1 score of 79.85 ± 3.91, AUC of 89.7 ± 2.54, and G-means of 89.31 ± 2.78. The SMOTEENN with Random Forest combination achieved an F1 score of 78.27 ± 1.54, AUC of 87.18 ± 1.12, and G-means of 86.47 ± 1.28.

CONCLUSION

These combinations effectively address class imbalance, improving the accuracy and reliability of diabetes predictions. The findings highlight the importance of using appropriate data-balancing techniques in medical data analysis.

igRNA Prediction and Selection AI Models (igRNA-PS) for Bystander-less ABE Base Editing

CRISPR derived base editing techniques tend to edit multiple bases in the targeted region, which impedes precise reversion of disease-associated single nucleotide variations (SNVs). We designed an imperfect gRNA (igRNA) editing strategy to achieve bystander-less single-base editing. To predict the performance and provide ready-to-use igRNAs, we employed a high-throughput method to edit 5000 loci, each with approximate 19 systematically designed ABE igRNAs. Through deep learning of the relationship of editing efficiency, original gRNA sequence and igRNA sequence, AI models were constructed and tested, designated igRNA Prediction and Selection AI models (igRNA-PS). The models have three functions, First, they can identify the major editing site from the bystanders on a gRNA protospacer with a near 90% accuracy. second, a modified single-base editing efficiency (SBE), considering both single-base editing efficiency and product purity, can be predicted for any given igRNAs. Third, for an editing locus, a set of 64 igRNAs derived from a gRNA can be generated, evaluated through igRNA-PS to select for the best performer, and provided to the user. In this work, we overcome one of the most significant obstacles of base editors, and provide a convenient and efficient approach for single-base bystander-less ABE base editing.

High resolution data visualization and machine learning prediction of free chlorine residual in a green building water system

People spend most of their time indoors and are exposed to numerous contaminants in the built environment. Water management plans implemented in buildings are designed to manage the risks of preventable diseases caused by drinking water contaminants such as opportunistic pathogens (e.g., Legionella spp.), metals, and disinfection by-products (DBPs). However, specialized training required to implement water management plans and heterogeneity in building characteristics limit their widespread adoption. Implementation of machine learning and artificial intelligence (ML/AI) models in building water settings presents an opportunity for faster, more widespread use of data-driven water quality management approaches. We demonstrate the utility of Random Forest and Long Short-Term Memory (LSTM) ML models for predicting a key public health parameter, free chlorine residual, as a function of data collected from building water quality sensors (ORP, pH, conductivity, and temperature) as well as WiFi signals as a proxy for building occupancy and water usage in a "green" Leadership in Energy and Environmental Design (LEED) commercial and institutional building. The models successfully predicted free chlorine residual declines below 0.2 ppm, a common minimum reference level for public health protection in drinking water distribution systems. The predictions were valid up to 5 min in advance, and in some cases reasonably accurate up to 24 h in advance, presenting opportunities for proactive water quality management as part of a sense-analyze-decide framework. An online data dashboard for visualizing water quality in the building is presented, with the potential to link these approaches for real-time water quality management.

Evaluating Active Learning Strategies for Automated Classification of Patient Safety Event Reports in Hospitals

Patient safety event (PSE) reports, which document incidents that compromise patient safety, are fundamental for improving healthcare quality. Accurate classification of these reports is crucial for analyzing trends, guiding interventions, and supporting organizational learning. However, this process is labor-intensive due to the high volume and complex taxonomy of reports. Previous work has shown that machine learning (ML) can automate PSE report classification; however, its success depends on large manually-labeled datasets. This study leverages Active Learning (AL) strategies with human expertise to streamline PSE-report labeling. We utilize pool-based AL sampling to selectively query reports for human annotation, developing a robust dataset for training ML classifiers. Our experiments demonstrate that AL significantly outperforms random sampling in accuracy across various text representations, reducing the need for labeled samples by 24% to 69%. Based on these findings, we suggest that incorporating AL strategies into PSE-report labeling can effectively reduce manual workload while maintaining high classification accuracy.

Non-Invasive Tumor Grade Evaluation in Von Hippel-Lindau-Associated Clear Cell Renal Cell Carcinoma: A Magnetic Resonance Imaging-Based Study

BACKGROUND

Pathology grading is an essential step for the treatment and evaluation of the prognosis in patients with clear cell renal cell carcinoma (ccRCC).

PURPOSE

To investigate the utility of texture analysis in evaluating Fuhrman grades of renal tumors in patients with Von Hippel-Lindau (VHL)-associated ccRCC, aiming to improve non-invasive diagnosis and personalized treatment.

STUDY TYPE

Retrospective analysis of a prospectively maintained cohort.

POPULATION

One hundred and thirty-six patients, 84 (61%) males and 52 (39%) females with pathology-proven ccRCC with a mean age of 52.8 ± 12.7 from 2010 to 2023.

FIELD STRENGTH AND SEQUENCES

1.5 and 3 T MRIs. Segmentations were performed on the T1-weighted 3-minute delayed sequence and then registered on pre-contrast, T1-weighted arterial and venous sequences.

ASSESSMENT

A total of 404 lesions, 345 low-grade tumors, and 59 high-grade tumors were segmented using ITK-SNAP on a T1-weighted 3-minute delayed sequence of MRI. Radiomics features were extracted from pre-contrast, T1-weighted arterial, venous, and delayed post-contrast sequences. Preprocessing techniques were employed to address class imbalances. Features were then rescaled to normalize the numeric values. We developed a stacked model combining random forest and XGBoost to assess tumor grades using radiomics signatures.

STATISTICAL TESTS

The model's performance was evaluated using positive predictive value (PPV), sensitivity, F1 score, area under the curve of receiver operating characteristic curve, and Matthews correlation coefficient. Using Monte Carlo technique, the average performance of 100 benchmarks of 85% train and 15% test was reported.

RESULTS

The best model displayed an accuracy of 0.79. For low-grade tumor detection, a sensitivity of 0.79, a PPV of 0.95, and an F1 score of 0.86 were obtained. For high-grade tumor detection, a sensitivity of 0.78, PPV of 0.39, and F1 score of 0.52 were reported.

