Machine learning sleep duration classification in Preschoolers using waist-worn ActiGraphs

ObjectiveTo create a sleep duration classification technique for waist-worn ActiGraph accelerometers in preschool-aged children.MethodsChildren wore ActiGraph wGT3X-BT accelerometers on their right hip for 7 days (24 h/day). Ground truth nap, sleep, and wake were estimated through visual inspection of accelerometer data, guided by sleep log-sheets and previously published visual inspection heuristics. Raw accelerometer data (30Hz) were used to generate 144 features aggregated to 1-min epochs. Machine learning classification (ie, Random Forest and Hidden Markov Modeling [HMM]) predicted nap, sleep, and wake. A simplified prediction formula was also created using features (n = 10) with the highest mean decrease in Gini index during training of Random Forests, and temporally smoothed with rolling median calculations.ResultsChildren (n = 89, mean age = 4.5 years, 67% boys) contributed >600,000 min of accelerometer data. Overall classification accuracy of the Random Forest and HMM classifier was 96.2% (95%CI: 96.1, 96.2%), with a Kappa score of 0.93. Additionally, overall classification accuracy for the temporally smoothed simplified formula was 93.7% (95%CI: 93.6, 93.7%) with Kappa = 0.87. Nap prediction accuracy was 99.8% for the final machine learning model, and 86.1% for the simplified formula. For participant-level daily summaries, generally small but statistically significant differences were found between machine learning and ground truth behaviour predictions, whereas non-significant differences were found between the simplified formulas and ground truth predictions.ConclusionsPredictions for both machine learning and the simplified formula had almost perfect agreement with visual inspection ground truth measurements. Future research is needed to confirm these findings using polysomnography ground truth sleep measurements.     

Constructing an In Silico Three-Class Predictor of Human Intestinal Absorption With Caco-2 Permeability and Dried-DMSO Solubility

Absorption of drugs is the first step after dosing, and it largely affects drug bioavailability. Hence, estimating the fraction of absorption (Fa) in humans is important in the early stages of drug discovery. To achieve correct exclusion of low Fa compounds and retention of potential compounds, we developed a freely available model to classify compounds into 3 levels of Fa capacity using only the chemical structure. To improve Fa prediction, we added predicted binary classification results of membrane permeability measured using Caco-2 cell line (Papp) and dried–dimethyl sulfoxide solubility (accuracy, 0.836; kappa, 0.560). The constructed models can be accessed via a web application.     

Machine intelligence in peptide therapeutics: A next-generation tool for rapid disease screening

Discovery and development of biopeptides are time-consuming, laborious, and dependent on various factors. Data-driven computational methods, especially machine learning (ML) approach, can rapidly and efficiently predict the utility of therapeutic peptides. ML methods offer an array of tools that can accelerate and enhance decision making and discovery for well-defined queries with ample and sophisticated data quality. Various ML approaches, such as support vector machines, random forest, extremely randomized tree, and more recently deep learning methods, are useful in peptide-based drug discovery. These approaches leverage the peptide data sets, created via high-throughput sequencing and computational methods, and enable the prediction of functional peptides with increased levels of accuracy. The use of ML approaches in the development of peptide-based therapeutics is relatively recent; however, these techniques are already revolutionizing protein research by unraveling their novel therapeutic peptide functions. In this review, we discuss several ML-based state-of-the-art peptide-prediction tools and compare these methods in terms of their algorithms, feature encodings, prediction scores, evaluation methodologies, and software utilities. We also assessed the prediction performance of these methods using well-constructed independent data sets. In addition, we discuss the common pitfalls and challenges of using ML approaches for peptide therapeutics. Overall, we show that using ML models in peptide research can streamline the development of targeted peptide therapies.     

肾嫌色细胞癌与不典型透明细胞癌的CT纹理分析研究

目的研究影像组学方法在肾嫌色细胞癌和强化方式不典型的透明细胞癌二者中的应用。方法搜集行肾动脉CTA扫描的108例肾细胞癌患者的临床资料及影像学图像。应用影像组学中的Lasso回归统计方法和机器学习中的随机森林算法提取病例的CTA图像特征并使计算机学习,通过20次重复试验得到平均诊断准确率。患者的临床特征处理采用SPSS 20.0软件,计量资料用t检验,计数资料用χ^2检验,P<0.05为差异具有统计学意义。结果108例肾细胞癌中,透明细胞癌57例,嫌色细胞癌51例。两组病例临床特征中的性别和吸烟史差异具有统计学意义(P<0.05),透明细胞癌更多见于吸烟的男性患者。放射科医师对两组病例诊断的平均准确性为(45.42±3.32)%,低于Lasso回归(76.5±12.26)%和随机森林算法(78.5±6.3)%。在两组病例中,随机森林算法给出的总准确性、对嫌色细胞癌诊断的特异性要高于Lasso回归,Lasso回归对透明细胞癌诊断的敏感性高于随机森林算法。结论影像组学方法可以对肾嫌色细胞癌及透明细胞癌做出有效的鉴别诊断,且诊断能力高于放射科医师的能力。影像组学作为一种新兴的研究方法,有望为医学发展带来重要变革。     

