253 Better and faster automatic sleep staging with artificial intelligence

SLEEP ◽  
2021 ◽  
Vol 44 (Supplement_2) ◽  
pp. A102-A102
Author(s):  
Massimiliano Grassi ◽  
Daniela Caldirola ◽  
Silvia Daccò ◽  
Giampaolo Perna ◽  
Archie Defillo
Keyword(s):  
Statistical Significance ◽  
Indirect Comparison ◽  
Machine Learning Algorithms ◽  
Bootstrap Resampling ◽  
Sleep Staging ◽  
Partial Reduction ◽  
Sleep Scoring ◽  
Percent Agreement ◽  
Automatic Staging ◽  
Very High

Abstract Introduction Sleep staging of polysomnography (PSG) is a time-consuming task, it requires significant training, and significant variability among scorers is expected. A new software (MEBsleep by Medibio Limited) was developed to automatically perform sleep scoring based on machine learning algorithms. This study aimed to perform an extensive investigation of its agreement with expert sleep technicians. Methods Forty polysomnography recordings of patients that were referred for sleep evaluation to three sleep clinics were retrospectively collected. Three experienced technicians independently staged the recording complying with the scoring rules of the American Academy of Sleep Medicine guidelines. Positive Percent Agreement (PPA), Positive Predictive Value (PPV), and other agreement statistics between the automatic and manual staging, among the staging performed by the three technicians, and their differences were calculated. Bootstrap resampling was used to calculate 95% confidence intervals and statistical significance of the differences. Results Automatic staging took less than two minutes per PSG on a consumer laptop. The automatic staging resulted for the most comparable (PPA difference of N1, N3, and REM; PPV difference of N1, N2, N3, and REM) or statistically significantly more in agreement with the technicians’ staging than the between-technician agreement (PPA difference of N2: 3.90%, 95% bootstrap CI 1.79%-6.01%; PPV difference of Wake: 1.16%, 95% bootstrap CI 0.64%/1.67%), with the sole exception of a partial reduction in the positive percent agreement of the Wake stage (PPA difference of Wake -7.04%, 95% bootstrap CI -10.40%/-3.85%). The automatic staging also demonstrated very high accuracy in an indirect comparison with other similar software. Conclusion Given these promising results, the use of this software may support sleep clinicians by improving efficiency in sleep scoring. Support (if any):

Novel Approach for Battery Type Determination: A Mere Electrical Alternative

2021 ◽  
Author(s):  
İsmail Can Dikmen ◽  
Teoman Karadağ
Keyword(s):  
Integrated Circuits ◽  
Decision Tree ◽  
Statistical Significance ◽  
High Capacity ◽  
Electrical Energy ◽  
Machine Learning Algorithms ◽  
Battery Management ◽  
Separation Function ◽  
Novel Approach ◽  
Tree Algorithms

Abstract Today, the storage of electrical energy is one of the most important technical challenges. The increasing number of high capacity, high-power applications, especially electric vehicles and grid energy storage, points to the fact that we will be faced with a large amount of batteries that will need to be recycled and separated in the near future. An alternative method to the currently used methods for separating these batteries according to their chemistry is discussed in this study. This method can be applied even on integrated circuits due to its ease of implementation and low operational cost. In this respect, it is also possible to use it in multi-chemistry battery management systems to detect the chemistry of the connected battery. For the implementation of the method, the batteries are connected to two different loads alternately. In this way, current and voltage values ​​are measured for two different loads without allowing the battery to relax. The obtained data is pre-processed with a separation function developed based on statistical significance. In machine learning algorithms, artificial neural network and decision tree algorithms are trained with processed data and used to determine battery chemistry with 100% accuracy. The efficiency and ease of implementation of the decision tree algorithm in such a categorization method are presented comparatively.

scholarly journals The Nutritional Preponderance and Propensity to COVID-19 in Children

2021 ◽  
pp. 1-2
Author(s):  
Dhanasekhar Kesavelu ◽  
◽  
Lekha VS ◽  
Sarah Nalliannan ◽  
◽  
...  
Keyword(s):  
Nutritional Status ◽  
Statistical Significance ◽  
Case Series ◽  
Age Group ◽  
Significant Relation ◽  
Single Centre ◽  
Equal Distribution ◽  
Nutritional Profile ◽  
Status Analysis ◽  
Very High