DATA CONCLUSION

Radiomics analysis shows promise in classifying pathology grades non-invasively for patients with VHL-associated ccRCC, potentially leading to better diagnosis and personalized treatment.

LEVEL OF EVIDENCE

1 TECHNICAL EFFICACY: Stage 2.

Effects of Post-Hospital Arrival Factors on Out-of-Hospital Cardiac Arrest Outcomes During the COVID-19 Pandemic

IMPORTANCE

The relationship between post-hospital arrival factors and out-of-hospital cardiac arrest (OHCA) outcomes remains unclear.

OBJECTIVES

This study assessed the impact of post-hospital arrival factors on OHCA outcomes during the COVID-19 pandemic using a prediction model.

DESIGN, SETTING, AND PARTICIPANTS

In this cohort study, data from the All-Japan Utstein Registry, a nationwide population-based database, between 2015 and 2021 were used. A total of 541,781 patients older than 18 years old who experienced OHCA of cardiac origin were included.

MAIN OUTCOMES AND MEASURES

The primary exposure was trends in COVID-19 cases. The study compared the predicted proportion of favorable neurologic outcomes 1 month after resuscitation with the actual outcomes. Neurologic outcomes were categorized based on the Cerebral Performance Category score (1, good cerebral function; 2, moderate cerebral function).

RESULTS

The prediction model, which had an area under the curve of 0.96, closely matched actual outcomes in 2019. However, a significant discrepancy emerged after the pandemic began in 2020, where outcomes continued to deteriorate as the virus spread, exacerbated by both pre- and post-hospital arrival factors.

CONCLUSIONS AND RELEVANCE

Post-hospital arrival factors were as important as pre-hospital factors in adversely affecting the prognosis of patients following OHCA during the COVID-19 pandemic. The results suggest that the overall response of the healthcare system needs to be improved during infectious disease outbreaks to improve outcomes.

Using machine learning methods to predict all-cause somatic hospitalizations in adults: A systematic review

AIM

In this review, we investigated how Machine Learning (ML) was utilized to predict all-cause somatic hospital admissions and readmissions in adults.

METHODS

We searched eight databases (PubMed, Embase, Web of Science, CINAHL, ProQuest, OpenGrey, WorldCat, and MedNar) from their inception date to October 2023, and included records that predicted all-cause somatic hospital admissions and readmissions of adults using ML methodology. We used the CHARMS checklist for data extraction, PROBAST for bias and applicability assessment, and TRIPOD for reporting quality.

RESULTS

We screened 7,543 studies of which 163 full-text records were read and 116 met the review inclusion criteria. Among these, 45 predicted admission, 70 predicted readmission, and one study predicted both. There was a substantial variety in the types of datasets, algorithms, features, data preprocessing steps, evaluation, and validation methods. The most used types of features were demographics, diagnoses, vital signs, and laboratory tests. Area Under the ROC curve (AUC) was the most used evaluation metric. Models trained using boosting tree-based algorithms often performed better compared to others. ML algorithms commonly outperformed traditional regression techniques. Sixteen studies used Natural language processing (NLP) of clinical notes for prediction, all studies yielded good results. The overall adherence to reporting quality was poor in the review studies. Only five percent of models were implemented in clinical practice. The most frequently inadequately addressed methodological aspects were: providing model interpretations on the individual patient level, full code availability, performing external validation, calibrating models, and handling class imbalance.

CONCLUSION

This review has identified considerable concerns regarding methodological issues and reporting quality in studies investigating ML to predict hospitalizations. To ensure the acceptability of these models in clinical settings, it is crucial to improve the quality of future studies.

Hybridizing mechanistic modeling and deep learning for personalized survival prediction after immune checkpoint inhibitor immunotherapy

We present a study where predictive mechanistic modeling is combined with deep learning methods to predict individual patient survival probabilities under immune checkpoint inhibitor (ICI) immunotherapy. This hybrid approach enables prediction based on both measures that are calculable from mechanistic models of key mechanisms underlying ICI therapy that may not be directly measurable in the clinic and easily measurable quantities or patient characteristics that are not always readily incorporated into predictive mechanistic models. A deep learning time-to-event predictive model trained on a hybrid mechanistic + clinical data set from 93 patients achieved higher per-patient predictive accuracy based on event-time concordance, Brier score, and negative binomial log-likelihood-based criteria than when trained on only mechanistic model-derived values or only clinical data. Feature importance analysis revealed that both clinical and model-derived parameters play prominent roles in increasing prediction accuracy, further supporting the advantage of our hybrid approach.

Customised-sampling approach for pipe failure prediction in water distribution networks

This paper presents a new methodology for addressing imbalanced class data for failure prediction in Water Distribution Networks (WDNs). The proposed methodology relies on existing approaches including under-sampling, over-sampling, and class weighting as primary strategies. These techniques aim to treat the imbalanced datasets by adjusting the representation of minority and majority classes. Under-sampling reduces data in the majority class, over-sampling adds data to the minority class, and class weighting assigns unequal weights based on class counts to balance the influence of each class during machine learning (ML) model training. In this paper, the mentioned approaches were used at levels other than "balance point" to construct pipe failure prediction models for a WDN with highly imbalanced data. F1-score, and AUC-ROC, were selected to evaluate model performance. Results revealed that under-sampling above the balance point yields the highest F1-score, while over-sampling below the balance point achieves optimal results. Employing class weights during training and prediction emphasises the efficacy of lower weights than the balance. Combining under-sampling and over-sampling to the same ratio for both majority and minority classes showed limited improvement. However, a more effective predictive model emerged when over-sampling the minority class and under-sampling the majority class to different ratios, followed by applying class weights to balance data.

