High Level Of Prediction Accuracy Research Articles

Demystifying the Black Box: The Importance of Interpretability of Predictive Models in Neurocritical Care

Neurocritical care patients are a complex patient population, and to aid clinical decision-making, many models and scoring systems have previously been developed. More recently, techniques from the field of machine learning have been applied to neurocritical care patient data to develop models with high levels of predictive accuracy. However, although these recent models appear clinically promising, their interpretability has often not been considered and they tend to be black box models, making it extremely difficult to understand how the model came to its conclusion. Interpretable machine learning methods have the potential to provide the means to overcome some of these issues but are largely unexplored within the neurocritical care domain. This article examines existing models used in neurocritical care from the perspective of interpretability. Further, the use of interpretable machine learning will be explored, in particular the potential benefits and drawbacks that the techniques may have when applied to neurocritical care data. Finding a solution to the lack of model explanation, transparency, and accountability is important because these issues have the potential to contribute to model trust and clinical acceptance, and, increasingly, regulation is stipulating a right to explanation for decisions made by models and algorithms. To ensure that the prospective gains from sophisticated predictive models to neurocritical care provision can be realized, it is imperative that interpretability of these models is fully considered.

Predicting Poverty Using Geospatial Data in Thailand

Poverty statistics are conventionally compiled using data from socioeconomic surveys. This study examines an alternative approach to estimating poverty by investigating whether readily available geospatial data can accurately predict the spatial distribution of poverty in Thailand. In particular, the geospatial data examined in this study include the intensity of night-time light (NTL), land cover, vegetation index, land surface temperature, built-up areas, and points of interest. The study also compares the predictive performance of various econometric and machine-learning methods such as generalized least squares, neural network, random forest, and support-vector regression. Results suggest that the intensity of NTL and other variables that approximate population density are highly associated with the proportion of an area’s population that are living in poverty. The random forest technique yielded the highest level of prediction accuracy among the methods considered in this study, primarily due to its capability to fit complex association structures even with small-to-medium-sized datasets. This obtained result suggests the potential applications of using publicly accessible geospatial data and machine-learning methods for timely monitoring of the poverty distribution. Moving forward, additional studies are needed to improve the predictive power and investigate the temporal stability of the relationships observed.

Forecasting the loss given default of bank loans with a hybrid multilayer LGD model by extending multidimensional signals

According to the Basel II and Basel III accords, loss given default (LGD) is an important component of credit risk assessment. To improve the accuracy of LGD prediction, this paper uses signaling theory and machine learning methods to study the LGD predictions of commercial banks by considering the venture quality and level of uncertainty. First, from the perspective of signal costs, signaling theory is used to analyze the impacts of venture quality and extended level of uncertainty signals on LGD. Then, a new hybrid multilayer LGD model-based hierarchical framework is proposed for studying the LGD predictions of commercial banks in a high-dimensional unbalanced data context. The experimental results demonstrate that the venture quality and level of uncertainty significantly affect LGD. Additionally, the hybrid multilayer LGD model – consisting of logistic classification, random oversampling examples combined with random forest classification and ordinary least squares regression – not only yields high levels of prediction accuracy and interpretability in high-dimensional unbalanced data contexts but is also robust regardless of the selection of the training set and samples used. The results of the study serve as an important reference for decision-making by commercial banks in their assessment of default losses and the risk management of small/startup companies.

A Novel Multi-Neural Ensemble Approach for Cancer Diagnosis

ABSTRACT Cancer is a complex worldwide health concern that resulted in 10 million cancer deaths in 2018; hence, early cancer detection is crucial. Early detection involves developing more precise technology that offers information about the patient’s cancer, allowing clinicians to make better-informed treatment options. This study provides an in-depth analysis of multiple cancers. This study also exhibits a good survey of the machine or deep learning techniques used in cancer research. Also, the study proposed a stacking-based multi-neural ensemble learning method’s prediction performance on eight datasets, including the benchmark datasets like Wisconsin Breast cancer dataset, mesothelioma, cervical cancer, non-small cell lung cancer survival dataset, and prostate cancer dataset. This study also analyzes the three real-time cancer datasets (Lung, Ovarian & Leukemia) of the Jammu and Kashmir region. The simulation findings indicate that the methodology described in our study attained the highest level of prediction accuracy across all types of cancer data sets. Additionally, the proposed approach has been statistically validated. The purpose of this investigation was to develop and evaluate a prediction model that might be used as a biomarker for malignancy based on anthropometric, clinical, imaging, and gene data.