Aims & Objectives: Nutrition plays a very important role in the Immunity and Immunoregulation in children. The risk of acquiring an infection is very high in malnourished and underweight children, which is what we venture out to find out in our study by researching the susceptibility for COVID-19 in children and comparing it with their nutritional status. Materials & Methods: Our Tertiary Childrens hospital had 46 COVID positive children admitted in 2020, the nutritional status analysis, showed that there were four children 8.6% (n=4) in Obese category, 19.5%, nine (n=9) children were Overweight, 17.39 %, 8 children Underweight (n=8) and 25 children were 54.3% well nourished. Results: This case series describes the various features in COVID-19 in children with and without co-morbidities primarily focusing on the nutritional profile. This is the first single centre case series globally on COVID-19. Our cohort showed no significant relation between COVID-19 and the nutritional status. We saw an equal distribution of COVID-19 in children irrespective of their nutritional status at admission. Conclusions: We did not notice any statistical significance in the age group or the nutritional status in children infected with COVID-19

scholarly journals Deep Learning Approach for Stages of Severity Classification in Diabetic Retinopathy Using Color Fundus Retinal Images

2021 ◽  
Vol 2021 ◽  
pp. 1-8
Author(s):  
Silky Goel ◽  
Siddharth Gupta ◽  
Avnish Panwar ◽  
Sunil Kumar ◽  
Madhushi Verma ◽  
...  
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Vision Loss ◽  
Early Stage ◽  
Machine Learning Algorithms ◽  
Fundus Photography ◽  
Huge Amount ◽  
Early Screening ◽  
Image Metadata ◽  
Very High

Diabetes is a very fast-growing disease in India, with currently more than 72 million patients. Prolonged diabetes (about almost 20 years) can cause serious loss to the tiny blood vessels and neurons in the patient eyes, called diabetic retinopathy (DR). This first causes occlusion and then rapid vision loss. The symptoms of the disease are not very conspicuous in its early stage. The disease is featured by the formation of bloated structures in the retinal area called microaneurysms. Because of negligence, the condition of the eye worsens into the generation of more severe blots and damage to retinal vessels causing complete loss of vision. Early screening and monitoring of DR can reduce the risk of vision loss in patients with high possibilities. But the diabetic retinopathy detection and its classification by a human, is a challenging and error-prone task, because of the complexity of the image captured by color fundus photography. Machine learning algorithms armed with some feature extraction techniques have been employed earlier to detect and classify the levels of DR. However, these techniques provide below-par accuracy. Now, with the advent of deep learning (DL) techniques in computer vision, it has become possible to achieve very high levels of accuracy. DL models are an abstraction of the human brain coupled with the eyes. To create a model from scratch and train it is a cumbersome task requiring a huge amount of images. This deficiency of the DL techniques can be patched up by employing another technique to a task called transfer learning. In this, a DL model is trained on image metadata, and to learn features it used hundreds of classes from the DR fundus images. This enables professionals to create models capable of classifying unseen images into a proper grade or level with acceptable accuracy. This paper proposed a DL model coupled with different classifiers to classify the fundus image into its correct class of severity. We have trained the model on IDRD images and it has proven to show very high accuracy.

scholarly journals A universal, open-source, high-performance tool for automated sleep staging

2021 ◽  
Author(s):  
Raphael Vallat ◽  
Matthew P Walker
Keyword(s):  
Open Source ◽  
Scoring System ◽  
High Performance ◽  
Small Samples ◽  
Sleep Staging ◽  
Sleep Scoring ◽  
Industry Standard ◽  
Staging Accuracy ◽  
The World ◽  
Novel Algorithm

The creation of a completely automated sleep-scoring system that is highly accurate, flexible, well validated, free and simple to use by anyone has yet to be accomplished. In part, this is due to the difficulty of use of existing algorithms, algorithms having been trained on too small samples, and paywall demotivation. Here we describe a novel algorithm trained and validated on +27,000 hours of polysomnographic sleep recordings across heterogeneous populations around the world. This tool offers high sleep-staging accuracy matching or exceeding human accuracy and interscorer agreement no matter the population kind. The software is easy to use, computationally low-demanding, open source, and free. Such software has the potential to facilitate broad adoption of automated sleep staging with the hope of becoming an industry standard.

scholarly journals Association between Human Development Index and Delay on Arrival to Stroke Unit

2021 ◽  
Author(s):  
Gabriel Souza Suzart ◽  
Ingrid Sanchez ◽  
Daniel Guimarães ◽  
Pedro Augusto Assis Lopes ◽  
Pedro Antonio Pereira de Jesus
Keyword(s):  
Human Development ◽  
Human Development Index ◽  
Statistical Significance ◽  
Cross Sectional Study ◽  
Cross Sectional ◽  
Development Index ◽  
Specialized Service ◽  
Time Of Admission ◽  
The Impact ◽  
Very High