Species distribution modeling for disease ecology: A multi-scale case study for schistosomiasis host snails in Brazil

Species distribution models (SDMs) are increasingly popular tools for profiling disease risk in ecology, particularly for infectious diseases of public health importance that include an obligate non-human host in their transmission cycle. SDMs can create high-resolution maps of host distribution across geographical scales, reflecting baseline risk of disease. However, as SDM computational methods have rapidly expanded, there are many outstanding methodological questions. Here we address key questions about SDM application, using schistosomiasis risk in Brazil as a case study. Schistosomiasis is transmitted to humans through contact with the free-living infectious stage of Schistosoma spp. parasites released from freshwater snails, the parasite's obligate intermediate hosts. In this study, we compared snail SDM performance across machine learning (ML) approaches (MaxEnt, Random Forest, and Boosted Regression Trees), geographic extents (national, regional, and state), types of presence data (expert-collected and publicly-available), and snail species (Biomphalaria glabrata, B. straminea, and B. tenagophila). We used high-resolution (1km) climate, hydrology, land-use/land-cover (LULC), and soil property data to describe the snails' ecological niche and evaluated models on multiple criteria. Although all ML approaches produced comparable spatially cross-validated performance metrics, their suitability maps showed major qualitative differences that required validation based on local expert knowledge. Additionally, our findings revealed varying importance of LULC and bioclimatic variables for different snail species at different spatial scales. Finally, we found that models using publicly-available data predicted snail distribution with comparable AUC values to models using expert-collected data. This work serves as an instructional guide to SDM methods that can be applied to a range of vector-borne and zoonotic diseases. In addition, it advances our understanding of the relevant environment and bioclimatic determinants of schistosomiasis risk in Brazil.

Enhancing microalgae classification accuracy in marine ecosystems through convolutional neural networks and support vector machines

Accurately classifying microalgae species is vital for monitoring marine ecosystems and managing the emergence of marine mucilage, which is crucial for monitoring mucilage phenomena in marine environments. Traditional methods have been inadequate due to time-consuming processes and the need for expert knowledge. The purpose of this article is to employ convolutional neural networks (CNNs) and support vector machines (SVMs) to improve classification accuracy and efficiency. By employing advanced computational techniques, including MobileNet and GoogleNet models, alongside SVM classification, the study demonstrates significant advancements over conventional identification methods. In the classification of a dataset consisting of 7820 images using four different SVM kernel functions, the linear kernel achieved the highest success rate at 98.79 %. It is followed by the RBF kernel at 98.73 %, the polynomial kernel at 97.84 %, and the sigmoid kernel at 97.20 %. This research not only provides a methodological framework for future studies in marine biodiversity monitoring but also highlights the potential for real-time applications in ecological conservation and understanding mucilage dynamics amidst climate change and environmental pollution.

Automated Detection of Pediatric Foreign Body Aspiration from Chest X-rays Using Machine Learning

OBJECTIVE/HYPOTHESIS

Standard chest radiographs are a poor diagnostic tool for pediatric foreign body aspiration. Machine learning may improve upon the diagnostic capabilities of chest radiographs. The objective is to develop a machine learning algorithm that improves the diagnostic capabilities of chest radiographs in pediatric foreign body aspiration.

METHOD

This retrospective, diagnostic study included a retrospective chart review of patients with a potential diagnosis of FBA from 2010 to 2020. Frontal view chest radiographs were extracted, processed, and uploaded to Google AutoML Vision. The developed algorithm was then evaluated against a pediatric radiologist.

RESULTS

The study selected 566 patients who were presented with a suspected diagnosis of foreign body aspiration. One thousand six hundred and eighty eight chest radiograph images were collected. The sensitivity and specificity of the radiologist interpretation were 50.6% (43.1-58.0) and 88.7% (85.3-91.5), respectively. The sensitivity and specificity of the algorithm were 66.7% (43.0-85.4) and 95.3% (90.6-98.1), respectively. The precision and recall of the algorithm were both 91.8% with an AuPRC of 98.3%.

CONCLUSION

Chest radiograph analysis augmented with machine learning can diagnose foreign body aspiration in pediatric patients at a level similar to a read performed by a pediatric radiologist despite only using single-view, fixed images. Overall, this study highlights the potential and capabilities of machine learning in diagnosing conditions with a wide range of clinical presentations.

LEVEL OF EVIDENCE

3 Laryngoscope, 134:3807-3814, 2024.

Using machine learning algorithms to detect fear of falling in people with multiple sclerosis in standardized gait analysis

INTRODUCTION

Multiple sclerosis (MS) is the most common chronic inflammatory disease of the central nervous system. The progressive impairment of gait is one of the most important pathognomic symptoms which are associated with falls and fear of falling (FOF) in people with MS (pwMS). 60 % of pwMS show a FOF, which leads to restrictions in mobility as well as physical activity and reduces the quality of life in general. Therefore, early detection of FOF is crucial because it enables early implementation of rehabilitation strategies as well as clinical decision-making to reduce progression. Qualitative and quantitative evaluation of gait pattern is an essential aspect of disease assessment and can provide valuable insights for personalized treatment decisions in pwMS. Our objective was to identify the most appropriate clinical gait analysis methods to identify FOF in pwMS and to detect the optimal machine learning (ML) algorithms to predict FOF using the complex multidimensional data from gait analysis.

METHODS

Data of 1240 pwMS was recorded at the MS Centre of the University Hospital Dresden between November 2020 and September 2021. Patients performed a multidimensional gait analysis with pressure and motion sensors, as well as patient-reported outcomes (PROs), according to a standardized protocol. A feature selection ensemble (FS-Ensemble) was developed to improve the classification performance. The FS-Ensemble consisted of four filtering methods: Chi-square test, information gain, minimum redundancy maximum relevance and ReliefF. Gaussian Naive Bayes, Decision Tree, k-Nearest Neighbor, and Support Vector Machines (SVM) were used to identify FOF.

RESULTS

The descriptive analysis showed that 37 % of the 1240 pwMS had a FOF (n = 458; age: 51 ± 16 years, 76 % women, median EDSS: 4.0). The FS-Ensemble improved classification performance in most cases. The SVM showed the best performance of the four classification models in detecting FOF. The PROs showed the best F1 scores (Early Mobility Impairment Questionnaire F1 = 0.81 ± 0.00 and 12-item Multiple Sclerosis Scale F1 = 0.80 ± 0.00).

CONCLUSION

FOF is an important psychological risk factor associated with an increased risk of falls. To integrate a functional early warning system for fall detection into MS management and progression monitoring, it is necessary to detect the relevant gait parameters as well as assessment methods. In this context, ML strategies allow the integration of gait parameters from clinical routine to support the initiation of early rehabilitation measures and adaptation of course-modifying therapeutics. The results of this study confirm that patients' self-assessments play an important role in disease management.