Joint 3-D vessel segmentation and centerline extraction using oblique Hough forests with steerable filters

ContributionsWe propose a novel framework for joint 3-D vessel segmentation and centerline extraction. The approach is based on multivariate Hough voting and oblique random forests (RFs) that we learn from noisy annotations. It relies on steerable filters for the efficient computation of local image features at different scales and orientations.ExperimentsWe validate both the segmentation performance and the centerline accuracy of our approach both on synthetic vascular data and four 3-D imaging datasets of the rat visual cortex at 700 nm resolution. First, we evaluate the most important structural components of our approach: (1) Orthogonal subspace filtering in comparison to steerable filters that show, qualitatively, similarities to the eigenspace filters learned from local image patches. (2) Standard RF against oblique RF. Second, we compare the overall approach to different state-of-the-art methods for (1) vessel segmentation based on optimally oriented flux (OOF) and the eigenstructure of the Hessian, and (2) centerline extraction based on homotopic skeletonization and geodesic path tracing.ResultsOur experiments reveal the benefit of steerable over eigenspace filters as well as the advantage of oblique split directions over univariate orthogonal splits. We further show that the learning-based approach outperforms different state-of-the-art methods and proves highly accurate and robust with regard to both vessel segmentation and centerline extraction in spite of the high level of label noise in the training data.     

Radiomics for classification of bone mineral loss: A machine learning study

PurposeThe purpose of this study was to develop predictive models to classify osteoporosis, osteopenia and normal patients using radiomics and machine learning approaches.Materials and methodsA total of 147 patients were included in this retrospective single-center study. There were 12 men and 135 women with a mean age of 56.88 ± 10.6 (SD) years (range: 28–87 years). For each patient, seven regions including four lumbar and three femoral including trochanteric, intertrochanteric and neck were segmented on bone mineral densitometry images and 54 texture features were extracted from the regions. The performance of four feature selection methods, including classifier attribute evaluation (CLAE), one rule attribute evaluation (ORAE), gain ratio attribute evaluation (GRAE) and principal components analysis (PRCA) along with four classification methods, including random forest (RF), random committee (RC), K-nearest neighbor (KN) and logit-boost (LB) were evaluated. Four classification categories, including osteopenia vs. normal, osteoporosis vs. normal, osteopenia vs. osteoporosis and osteoporosis + osteopenia vs. osteoporosis were examined for the defined seven regions. The classification model performances were evaluated using the area under the receiver operator characteristic curve (AUC).ResultsThe AUC values ranged from 0.50 to 0.78. The combination of methods RF + CLAE, RF + ORAE and RC + ORAE yielded highest performance (AUC = 0.78) in discriminating between osteoporosis and normal state in the trochanteric region. The combinations of RF + PRCA and LB + PRCA had the highest performance (AUC = 0.76) in discriminating between osteoporosis and normal state in the neck region.ConclusionThe machine learning radiomic approach can be considered as a new method for bone mineral deficiency disease classification using bone mineral densitometry image features.     

基于随机森林与logistic回归的高血压影响因素研究

目的 针对铜陵市天桥社区居民体检数据中多因素、有效样本有限的情况，挖掘与分析高血压影响因素与因素间的交互效应，为高血压干预提供参考。方法 选取2017年该社区801例体检数据为研究对象，采用随机森林方法，筛选出重要性评分较大的特征，代入logistic完全二次回归模型，逐步回归分析影响因素及因素间的交互效应。结果 随机森林模型准确率83.67%，特征重要性前10项为年龄、糖尿病、锻炼频率、体质指数、总胆固醇、吸烟情况、饮酒情况、中心性肥胖、甘油三酯、血尿素氨。Logistic完全二次回归模型准确率84.17%，输出2条主效应、8条二次交互效应。主效应中有统计学意义(P<0.05)的特征有年龄、锻炼频率，二次交互效应中有统计学意义(P<0.05)的特征有年龄、糖尿病、体质指数、总胆固醇、吸烟情况、饮酒情况、甘油三酯、血尿素氨。结论 随机森林与logistic完全二次回归模型相结合，解决了经典方法难以从多因素、样本有限的数据中挖掘交互效应的问题，获得高血压影响因素与因素间的交互效应，为高血压干预提供有益的指导。     