Background: Stroke outcomes depend somehow on the time taken from the symptoms onset until arrival to the specialized service. However, as it lacks literature exploring the impact of socio-demographics factors on this time, we investigated the association between Human Development Index and delay on arrival to specialized service. Design and setting: Cross-sectional study from a prospective cohort (PMID=33719516) at Hospital Geral Roberto Santos. Methods: From a total of 454 stroke patients, 156 were included in this study because they had registered address, time of admission and of symptoms onset. Patients had HDI defined by their address and were grouped into HDI categories. Results: In our sample, 57 (36,5%) individuals had medium HDI, 70 (44,9%) high HDI and 29 (18,6%) very high HDI. Very high HDI patients’ delay (2:01; 1:22-2:57) was lower than high HDI (3:05; 2:05-5:26) and medium HDI (2:25;1:45-4:04) patients. There was statistical significance comparing these groups (X²=11,41;p<0,05), but a post-hoc test revealed statistical difference just between the very high HDI and high HDI groups (p<0.05). Conclusions: We expected to find a direct relation between delay on arrival to the stroke service and HDI categories. However, this was not observed. *Authors contributed equally.

scholarly journals Outcome of Patients Younger Than 50 Years Old Diagnosed with Myelodysplastic Syndromes (MDS): Single Institution Experience

Blood ◽  
2016 ◽  
Vol 128 (22) ◽  
pp. 5541-5541
Author(s):  
Aref Al-Kali ◽  
Rong He ◽  
Mrinal M Patnaik ◽  
David S Viswanatha ◽  
Patricia T. Greipp ◽  
...  
Keyword(s):  
High Risk ◽  
Myelodysplastic Syndromes ◽  
Older Patients ◽  
Research Funding ◽  
Multivariate Analyses ◽  
Statistical Significance ◽  
Young Patients ◽  
Risk Groups ◽  
World Health ◽  
Very High

Abstract Background: Myelodysplastic syndromes (MDS) are rare hematological neoplasms that are more typically seen in elderly patients. Young patients (< 50 years old) have been reported to comprise between 3-6% in the Surveillance, Epidemiology and End Results (SEER) (Ma X et al, Cancer 2007; Rollison R et al, Blood 2008) with a better overall survival (OS). We hereby report the characteristics of young MDS patients with long survival follow-up. Methods: A total of 1012 MDS patients' data from Oct 1993 to Dec 2015 at Mayo Clinic were reviewed after appropriate IRB approval was obtained. All cases had their bone marrow slides reviewed at our institution. Patients youngers than 50 years old (yMDS) were included as cohort 1, while the rest was included as control group (cohort 2). Prognostic factors were analyzed by univariate and multivariate analyses. Survival estimates were calculated using Kaplan-Meier curves and univariate and multivariate analyses was based on log-rank testing using JMP software version 10. Results: Characteristics: We identified 68 (7%) yMDS patients with a median age of 42 years (range, 18-49). Female gender was more common in yMDS (43% vs 31%, p= 0.05). Upon comparison between cohort 1 and 2, only platelets were significantly lower in yMDS (61 vs 102, p <0.0001), but not white blood cells, hemoglobin or degree of marrow fibrosis. MDS subtyping according to World Health Organization 2016 showed single lineage dysplasia in 15% vs 2%, multilineage dysplasia 24% vs 36% , ring sideroblasts 9% vs 15%, isolated del (5q) 1% vs 3%, excess blasts 44% vs 31%, and unclassifiable in 6% vs 5%. As expected, therapy related MDS (t-MDS) was more frequent in yMDS (33% vs 17%, p= 0.005). Transformation to acute myeloid leukemia (AML) was also more frequent in yMDS (28 % vs 11%, p= 0.0004). Compared to cohort 2, yMDS IPSS-R scores were very high, high, intermediate, low, and very low in 24% vs 13%, 29% vs 16%, 21% vs 20%, 19% vs 35%, and 7% vs 17%, respectively. Allogenic hematopoietic cell transplantation (HCT) was more frequent in yMDS (42% vs 4%, p< 0.0001). Survival outcome: Median OS was longer for cohort 1 vs cohort 2 but did not reach statistical significance (43 vs 21 months, p= 0.1). Median progression free survival (PFS) was shorter for cohort 1 vs cohort 2 but did not reach statistical significance (8 vs 12 months, p= 0.3). Median OS for cohort 1 based on R-IPSS was 44, 105, 40, 18, and 12 months for very low, low, intermediate, high and high risk groups, respectively (p= 0.09). Median OS was shorter in t-MDS vs de novo MDS in cohort 1 (13 vs 47 months, p = 0.04). Young patients who had transformed to AML had a worse median OS (18 vs 93 months, p=0.001). On multivariate analysis neither t-MDS nor R-IPSS had a statistically significant impact on OS. Conclusions: MDS is rarely diagnosed under the age of 50. IPSS-R was less powerful in detecting differences between its risk groups for this patient population, although more than half of the patients with yMDS had either high or very high risk. Among this cohort of yMDS patients, there was a significantly higher proportion with therapy-related myeloid neoplasms compared to older patients (one third of patients), with subsequent higher rates of transformation to AML and higher allogeneic HCT. In our study, we did not find an improved OS for yMDS patients compared to older patients. Disclosures Al-Kali: Celgene: Research Funding; Onconova Therapeutics, Inc.: Research Funding.