MediDRNet: Tackling category imbalance in diabetic retinopathy classification with dual-branch learning and prototypical contrastive learning

BACKGROUND AND OBJECTIVE

The classification of diabetic retinopathy (DR) aims to utilize the implicit information in images for early diagnosis, to prevent and mitigate the further worsening of the condition. However, existing methods are often limited by the need to operate within large, annotated datasets to show significant advantages. Additionally, the number of samples for different categories within the dataset needs to be evenly distributed, because the characteristic of sample imbalance distribution can lead to an excessive focus on high-frequency disease categories, while neglecting the less common but equally important disease categories. Therefore, there is an urgent need to develop a new classification method that can effectively alleviate the issue of sample distribution imbalance, thereby enhancing the accuracy of diabetic retinopathy classification.

METHODS

In this work, we propose MediDRNet, a dual-branch network model based on prototypical contrastive learning. This model adopts prototype contrastive learning, creating prototypes for different levels of lesions, ensuring they represent the core features of each lesion level. It classifies by comparing the similarity between data points and their category prototypes. Our dual-branch network structure effectively resolves the issue of category imbalance and improves classification accuracy by emphasizing subtle differences in retinal lesions. Moreover, our approach combines a dual-branch network with specific lesion-level prototypes for core feature representation and incorporates the convolutional block attention module for enhanced lesion feature identification.

RESULTS

Our experiments using both the Kaggle and UWF classification datasets have demonstrated that MediDRNet exhibits exceptional performance compared to other advanced models in the industry, especially on the UWF DR classification dataset where it achieved state-of-the-art performance across all metrics. On the Kaggle DR classification dataset, it achieved the highest average classification accuracy (0.6327) and Macro-F1 score (0.6361). Particularly in the classification tasks for minority categories of diabetic retinopathy on the Kaggle dataset (Grades 1, 2, 3, and 4), the model reached high classification accuracies of 58.08%, 55.32%, 69.73%, and 90.21%, respectively. In the ablation study, the MediDRNet model proved to be more effective in feature extraction from diabetic retinal fundus images compared to other feature extraction methods.

CONCLUSIONS

This study employed prototype contrastive learning and bidirectional branch learning strategies, successfully constructing a grading system for diabetic retinopathy lesions within imbalanced diabetic retinopathy datasets. Through a dual-branch network, the feature learning branch effectively facilitated a smooth transition of features from the grading network to the classification learning branch, accurately identifying minority sample categories. This method not only effectively resolved the issue of sample imbalance but also provided strong support for the precise grading and early diagnosis of diabetic retinopathy in clinical applications, showcasing exceptional performance in handling complex diabetic retinopathy datasets. Moreover, this research significantly improved the efficiency of prevention and management of disease progression in diabetic retinopathy patients within medical practice. We encourage the use and modification of our code, which is publicly accessible on GitHub: https://github.com/ReinforceLove/MediDRNet.

Machine learning allows robust classification of visceral fat in women with obesity using common laboratory metrics

The excessive accumulation and malfunctioning of visceral adipose tissue (VAT) is a major determinant of increased risk of obesity-related comorbidities. Thus, risk stratification of people living with obesity according to their amount of VAT is of clinical interest. Currently, the most common VAT measurement methods include mathematical formulae based on anthropometric dimensions, often biased by human measurement errors, bio-impedance, and image techniques such as X-ray absorptiometry (DXA) analysis, which requires specialized equipment. However, previous studies showed the possibility of classifying people living with obesity according to their VAT through blood chemical concentrations by applying machine learning techniques. In addition, most of the efforts were spent on men living with obesity while little was done for women. Therefore, this study aims to compare the performance of the multilinear regression model (MLR) in estimating VAT and six different supervised machine learning classifiers, including logistic regression (LR), support vector machine and decision tree-based models, to categorize 149 women living with obesity. For clustering, the study population was categorized into classes 0, 1, and 2 according to their VAT and the accuracy of each MLR and classification model was evaluated using DXA-data (DXAdata), blood chemical concentrations (BLDdata), and both DXAdata and BLDdata together (ALLdata). Estimation error and R 2 were computed for MLR, while receiver operating characteristic (ROC) and precision-recall curves (PR) area under the curve (AUC) were used to assess the performance of every classification model. MLR models showed a poor ability to estimate VAT with mean absolute error ≥ 401.40 andR 2 ≤ 0.62 in all the datasets. The highest accuracy was found for LR with values of 0.57, 0.63, and 0.53 for ALLdata, DXAdata, and BLDdata, respectively. The ROC AUC showed a poor ability of both ALLdata and DXAdata to distinguish class 1 from classes 0 and 2 (AUC = 0.31, 0.71, and 0.85, respectively) as also confirmed by PR (AUC = 0.24, 0.57, and 0.73, respectively). However, improved performances were obtained when applying LR model to BLDdata (ROC AUC ≥ 0.61 and PR AUC ≥ 0.42), especially for class 1. These results seem to suggest that, while a direct and reliable estimation of VAT was not possible in our cohort, blood sample-derived information can robustly classify women living with obesity by machine learning-based classifiers, a fact that could benefit the clinical practice, especially in those health centres where medical imaging devices are not available. Nonetheless, these promising findings should be further validated over a larger population.

AI-driven automatic compression system for colorectal anastomosis

Although circular staplers offer technical advancements over traditional hand-sewn techniques, their use remains challenging for unskilled users, necessitating substantial time and experience for mastery. In particular, it is challenging to apply a consistent pressure of an appropriate magnitude. We developed an automated circular anastomosis device using artificial intelligence (AI) to solve this problem. Automation through AI reduces experiential factors during the anastomosis process. We defined damage occurring during the anastomosis process, noting that a greater depth of damage indicated a more severe injury. For automated anastomosis, data at a tissue strain of 40% were used for the AI model, as this strain level showed optimal performance based on the accuracy and cost matrix. We compared the outcomes of automated anastomosis using a trained AI with those of unskilled users. The results were validated using the Shapiro-Wilk test and t tests. Compression damage was verified on collagen sheets. The AI-driven automatic compression system resulted in less damage compared to unskilled users. In particular, a more significant difference in damage was observed in poor-condition collagen than in good-condition collagen. Damage to the collagen under poor conditions was 54.8% when handled by unskilled users, while the AI-driven automatic compression system resulted in 38.9% damage. This study confirmed that novices' use of AI for automated anastomosis reduces the risk of damage, especially for tissues in poor condition.