Seasonality of Clostridium difficile in the natural environment

Clostridium difficile is considered the leading cause of antibiotic‐associated disease worldwide. In the past decade, a large number of studies have focused on identifying the main sources of contamination in order to elucidate the complete life cycle of the infection. Hospitals, animals and retail foods have been considered as potential vectors. However, the prevalence of C. difficile in these types of samples was found to be rather low, suggesting that other contamination routes must exist. This study explores the presence of C. difficile in the natural environment and the seasonal dynamics of the bacterium. C. difficile was isolated from a total of 45 samples out of 112 collected (40.2%) on 56 sampling points. A total of 17 points were positive only during the winter sampling (30.4%), 10 were positive only during the summer sampling (17.9%) and 9 sampling points (16.1%) were positive in both summer sampling and winter sampling. Spore counts in soil samples ranged between 50 and 250 cfu/g for 24.4% of the positive samples, with the highest concentrations detected in samples collected in the forest during winter campaign (200–250 cfu/g). A total of 17 different PCR ribotypes were identified, and 15 of them had the genes coding for toxins A and B. Most of those ribotypes had not previously been found or had been isolated only sporadically (<1% of samples) from hospitals in Belgium. Regarding antimicrobial susceptibility, most of the resistant strains were found during the summer campaign. These findings bear out that C. difficile is present in the natural environment, where the bacterium undergoes seasonal variations.     

Advancing the deer calcaneus model for bone adaptation studies: ex vivo strains obtained after transecting the tension members suggest an unrecognized important role for shear strains

Sheep and deer calcanei are finding increased use as models for studies of bone adaptation, including advancing understanding of how the strain (deformation) environment influences the ontogenetic emergence of biomechanically relevant structural and material variations in cortical and trabecular bone. These artiodactyl calcanei seem ideal for these analyses because they function like simply loaded short‐cantilevered beams with net compression and tension strains on the dorsal and plantar cortices, respectively. However, this habitual strain distribution requires more rigorous validation because it has been shown by limited in vivo and ex vivo strain measurements obtained during controlled ambulation (typically walking and trotting). The conception that these calcanei are relatively simply and habitually loaded ‘tension/compression bones’ could be invalid if infrequent, though biologically relevant, loads substantially change the location of the neutral axis (NA) that separates ‘compression’ and ‘tension’ regions. The effect on calcaneus strains of the tension members (plantar ligament and flexor tendon) is also not well understood and measuring strains after transecting them could reveal that they significantly modulate the strain distribution. We tested the hypothesis that the NA location previously described during simulated on‐axis loads of deer calcanei would exhibit limited variations even when load perturbations are unusual (e.g. off‐axis loads) or extreme (e.g. after transection of the tension members). We also examined regional differences in the predominance of the three strain modes (tension, compression, and shear) in these various load conditions in dorsal, plantar, medial, and lateral cortices. In addition to considering principal strains (tension and compression) and maximum shear strains, we also considered material‐axis (M‐A) shear strains. M‐A shear strains are those that are aligned along the long axis of the bone and are considered to have greater biomechanical relevance than maximum shear strains because failure theories of composite materials and bone are often based on stresses or strains in the principal material directions. We used the same load apparatus from our prior study of mule deer calcanei. Results showed that although the NA rotated up to 8° medially and 15° laterally during these off‐axis loads, it did not shift dramatically until after transection of all tension members. When comparing results based on maximum shear strain data vs. M‐A shear strain data, the dominant strain mode changed only in the plantar cortex – as expected (in accordance with our a priori view) it was tension when M‐A shear strains were considered (shear : tension = 0.2) but changed to dominant shear when maximum shear strain data were considered (shear : tension = 1.3). This difference leads to different conclusions and speculations regarding which specific strain modes and magnitudes most strongly influence the emergence of the marked mineralization and histomorphological differences in the dorsal vs. plantar cortices. Consequently, our prior simplification of the deer calcaneus model as a simply loaded ‘tension/compression bone’ (i.e. plantar/dorsal) might be incorrect. In vivo and in finite element analyses are needed to determine whether describing it as a ‘shear‐tension/compression’ bone is more accurate. Addressing this question will help to advance the artiodactyl calcaneus as an experimental model for bone adaptation studies.