scholarly journals Tree Species Classification in Mixed Deciduous Forests Using Very High Spatial Resolution Satellite Imagery and Machine Learning Methods

2020 ◽  
Vol 12 (23) ◽  
pp. 3926
Author(s):  
Martina Deur ◽  
Mateo Gašparović ◽  
Ivan Balenović
Keyword(s):  
Machine Learning ◽  
Satellite Imagery ◽  
Tree Species ◽  
Learning Algorithms ◽  
Spectral Characteristics ◽  
Machine Learning Algorithms ◽  
Support Vector ◽  
Species Classification ◽  
Tree Species Classification ◽  
Very High

Spatially explicit information on tree species composition is important for both the forest management and conservation sectors. In combination with machine learning algorithms, very high-resolution satellite imagery may provide an effective solution to reduce the need for labor-intensive and time-consuming field-based surveys. In this study, we evaluated the possibility of using multispectral WorldView-3 (WV-3) satellite imagery for the classification of three main tree species (Quercus robur L., Carpinus betulus L., and Alnus glutinosa (L.) Geartn.) in a lowland, mixed deciduous forest in central Croatia. The pixel-based supervised classification was performed using two machine learning algorithms: random forest (RF) and support vector machine (SVM). Additionally, the contribution of gray level cooccurrence matrix (GLCM) texture features from WV-3 imagery in tree species classification was evaluated. Principal component analysis confirmed GLCM variance to be the most significant texture feature. Of the 373 visually interpreted reference polygons, 237 were used as training polygons and 136 were used as validation polygons. The validation results show relatively high overall accuracy (85%) for tree species classification based solely on WV-3 spectral characteristics and the RF classification approach. As expected, an improvement in classification accuracy was achieved by a combination of spectral and textural features. With the additional use of GLCM variance, the overall accuracy improved by 10% and 7% for RF and SVM classification approaches, respectively.

scholarly journals An Empirical Study of Machine Learning Algorithms for Stock Daily Trading Strategy

2019 ◽  
Vol 2019 ◽  
pp. 1-30 ◽  
Author(s):  
Dongdong Lv ◽  
Shuhan Yuan ◽  
Meizi Li ◽  
Yang Xiang
Keyword(s):  
Machine Learning ◽  
Transaction Cost ◽  
Stock Price ◽  
Large Scale ◽  
Statistical Significance ◽  
Significance Test ◽  
Machine Learning Algorithms ◽  
Experimental Results ◽  
Price Trends ◽  
The Impact

According to the forecast of stock price trends, investors trade stocks. In recent years, many researchers focus on adopting machine learning (ML) algorithms to predict stock price trends. However, their studies were carried out on small stock datasets with limited features, short backtesting period, and no consideration of transaction cost. And their experimental results lack statistical significance test. In this paper, on large-scale stock datasets, we synthetically evaluate various ML algorithms and observe the daily trading performance of stocks under transaction cost and no transaction cost. Particularly, we use two large datasets of 424 S&P 500 index component stocks (SPICS) and 185 CSI 300 index component stocks (CSICS) from 2010 to 2017 and compare six traditional ML algorithms and six advanced deep neural network (DNN) models on these two datasets, respectively. The experimental results demonstrate that traditional ML algorithms have a better performance in most of the directional evaluation indicators. Unexpectedly, the performance of some traditional ML algorithms is not much worse than that of the best DNN models without considering the transaction cost. Moreover, the trading performance of all ML algorithms is sensitive to the changes of transaction cost. Compared with the traditional ML algorithms, DNN models have better performance considering transaction cost. Meanwhile, the impact of transparent transaction cost and implicit transaction cost on trading performance are different. Our conclusions are significant to choose the best algorithm for stock trading in different markets.

Sign in / Sign up

Export Citation Format

Share Document