Transcriptional markers classifying Escherichia coli and Staphylococcus aureus induced sepsis in adults: A data-driven approach

Sepsis is a life-threatening condition mainly caused by gram-negative and gram-positive bacteria. Understanding the type of causative agent in the early stages is essential for precise antibiotic therapy. This study sought to identify a host gene set capable of distinguishing between sepsis induced by gram-negative bacteria; Escherichia coli and gram-positive bacteria; Staphylococcus aureus in community-onset adult patients. In the present study, microarray expression information was used to apply the Least Absolute Shrinkage and Selection Operator (Lasso) technique to select the predictive gene set for classifying sepsis induced by E. coli or S. aureus pathogens. We identified 25 predictive genes, including LILRA5 and TNFAIP6, which had previously been associated with sepsis in other research. Using these genes, we trained a logistic regression classifier to distinguish whether a sample contains an E. coli or S. aureus infection or belongs to a healthy control group, and subsequently assessed its performance. The classifier achieved an Area Under the Curve (AUC) of 0.96 for E. coli and 0.98 for S. aureus-induced sepsis, and perfect discrimination (AUC of 1) for healthy controls from the other conditions in a 10-fold cross-validation. The genes demonstrated an AUC of 0.75 in distinguishing between sepsis patients with E. coli and S. aureus pathogens. These findings were further confirmed in two distinct independent validation datasets which gave high prediction AUC ranging from 0.72-0.87 and 0.62 in distinguishing three groups of participants and two groups of patients respectively. These genes were significantly enriched in the immune system, cytokine signaling in immune system, innate immune system, and interferon signaling. Transcriptional patterns in blood can differentiate patients with E. coli-induced sepsis from those with S. aureus-induced sepsis. These diagnostic markers, upon validation in larger trials, may serve as a foundation for a reliable differential diagnostics assay.

Video-based diagnosis support system for pianists with Musician's dystonia

Background

Musician's dystonia is a task-specific movement disorder that deteriorates fine motor control of skilled movements in musical performance. Although this disorder threatens professional careers, its diagnosis is challenging for clinicians who have no specialized knowledge of musical performance.

Objectives

To support diagnostic evaluation, the present study proposes a novel approach using a machine learning-based algorithm to identify the symptomatic movements of Musician's dystonia.

Methods

We propose an algorithm that identifies the dystonic movements using the anomaly detection method with an autoencoder trained with the hand kinematics of healthy pianists. A unique feature of the algorithm is that it requires only the video image of the hand, which can be derived by a commercially available camera. We also measured the hand biomechanical functions to assess the contribution of peripheral factors and improve the identification of dystonic symptoms.

Results

The proposed algorithm successfully identified Musician's dystonia with an accuracy and specificity of 90% based only on video footages of the hands. In addition, we identified the degradation of biomechanical functions involved in controlling multiple fingers, which is not specific to musical performance. By contrast, there were no dystonia-specific malfunctions of hand biomechanics, including the strength and agility of individual digits.

Conclusion

These findings demonstrate the effectiveness of the present technique in aiding in the accurate diagnosis of Musician's dystonia.

Predicting 30-day unplanned hospital readmission after revision total knee arthroplasty: machine learning model analysis of a national patient cohort

Revision total knee arthroplasty (TKA) is associated with a higher risk of readmission than primary TKA. Identifying individual patients predisposed to readmission can facilitate proactive optimization and increase care efficiency. This study developed machine learning (ML) models to predict unplanned readmission following revision TKA using a national-scale patient dataset. A total of 17,443 revision TKA cases (2013-2020) were acquired from the ACS NSQIP database. Four ML models (artificial neural networks, random forest, histogram-based gradient boosting, and k-nearest neighbor) were developed on relevant patient variables to predict readmission following revision TKA. The length of stay, operation time, body mass index (BMI), and laboratory test results were the strongest predictors of readmission. Histogram-based gradient boosting was the best performer in distinguishing readmission (AUC: 0.95) and estimating the readmission probability for individual patients (calibration slope: 1.13; calibration intercept: -0.00; Brier score: 0.064). All models produced higher net benefit than the default strategies of treating all or no patients, supporting the clinical utility of the models. ML demonstrated excellent performance for the prediction of readmission following revision TKA. Optimization of important predictors highlighted by our model may decrease preventable hospital readmission following surgery, thereby leading to reduced financial burden and improved patient satisfaction.

Multiparametric MRI subregion radiomics for preoperative assessment of high-risk subregions in microsatellite instability of rectal cancer patients: a multicenter study

BACKGROUND

Microsatellite instability (MSI) is associated with treatment response and prognosis in patients with rectal cancer (RC). However, intratumoral heterogeneity limits MSI testing in patients with RC. The authors developed a subregion radiomics model based on multiparametric MRI to preoperatively assess high-risk subregions with MSI and predict the MSI status of patients with RC.

METHODS

This retrospective study included 475 patients (training cohort, 382; external test cohort, 93) with RC from two participating hospitals between April 2017 and June 2023. In the training cohort, subregion radiomic features were extracted from multiparametric MRI, which included T2-weighted, T1-weighted, diffusion-weighted, and contrast-enhanced T1-weighted imaging. MSI-related subregion radiomic features, classical radiomic features, and clinicoradiological variables were gathered to build five predictive models using logistic regression. Kaplan-Meier survival analysis was conducted to explore the prognostic information.

RESULTS

Among the 475 patients [median age, 64 years (interquartile range, IQR: 55-70 years); 304 men and 171 women], the prevalence of MSI was 11.16% (53/475). The subregion radiomics model outperformed the classical radiomics and clinicoradiological models in both training [area under the curve (AUC)=0.86, 0.72, and 0.59, respectively] and external test cohorts (AUC=0.83, 0.73, and 0.62, respectively). The subregion-clinicoradiological model combining clinicoradiological variables and subregion radiomic features performed the optimal, with AUCs of 0.87 and 0.85 in the training and external test cohorts, respectively. The 3-year disease-free survival rate of MSI groups predicted based on the model was higher than that of the predicted microsatellite stability groups in both patient cohorts (training, P =0.032; external test, P =0.046).

