Classification of radiographic lung pattern based on texture analysis and machine learning

BackgroundTo elucidate the value of texture analysis (TA) in detecting and differentiating myocardial tissue alterations on T2-weighted CMR (cardiovascular magnetic resonance imaging) in patients with cardiac amyloidosis (CA) and hypertrophic cardiomyopathy (HCM).MethodsIn this retrospective study, 100 CA (58.5 ± 10.7 years; 41 (41%) females) and 217 HCM (50.7 ± 14.8 years, 101 (46.5%) females) patients who underwent CMR scans were included. Regions of interest for TA were delineated by two radiologists independently on T2-weighted imaging (T2WI). Stepwise dimension reduction and texture feature selection based on reproducibility, machine learning algorithms, and correlation analyses were performed to select features. Both the CA and HCM groups were randomly divided into a training dataset and a testing dataset (7:3). After the TA model was established in the training set, the diagnostic performance of the model was validated in the testing set and further validated in a subgroup of patients with similar hypertrophy.ResultsThe 7 independent texture features provided, in combination, a diagnostic accuracy of 86.0% (AUC = 0.915; 95% CI 0.879–0.951) in the training dataset and 79.2% (AUC = 0.842; 95% CI 0.759–0.924) in the testing dataset. The differential diagnostic accuracy in the similar hypertrophy subgroup was 82.2% (AUC = 0.864, 95% CI 0.805–0.922). The significance of the difference between the AUCs of the TA model and late gadolinium enhancement (LGE) was verified by Delong’s test (p = 0.898). All seven texture features showed significant differences between CA and HCM (all p < 0.001).ConclusionsOur study demonstrated that texture analysis based on T2-weighted images could feasibly differentiate CA from HCM, even in patients with similar hypertrophy. The selected final texture features could achieve a comparable diagnostic capacity to the quantification of LGE.Trial registration Since this study is a retrospective observational study and no intervention had been involved, trial registration is waived.

Machine learning to identify lymph node metastasis from thyroid cancer in patients undergoing contrast-enhanced CT studies

Machine learning (ML) algorithms can accurately extract quantitative patterns from datasets without requiring prior knowledge, playing an increasingly crucial role in tasks such as information extraction from remotely sensed data. This study employs several ML algorithms, including Back Propagation Network (BPN), Support Vector Machine (SVM), Random Forest (RF), and Extreme Gradient Boosting (XGBoost), to develop global oceanic chlorophyll-a (Chl-a) concentration inversion models for MODIS/Aqua (MODISA). The models utilize the SeaWiFS Bio-optical Archive and Storage System (SeaBASS) dataset, which contains in situ Chl-a concentration measurements paired with MODISA remote sensing reflectance (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\mathrm {rs}}$ </tex-math></inline-formula>) for various water color bands from 2002 to 2017. This study includes the following processing steps. First, the SeaBASS dataset is randomly divided into training dataset and testing dataset. Second the Chl-a concentration inversion models are developed using the training dataset. Third, the testing dataset is utilized to evaluate the developed models. Finally, the inversion experiment is conducted using MODISA data. Experimental results indicate that all ML algorithms achieve higher inversion accuracy compared to the traditional ocean color index (OCI) algorithm. Particularly, the SVR model demonstrates the best performance. Additionally, to address the challenge of multidimensional parameter optimization in ML algorithms, this study introduces the Differential Evolution (DE) algorithm and develops a DE-based SVM model (DE-SVM). Compared to the standard SVM model, the DE-SVM model significantly reduces the mean deviation, mean absolute deviation, and root mean square error of the inverted chlorophyll-a concentration, while increasing the coefficient of determination (R2) from 0.87 to 0.926. These results indicate that the DE-SVM model further improves the inversion accuracy of Chl-a concentration, and holds substantial potential in satellite remote sensing inversion of Chl-a concentration.

SummaryAccurate prediction of the solubility of chemical substances in solvents remains a challenge. The sparsity of high-quality solubility data is recognized as the biggest hurdle in the development of robust data-driven methods for practical use. Nonetheless, the effects of the quality and quantity of data on aqueous solubility predictions have not yet been scrutinized. In this study, the roles of the size and the quality of data sets on the performances of the solubility prediction models are unraveled, and the concepts of actual and observed performances are introduced. In an effort to curtail the gap between actual and observed performances, a quality-oriented data selection method, which evaluates the quality of data and extracts the most accurate part of it through statistical validation, is designed. Applying this method on the largest publicly available solubility database and using a consensus machine learning approach, a top-performing solubility prediction model is achieved.