Modeling of BRT System Travel Time Prediction Using AVL Data and ANN Approach

Improving the quality of public transportation systems and encouraging passengers to use them are effective solutions for reducing transportation problems in metropolitan. Prediction of travel time and providing information to passengers are significant factors in this process. In this research not only the travel time components in Bus Rapid Transit (BRT) system were investigated but also an Artificial Neural Network (ANN) model and a regression model for travel time prediction were presented. To enhance this aim, data was collected by AVL data and field observation and after investigating the primary independent variables, the significant ones were determined using statistical analysis, then ANN development was done. Moreover, linear regression method was used for this purpose. The results prove that although both models have high level of prediction accuracy, ANN model outperform the regression model and the accuracy for the route sections with no signalized intersections is higher than the others.

Forecasting the CO2 Emissions at the Global Level: A Multilayer Artificial Neural Network Modelling

Better accuracy in short-term forecasting is required for intermediate planning for the national target to reduce CO2 emissions. High stake climate change conventions need accurate predictions of the future emission growth path of the participating countries to make informed decisions. The current study forecasts the CO2 emissions of the 17 key emitting countries. Unlike previous studies where linear statistical modeling is used to forecast the emissions, we develop a multilayer artificial neural network model to forecast the emissions. This model is a dynamic nonlinear model that helps to obtain optimal weights for the predictors with a high level of prediction accuracy. The model uses the gross domestic product (GDP), urban population ratio, and trade openness, as predictors for CO2 emissions. We observe an average of 96% prediction accuracy among the 17 countries which is much higher than the accuracy of the previous models. Using the optimal weights and available input data the forecasting of CO2 emissions is undertaken. The results show that high emitting countries, such as China, India, Iran, Indonesia, and Saudi Arabia are expected to increase their emissions in the near future. Currently, low emitting countries, such as Brazil, South Africa, Turkey, and South Korea will also tread on a high emission growth path. On the other hand, the USA, Japan, UK, France, Italy, Australia, and Canada will continuously reduce their emissions. These findings will help the countries to engage in climate mitigation and adaptation negotiations.

New Insights Into Drug Repurposing for COVID-19 Using Deep Learning.

The coronavirus disease 2019 (COVID-19) has continued to spread worldwide since late 2019. To expedite the process of providing treatment to those who have contracted the disease and to ensure the accessibility of effective drugs, numerous strategies have been implemented to find potential anti-COVID-19 drugs in a short span of time. Motivated by this critical global challenge, in this review, we detail approaches that have been used for drug repurposing for COVID-19 and suggest improvements to the existing deep learning (DL) approach to identify and repurpose drugs to treat this complex disease. By optimizing hyperparameter settings, deploying suitable activation functions, and designing optimization algorithms, the improved DL approach will be able to perform feature extraction from quality big data, turning the traditional DL approach, referred to as a “black box,” which generalizes and learns the transmitted data, into a “glass box” that will have the interpretability of its rationale while maintaining a high level of prediction accuracy. When adopted for drug repurposing for COVID-19, this improved approach will create a new generation of DL approaches that can establish a cause and effect relationship as to why the repurposed drugs are suitable for treating COVID-19. Its ability can also be extended to repurpose drugs for other complex diseases, develop appropriate treatment strategies for new diseases, and provide precision medical treatment to patients, thus paving the way to discover new drugs that can potentially be effective for treating COVID-19.