CONCLUSIONS

The authors developed and validated a model based on subregion radiomic features of multiparametric MRI to evaluate high-risk subregions with MSI and predict the MSI status of RC preoperatively, which may assist in individualized treatment decisions and positioning for biopsy.

Automatic and robust estimation of sex and chronological age from panoramic radiographs using a multi-task deep learning network: a study on a South Korean population

Sex and chronological age estimation are crucial in forensic investigations and research on individual identification. Although manual methods for sex and age estimation have been proposed, these processes are labor-intensive, time-consuming, and error-prone. The purpose of this study was to estimate sex and chronological age from panoramic radiographs automatically and robustly using a multi-task deep learning network (ForensicNet). ForensicNet consists of a backbone and both sex and age attention branches to learn anatomical context features of sex and chronological age from panoramic radiographs and enables the multi-task estimation of sex and chronological age in an end-to-end manner. To mitigate bias in the data distribution, our dataset was built using 13,200 images with 100 images for each sex and age range of 15-80 years. The ForensicNet with EfficientNet-B3 exhibited superior estimation performance with mean absolute errors of 2.93 ± 2.61 years and a coefficient of determination of 0.957 for chronological age, and achieved accuracy, specificity, and sensitivity values of 0.992, 0.993, and 0.990, respectively, for sex prediction. The network demonstrated that the proposed sex and age attention branches with a convolutional block attention module significantly improved the estimation performance for both sex and chronological age from panoramic radiographs of elderly patients. Consequently, we expect that ForensicNet will contribute to the automatic and accurate estimation of both sex and chronological age from panoramic radiographs.

Assessment of Bone Age Based on Hand Radiographs Using Regression-Based Multi-Modal Deep Learning

(1) Objective: In this study, a regression-based multi-modal deep learning model was developed for use in bone age assessment (BAA) utilizing hand radiographic images and clinical data, including patient gender and chronological age, as input data. (2) Methods: A dataset of hand radiographic images from 2974 pediatric patients was used to develop a regression-based multi-modal BAA model. This model integrates hand radiographs using EfficientNetV2S convolutional neural networks (CNNs) and clinical data (gender and chronological age) processed by a simple deep neural network (DNN). This approach enhances the model's robustness and diagnostic precision, addressing challenges related to imbalanced data distribution and limited sample sizes. (3) Results: The model exhibited good performance on BAA, with an overall mean absolute error (MAE) of 0.410, root mean square error (RMSE) of 0.637, and accuracy of 91.1%. Subgroup analysis revealed higher accuracy in females ≤ 11 years (MAE: 0.267, RMSE: 0.453, accuracy: 95.0%) and >11 years (MAE: 0.402, RMSE: 0.634, accuracy 92.4%) compared to males ≤ 13 years (MAE: 0.665, RMSE: 0.912, accuracy: 79.7%) and >13 years (MAE: 0.647, RMSE: 1.302, accuracy: 84.6%). (4) Conclusion: This model showed a generally good performance on BAA, showing a better performance in female pediatrics compared to male pediatrics and an especially robust performance in female pediatrics ≤ 11 years.

Large-scale foundation models and generative AI for BigData neuroscience

Recent advances in machine learning have led to revolutionary breakthroughs in computer games, image and natural language understanding, and scientific discovery. Foundation models and large-scale language models (LLMs) have recently achieved human-like intelligence thanks to BigData. With the help of self-supervised learning (SSL) and transfer learning, these models may potentially reshape the landscapes of neuroscience research and make a significant impact on the future. Here we present a mini-review on recent advances in foundation models and generative AI models as well as their applications in neuroscience, including natural language and speech, semantic memory, brain-machine interfaces (BMIs), and data augmentation. We argue that this paradigm-shift framework will open new avenues for many neuroscience research directions and discuss the accompanying challenges and opportunities.

The receiver operating characteristic curve accurately assesses imbalanced datasets

Many problems in biology require looking for a "needle in a haystack," corresponding to a binary classification where there are a few positives within a much larger set of negatives, which is referred to as a class imbalance. The receiver operating characteristic (ROC) curve and the associated area under the curve (AUC) have been reported as ill-suited to evaluate prediction performance on imbalanced problems where there is more interest in performance on the positive minority class, while the precision-recall (PR) curve is preferable. We show via simulation and a real case study that this is a misinterpretation of the difference between the ROC and PR spaces, showing that the ROC curve is robust to class imbalance, while the PR curve is highly sensitive to class imbalance. Furthermore, we show that class imbalance cannot be easily disentangled from classifier performance measured via PR-AUC.

Developing and comparing deep learning and machine learning algorithms for osteoporosis risk prediction

Introduction

Osteoporosis, characterized by low bone mineral density (BMD), is an increasingly serious public health issue. So far, several traditional regression models and machine learning (ML) algorithms have been proposed for predicting osteoporosis risk. However, these models have shown relatively low accuracy in clinical implementation. Recently proposed deep learning (DL) approaches, such as deep neural network (DNN), which can discover knowledge from complex hidden interactions, offer a new opportunity to improve predictive performance. In this study, we aimed to assess whether DNN can achieve a better performance in osteoporosis risk prediction.

Methods

By utilizing hip BMD and extensive demographic and routine clinical data of 8,134 subjects with age more than 40 from the Louisiana Osteoporosis Study (LOS), we developed and constructed a novel DNN framework for predicting osteoporosis risk and compared its performance in osteoporosis risk prediction with four conventional ML models, namely random forest (RF), artificial neural network (ANN), k-nearest neighbor (KNN), and support vector machine (SVM), as well as a traditional regression model termed osteoporosis self-assessment tool (OST). Model performance was assessed by area under 'receiver operating curve' (AUC) and accuracy.