Machine learning (ML) algorithms seek to extract the most beneficial information out of the raw data. As mentioned in previous chapters, preprocess data preparation actions might be needed in order to make the data set ready to be fed into the ML algorithms. ML algorithms have gotten the attention of researchers all over the world during the last few decades. In traditional algorithms, the analyst had to define the rules, which was not always accessible or easily possible, in order to obtain the output. ML algorithms develop the models or rules based on the training data set which include input and output data points. ML algorithms try to understand more beneficial information regarding the system based on the training data set. The built model can be tested using the verification data set which does not have any overlap with the training sets. If the model acquires the acceptable performance measures, it can be applied for other cases to perform the prediction process. It should be considered that ML algorithms are not supposed to do the magic. Indeed, they are supposed to explore and analyze the training data, which the human brain is not able to handle that, to develop a predictive model in which its performance is acceptable for the verification data set. Predictive models should be generalized which means performing satisfactorily for both training and verifying data sets.

A major objective of this project was to apply GE’s commercial machine learning and data analytics toolsets to large-scale, real-world, anonymized Phasor Measurement Unit (PMU) datasets in order to extract signatures, correlated and/or causal factors, and precursor patterns associated with significant power system phenomena. The project had a particular emphasis on extraction of insights relevant to asset health monitoring, real-time load modeling and cybersecurity monitoring. Additionally, the team was directed to undertake a comprehensive data quality analysis for the provided datasets and encouraged to estimate the ‘machine-learning readiness’ of the datasets by documenting any major obstacles to the application of commercial machine learning algorithms. To accomplish the aforementioned objectives, the project team’s work centered around the identification of key event signatures and application of the identified event signatures for event detection and event classification. The industry-validated, semi-supervised machine learning strategy employed for event signature identification involved several major tasks, including data-preprocessing, generation of an overabundance of features, normal data identification, normality modeling, and event signature identification through a methodical, quantitative ranking of features in order of relevance to each studied event type. Throughout the project, data quality issues and mitigation techniques were investigated. In this report, insights are provided regarding the readiness of the provided synchrophasor datasets for application of machine learning and data analytics. The methodologies employed for this technical strategy are summarized in this report. With regards to data preprocessing and feature generation, the provided Training and Test Datasets were ingested into GE’s big data environment. Subsequently, the team applied bad data cleansing and data imputation scripts, event detection scripts, and application programming interfaces (APIs) to the datasets for convenient data access. The project team completed development and validation of dozens of physics-based, statistics-based and transformation-based feature functions used for the extraction of over 60 synchrophasor features. Using a new parallel feature generation technology developed on this project, over 60 features have been rapidly generated for the full two years’ worth of Training and Test Dataset data associated with both the Eastern and Western interconnects. Even accommodating for temporal down-sampling inherent to the feature extraction procedure, this parallel feature generation activity resulted in a massive feature set with a storage requirement approximately equal to that of the raw training dataset itself. With regards to normal data identification and normality modeling, a normality model was built using the feature data extracted from the Training Dataset and iteratively refined subsequent to incremental adjustments and expansions of the Training Dataset feature data. With respect to event characterization and signature identification, an event signature identification pipeline was developed and used in conjunction with the normality model to identify over 15 event signatures for key event categories within the Training Dataset. The identified event signatures were used to characterize hundreds of key events in terms of relative severity, duration, and location of the event. An investigation was undertaken to identify correlated and causal factors involved in transformer events. A separate investigation into temporal trends in ring-down analysis results was undertaken to determine possible associations between system dynamics and various other factors such as loading, season or year. To validate the identified event signatures, additional work was undertaken to develop signature-based anomaly detection and classification tools suitable for convenient application to the synchrophasor datasets. The anomaly detection and classification tools, suitable for online application, were then applied to the entirety of the Eastern Interconnect Training and Test Datasets. Performance of the event detection and classification tools was evaluated upon receipt of the Test Dataset event logs (i.e., the labels for events contained in the Test Dataset), and promising results were obtained despite several challenges (documented herein) associated with application of supervised or semi-supervised machine learning methods to large-scale, anonymized datasets. Finally, the detection and classification tools were used to detect, classify, and characterize thousands of new events not included in the original event logs provided by the DOE within both the Training and Test Datasets.