Results

By using 16 discriminative variables, we observed that the DNN approach achieved the best predictive performance (AUC = 0.848) in classifying osteoporosis (hip BMD T-score ≤ -1.0) and non-osteoporosis risk (hip BMD T-score > -1.0) subjects, compared to the other approaches. Feature importance analysis showed that the top 10 most important variables identified by the DNN model were weight, age, gender, grip strength, height, beer drinking, diastolic pressure, alcohol drinking, smoke years, and economic level. Furthermore, we performed subsampling analysis to assess the effects of varying number of sample size and variables on the predictive performance of these tested models. Notably, we observed that the DNN model performed equally well (AUC = 0.846) even by utilizing only the top 10 most important variables for osteoporosis risk prediction. Meanwhile, the DNN model can still achieve a high predictive performance (AUC = 0.826) when sample size was reduced to 50% of the original dataset.

Conclusion

In conclusion, we developed a novel DNN model which was considered to be an effective algorithm for early diagnosis and intervention of osteoporosis in the aging population.

Deep learning with mixup augmentation for improved pore detection during additive manufacturing

In additive manufacturing (AM), process defects such as keyhole pores are difficult to anticipate, affecting the quality and integrity of the AM-produced materials. Hence, considerable efforts have aimed to predict these process defects by training machine learning (ML) models using passive measurements such as acoustic emissions. This work considered a dataset in which keyhole pores of a laser powder bed fusion (LPBF) experiment were identified using X-ray radiography and then registered both in space and time to acoustic measurements recorded during the LPBF experiment. Due to AM's intrinsic process controls, where a pore-forming event is relatively rare, the acoustic datasets collected during monitoring include more non-pores than pores. In other words, the dataset for ML model development is imbalanced. Moreover, this imbalanced and sparse data phenomenon remains ubiquitous across many AM monitoring schemes since training data is nontrivial to collect. Hence, we propose a machine learning approach to improve this dataset imbalance and enhance the prediction accuracy of pore-labeled data. Specifically, we investigate how data augmentation helps predict pores and non-pores better. This imbalance is improved using recent advances in data augmentation called Mixup, a weak-supervised learning method. Convolutional neural networks (CNNs) are trained on original and augmented datasets, and an appreciable increase in performance is reported when testing on five different experimental trials. When ML models are trained on original and augmented datasets, they achieve an accuracy of 95% and 99% on test datasets, respectively. We also provide information on how dataset size affects model performance. Lastly, we investigate the optimal Mixup parameters for augmentation in the context of CNN performance.

PerSEveML: a web-based tool to identify persistent biomarker structure for rare events using an integrative machine learning approach

Omics data sets often pose a computational challenge due to their high dimensionality, large size, and non-linear structures. Analyzing these data sets becomes especially daunting in the presence of rare events. Machine learning (ML) methods have gained traction for analyzing rare events, yet there has been limited exploration of bioinformatics tools that integrate ML techniques to comprehend the underlying biology. Expanding upon our previously developed computational framework of an integrative machine learning approach, we introduce PerSEveML, an interactive web-based tool that uses crowd-sourced intelligence to predict rare events and determine feature selection structures. PerSEveML provides a comprehensive overview of the integrative approach through evaluation metrics that help users understand the contribution of individual ML methods to the prediction process. Additionally, PerSEveML calculates entropy and rank scores, which visually organize input features into a persistent structure of selected, unselected, and fluctuating categories that help researchers uncover meaningful hypotheses regarding the underlying biology. We have evaluated PerSEveML on three diverse biologically complex data sets with extremely rare events from small to large scale and have demonstrated its ability to generate valid hypotheses. PerSEveML is available at https://biostats-shinyr.kumc.edu/PerSEveML/ and https://github.com/sreejatadutta/PerSEveML.

Three-layer heterogeneous network based on the integration of CircRNA information for MiRNA-disease association prediction

Increasing research has shown that the abnormal expression of microRNA (miRNA) is associated with many complex diseases. However, biological experiments have many limitations in identifying the potential disease-miRNA associations. Therefore, we developed a computational model of Three-Layer Heterogeneous Network based on the Integration of CircRNA information for MiRNA-Disease Association prediction (TLHNICMDA). In the model, a disease-miRNA-circRNA heterogeneous network is built by known disease-miRNA associations, known miRNA-circRNA interactions, disease similarity, miRNA similarity, and circRNA similarity. Then, the potential disease-miRNA associations are identified by an update algorithm based on the global network. Finally, based on global and local leave-one-out cross validation (LOOCV), the values of AUCs in TLHNICMDA are 0.8795 and 0.7774. Moreover, the mean and standard deviation of AUC in 5-fold cross-validations is 0.8777+/-0.0010. Especially, the two types of case studies illustrated the usefulness of TLHNICMDA in predicting disease-miRNA interactions.

Machine Learning Approach for Cardiovascular Death Prediction among Nonalcoholic Steatohepatitis (NASH) Liver Transplant Recipients

Cardiovascular disease is the leading cause of mortality among nonalcoholic steatohepatitis (NASH) patients who undergo liver transplants. In the present study, machine learning algorithms were used to identify important risk factors for cardiovascular death and to develop a prediction model. The Standard Transplant Analysis and Research data were gathered from the Organ Procurement and Transplantation Network. After cleaning and preprocessing, the dataset comprised 10,871 patients and 92 features. Recursive feature elimination (RFE) and select from model (SFM) were applied to select relevant features from the dataset and avoid overfitting. Multiple machine learning algorithms, including logistic regression, random forest, decision tree, and XGBoost, were used with RFE and SFM. Additionally, prediction models were developed using a support vector machine, Gaussian naïve Bayes, K-nearest neighbors, random forest, and XGBoost algorithms. Finally, SHapley Additive exPlanations (SHAP) were used to increase interpretability. The findings showed that the best feature selection method was RFE with a random forest estimator, and the most critical features were recipient and donor blood type, body mass index, recipient and donor state of residence, serum creatinine, and year of transplantation. Furthermore, among all the outcomes, the XGBoost model had the highest performance, with an accuracy value of 0.6909 and an area under the curve value of 0.86. The findings also revealed a predictive relationship between features and cardiovascular death after liver transplant among NASH patients. These insights may assist clinical decision-makers in devising strategies to prevent cardiovascular complications in post-liver transplant NASH patients.

Enhancing Machine-Learning Prediction of Enzyme Catalytic Temperature Optima through Amino Acid Conservation Analysis

Enzymes play a crucial role in various industrial production and pharmaceutical developments, serving as catalysts for numerous biochemical reactions. Determining the optimal catalytic temperature (Topt) of enzymes is crucial for optimizing reaction conditions, enhancing catalytic efficiency, and accelerating the industrial processes. However, due to the limited availability of experimentally determined Topt data and the insufficient accuracy of existing computational methods in predicting Topt, there is an urgent need for a computational approach to predict the Topt values of enzymes accurately. In this study, using phosphatase (EC 3.1.3.X) as an example, we constructed a machine learning model utilizing amino acid frequency and protein molecular weight information as features and employing the K-nearest neighbors regression algorithm to predict the Topt of enzymes. Usually, when conducting engineering for enzyme thermostability, researchers tend not to modify conserved amino acids. Therefore, we utilized this machine learning model to predict the Topt of phosphatase sequences after removing conserved amino acids. We found that the predictive model's mean coefficient of determination (R2) value increased from 0.599 to 0.755 compared to the model based on the complete sequences. Subsequently, experimental validation on 10 phosphatase enzymes with undetermined optimal catalytic temperatures shows that the predicted values of most phosphatase enzymes based on the sequence without conservative amino acids are closer to the experimental optimal catalytic temperature values. This study lays the foundation for the rapid selection of enzymes suitable for industrial conditions.

"Using network analysis modularity to group health code systems and decrease dimensionality in machine learning models"

Background

Machine learning (ML) prediction models in healthcare and pharmacy-related research face challenges with encoding high-dimensional Healthcare Coding Systems (HCSs) such as ICD, ATC, and DRG codes, given the trade-off between reducing model dimensionality and minimizing information loss.

Objectives

To investigate using Network Analysis modularity as a method to group HCSs to improve encoding in ML models.

Methods

The MIMIC-III dataset was utilized to create a multimorbidity network in which ICD-9 codes are the nodes and the edges are the number of patients sharing the same ICD-9 code pairs. A modularity detection algorithm was applied using different resolution thresholds to generate 6 sets of modules. The impact of four grouping strategies on the performance of predicting 90-day Intensive Care Unit readmissions was assessed. The grouping strategies compared: 1) binary encoding of codes, 2) encoding codes grouped by network modules, 3) grouping codes to the highest level of ICD-9 hierarchy, and 4) grouping using the single-level Clinical Classification Software (CCS). The same methodology was also applied to encode DRG codes but limiting the comparison to a single modularity threshold to binary encoding.The performance was assessed using Logistic Regression, Support Vector Machine with a non-linear kernel, and Gradient Boosting Machines algorithms. Accuracy, Precision, Recall, AUC, and F1-score with 95% confidence intervals were reported.

Results

Models utilized modularity encoding outperformed ungrouped codes binary encoding models. The accuracy improved across all algorithms ranging from 0.736 to 0.78 for the modularity encoding, to 0.727 to 0.779 for binary encoding. AUC, recall, and precision also improved across almost all algorithms. In comparison with other grouping approaches, modularity encoding generally showed slightly higher performance in AUC, ranging from 0.813 to 0.837, and precision, ranging from 0.752 to 0.782.

Conclusions

Modularity encoding enhances the performance of ML models in pharmacy research by effectively reducing dimensionality and retaining necessary information. Across the three algorithms used, models utilizing modularity encoding showed superior or comparable performance to other encoding approaches. Modularity encoding introduces other advantages such as it can be used for both hierarchical and non-hierarchical HCSs, the approach is clinically relevant, and can enhance ML models' clinical interpretation. A Python package has been developed to facilitate the use of the approach for future research.

Enhancing psychiatric rehabilitation outcomes through a multimodal multitask learning model based on BERT and TabNet: An approach for personalized treatment and improved decision-making

Evaluating the rehabilitation status of individuals with serious mental illnesses (SMI) necessitates a comprehensive analysis of multimodal data, including unstructured text records and structured diagnostic data. However, progress in the effective assessment of rehabilitation status remains limited. Our study develops a deep learning model integrating Bidirectional Encoder Representations from Transformers (BERT) and TabNet through a late fusion strategy to enhance rehabilitation prediction, including referral risk, dangerous behaviors, self-awareness, and medication adherence, in patients with SMI. BERT processes unstructured textual data, such as doctor's notes, whereas TabNet manages structured diagnostic information. The model's interpretability function serves to assist healthcare professionals in understanding the model's predictive decisions, improving patient care. Our model exhibited excellent predictive performance for all four tasks, with an accuracy exceeding 0.78 and an area under the curve of 0.70. In addition, a series of tests proved the model's robustness, fairness, and interpretability. This study combines multimodal and multitask learning strategies into a model and applies it to rehabilitation assessment tasks, offering a promising new tool that can be seamlessly integrated with the clinical workflow to support the provision of optimized patient care.

Adversarial pair-wise distribution matching for remote sensing image cross-scene classification

Remarkable achievements have been made in the field of remote sensing cross-scene classification in recent years. However, most methods directly align the entire image features for cross-scene knowledge transfer. They usually ignore the high background complexity and low category consistency of remote sensing images, which can significantly impair the performance of distribution alignment. Besides, shortcomings of the adversarial training paradigm and the inability to guarantee the prediction discriminability and diversity can also hinder cross-scene classification performance. To alleviate the above problems, we propose a novel cross-scene classification framework in a discriminator-free adversarial paradigm, called Adversarial Pair-wise Distribution Matching (APDM), to avoid irrelevant knowledge transfer and enable effective cross-domain modeling. Specifically, we propose the pair-wise cosine discrepancy for both inter-domain and intra-domain prediction measurements to fully leverage the prediction information, which can suppress negative semantic features and implicitly align the cross-scene distributions. Nuclear-norm maximization and minimization are introduced to enhance the target prediction quality and increase the applicability of the source knowledge, respectively. As a general cross-scene framework, APDM can be easily embedded with existing methods to boost the performance. Experimental results and analyses demonstrate that APDM can achieve competitive and effective performance on cross-scene classification tasks.