Bladder Urothelial Carcinoma: Machine Learning-based Computed Tomography Radiomics for Prediction of Histological Variant

Background/Aims On average, there is a delay of 6.7 years between symptom onset and diagnosis of axial spondyloarthritis (axSpA). Since traditional approaches to improving early axSpA identification have had limited success, predictive automated analyses using patient records may help alleviate the burden on healthcare providers. We report results from a machine learning (ML) algorithm developed with UK electronic health records (EHRs) Clinical Practice Research Datalink (CPRD) data to estimate the probability or likelihood of a patient being diagnosed with axSpA based on prior clinical indicators and patient history. Methods Primary care UK EHR data - CPRD GOLD was used to identify patients with axSpA and healthy controls (HC). Patients aged ≥18 years with first diagnosis date of axSpA within the identification period (01-Jan-2005 to 31-Dec-2018) and fulfilling CPRD research acceptability criteria were included. Data pertaining to clinical presentation, consultation, referral, test, and therapy history were extracted for individual patients prior to diagnosis of axSpA. A total of 5,090 patients with axSpA satisfied the acceptability criteria. HC were randomly sampled to create a subset of one unique HC matched to each patient with axSpA, resulting in 5,089 HC. ML usable features derived from the total population (patients with axSpA and HC) numbered 820. After using a further exclusion criterion for the patients with axSpA and HC who had ≥1 of 820 usable features, the final dataset included 7,813 patients (3,902 with axSpA and 3,911 HC). This combined dataset was randomly split (67:33) into a train (n = 5237) and a test (n = 2576) dataset. A random forest (RF) model was trained on the train dataset. Cross-validation was performed for hyper-parameter tuning of the RF classifier. Once the model was trained, accuracy, precision, and F-1 scores were obtained with the test dataset. Results The RF-based algorithm resulted in a high level of accuracy (88.12%), with precision of 0.95 for patients with axSpA and 0.83 for HC. The RF algorithm identified 89 best clinical predictors (out of 820 used as inputs) that differentiated between patient and HC such as: total number of tests, total number of referrals, first age of consultation, first symptom age, and number of low back pain symptoms. The model sensitivity was 0.75 and positive predictive value was 80.88%. The model specificity was 0.96 and negative predictive value was 82.56%. Conclusion The ML algorithm demonstrated a high level of accuracy and precision in the identification of possible cases of axSpA, which may be useful in reducing the delay in diagnosis. Previous studies have successfully demonstrated automated cohort identification of axSpA in large datasets, with only a few using ML-based approaches for diagnosis from patient medical history. While our model supports previous work in axSpA, it needs further validation in routine clinical practice (exploration ongoing). Disclosure R. Sengupta: Honoraria; AbbVie, Biogen, Celgene, Lilly, MSD, Novartis, Roche, UCB. Grants/research support; AbbVie, Celgene, Novartis, UCB. S. Narasimham: Shareholder/stock ownership; Novartis. Other; Employee of Novartis. B.S. Mato: Shareholder/stock ownership; Novartis. Other; Employee of Novartis. M. Meglic: Other; Employee of Novartis. C. Perella: Other; Employee of Novartis. P. Pamies: Other; Employee of Novartis. P. Emery: Consultancies; AbbVie, Astra-Zeneca, BMS, Boehringer Ingelheim, Celltrion, Gilead, Janssen, MSD, Lilly, Novartis, Pfizer, Roche, Samsung, UCB. Grants/research support; AbbVie, BMS, Lilly, Novartis, Pfizer, Roche, Samsung.

The role of pertinently diversified and balanced training as well as testing data sets in achieving the true performance of classifiers in predicting the antifreeze proteins

Due to high costs, resources and managemant associated with readmission into Intensive Care Units (ICU), it has been a center of clinical research. Previous research successfully identified several common risk factors and proposed a variety of frameworks to predict ICU readmissions, whereas, some studies reported that many risk factors were too specific and/or had limited focus. This study aims to investigate and analyze if the relevance of ICU readmission risk factors may have changed overtime. We used MIMIC-III database with 42,307 ICU stays of 31,749 patients from a US hospital, related to medical services provided from 2001 to 2012. The dataset was initially split into two chronological subsets (2001–2008 and 2008–2012), and then split again into train (70%) and test (30%) datasets. The training datasets were rebalanced through undersampling technique. To identify if the most relevant risk factors changes over time, 13 variables (12 features and one class) were selected and a three-step machine learning approach was executed: (i) Numerical Analysis, to identify overall quantitative changes; (ii) Feature Correlation Value Analysis, to rank the most important risk factors in each subset and compare them to identify any significant changes; and (iii) Classifier Performance Analysis, to identify changes in the risk factors prediction capability, based on the three machine learning algorithms - Multilayer Perceptron, Random Forest and Support Vector Machine. When considering readmission rates, some changes were observed for patients using private insurance (variability of +3.0%) and first admitted in ICU through Medical Intensive Care Unit (−3.1%). Regarding the feature analysis, the two most relevant variables were the same in both datasets, having similar correlation value. When applying the machine learning algorithms in test datasets, the model presented similar results for both periods, achieving the best accuracy of 86.4%, and Area Under ROC Curve (AUC) of 0.642. The difference in AUC values between the first and the second periods varied up to 0.05 (better in the first dataset) and in accuracy up to 4% (better in the second period). Overall results indicate that the most relevant risk factors were stable over the years, with some minor changes. Further research is required to incorporate other readmission risk factors, such as social determinants and mental health and well-being.

Objective: To compare the performance of multiple machine learning algorithms in predicting recurrence after resection of early-stage hepatocellular carcinoma(HCC). Methods: Clinical data of 882 early-stage HCC patients who were admitted to the First Affiliated Hospital of Nanjing Medical University from May 2009 to December 2019 and treated with curative surgical resection were retrospectively collected. There were 701 males and 181 females,with an age of (57.3±10.5)years(range:21 to 86 years). All patients were randomly assigned in a 2∶1 ratio, the training dataset consisted of 588 patients and the test dataset consisted of 294 patients. The construction of machine learning-based prediction models included random survival forest(RSF),gradient boosting machine,elastic net regression and Cox regression model. The prediction accuracy of the model was measured by the concordance index(C-index). The prediction error of the model was measured by the integrated Brier score. Model fit was assessed by the calibration plot. The performance of machine learning models with that of rival model and HCC staging systems was compared. All models were validated in the independent test dataset. Results: Median recurrence-free survival was 61.7 months in the training dataset while median recurrence-free survival was 61.9 months in the validation dataset, there was no significant difference between two datasets in terms of recurrence-free survival(χ²=0.029,P=0.865). The RSF model consisted of 5 commonly used clinicopathological characteristics, including albumin-bilirubin grade,serum alpha fetoprotein,tumor number,type of hepatectomy and microvascular invasion. In both training and test datasets,the RSF model provided the best prediction accuracy,with respective C-index of 0.758(95%CI:0.725 to 0.791) and 0.749(95%CI:0.700 to 0.797),and the lowest prediction error,with respective integrated Brier score of 0.171 and 0.151. The prediction accuracy of RSF model for recurrence after resection of early-stage HCC was superior to that of other machine learning models,rival model(ERASL model) as well as HCC staging systems(BCLC,CNLC and TNM staging),with statistically significant difference(P<0.01). Calibration curves demonstrated good agreement between RSF model-predicted probabilities and observed outcomes.All patients could be stratified into low-risk,intermediate-risk or high-risk group based on RSF model;statistically significant differences among three risk groups were observed in both training and test datasets(all P<0.01). The risk stratification of RSF model was superior to that of TNM staging. Conclusion: The proposed RSF model assembled with 5 commonly used clinicopathological characteristics in this study can predict the recurrence risk with favorable accuracy that may facilitate clinical decision-support for patients with early-stage HCC.

Machine learning (ML) algorithms are a portion of artificial intelligence that may be used to create more accurate algorithmic procedures for estimating an individual's dental age or defining an age classification. This study aims to use ML algorithms to evaluate the efficacy of pulp/tooth area ratio (PTR) in cone-beam CT (CBCT) images to predict dental age classification in adults. CBCT images of 236 Turkish individuals (121 males and 115 females) from 18 to 70 years of age were included. PTRs were calculated for six teeth in each individual, and a total of 1416 PTRs encompassed the study dataset. Support vector machine, classification and regression tree, and random forest (RF) models for dental age classification were employed. The accuracy of these techniques was compared. To facilitate this evaluation process, the available data were partitioned into training and test datasets, maintaining a proportion of 70% for training and 30% for testing across the spectrum of ML algorithms employed. The correct classification performances of the trained models were evaluated. The models' performances were found to be low. The models' highest accuracy and confidence intervals were found to belong to the RF algorithm. According to our results, models were found to be low in performance but were considered as a different approach. We suggest examining the different parameters derived from different measuring techniques in the data obtained from CBCT images in order to develop ML algorithms for age classification in forensic situations.

P68. Machine learning algorithms for predicting patient-controlled analgesia use after minimally invasive spine surgery

Machine learning models play a vital role in the prediction task in several fields of study. In this work, we utilize the ability of machine learning algorithms to predict the occurrence of extreme events in a nonlinear mechanical system. Extreme events are rare events that occur ubiquitously in nature. We consider four machine learning models, namely Logistic Regression, Support Vector Machine, Random Forest and Multi-Layer Perceptron in our prediction task. We train these four machine learning models using training set data and compute the performance of each model using the test set data. We show that the Multi-Layer Perceptron model performs better among the four models in the prediction of extreme events in the considered system. The persistent behaviour of the considered machine learning models is cross-checked with randomly shuffled training set and test set data.

BackgroundCerebral arteriovenous malformation (AVM) is a cerebrovascular disorder posing a risk for intracranial hemorrhage. However, there are few reliable quantitative indices to predict hemorrhage risk accurately. This study aimed to...