Point-cloud deep learning of porous media for permeability prediction

We propose a novel deep learning framework for predicting the permeability of porous media from their digital images. Unlike convolutional neural networks, instead of feeding the whole image volume as inputs to the network, we model the boundary between solid matrix and pore spaces as point clouds and feed them as inputs to a neural network based on the PointNet architecture. This approach overcomes the challenge of memory restriction of graphics processing units and its consequences on the choice of batch size and convergence. Compared to convolutional neural networks, the proposed deep learning methodology provides freedom to select larger batch sizes due to reducing significantly the size of network inputs. Specifically, we use the classification branch of PointNet and adjust it for a regression task. As a test case, two and three dimensional synthetic digital rock images are considered. We investigate the effect of different components of our neural network on its performance. We compare our deep learning strategy with a convolutional neural network from various perspectives, specifically for maximum possible batch size. We inspect the generalizability of our network by predicting the permeability of real-world rock samples as well as synthetic digital rocks that are statistically different from the samples used during training. The network predicts the permeability of digital rocks a few thousand times faster than a lattice Boltzmann solver with a high level of prediction accuracy.

Показники ендотеліальної дисфункції при цукровому діабеті 2-го типу як визначальні чинники розвитку ускладнень

Актуальність. Для з’ясування патогенетичної ролі основних чинників, що відображають розвиток ендотеліальної дисфункції (ЕДФ), була зроблена спроба у рамках одного дослідження розрахувати їх вплив на розвиток судинних ускладнень цукрового діабету (ЦД) 2-го типу. До числа таких показників увійшли ендотелін-1 (ЕТ-1), фактор некрозу пухлини (TNF-α) та дієнові кон’югати (ДК). Також було обрано два захисні фактори — оксид азоту (NO) і ендотеліальна NO-синтаза (eNOS). Мета дослідження: вивчити вплив показників ЕДФ при ЦД 2-го типу на розвиток його ускладнень. Матеріали та методи. До дослідження залучено дані 152 хворих на ЦД 2-го типу, вік пацієнтів становив від 34 до 80 років, у середньому 53,9 ± 8,4 року. Жінок було 95 (62,5 %), чоловіків — 57 (37,5 %). Відповідно до клінічних рекомендацій, за результатами клініко-лабораторних обстежень визначали наявність ретинопатії, нефропатії за рівнями мікроальбумінурії та швидкості клубочкової фільтрації (ШКФ), сенсорної полінейропатії, макроангіопатії нижніх кінцівок та артеріальної гіпертензії. У плазмі крові імуноферментним і біохімічним методами визначали вміст чинників ЕДФ. Для математичної обробки отриманих даних використовували програму Statistica 10 (StatSoft, Inc., США), пакет GLZ. Результати. На наявність ретинопатії сильно впливав рівень у крові ЕТ-1 (β = 3,1; р = 2,77Е-05), а також TNF-α (β = 0,019; р = 0,050). На наявність макроангіопатії нижніх кінцівок — рівень в крові NO (β = 27,5; р = 0,021) і TNF-α (β = –0,48; р = 0,017). Для полінейропатії значущим виявився тільки один показник — ЕТ-1 (β = 1,8; р = 0,016). На наявність нефропатії за ШКФ впливали рівні у крові ЕТ-1 (β = 2,7; р = 7,58Е-07) і TNF-α (β = 0,013; р = 0,047). Розвитку нефропатії за мікроальбумінурією сприяв ЕТ-1 (β = 4,5; р = 0,043). На наявність артеріальної гіпертензії впливали ЕТ-1 (β = 1,69; р = 2,54Е-04) і TNF-α (β = 0,017; р = 0,014). Шляхом багатофакторної логістичної регресії побудовані моделі прогнозування ймовірності розвитку ускладнень ЦД 2-го типу на підставні вивчених чинників ЕДФ. Усі моделі адекватні і мають високий рівень прогностичної точності (60,5–90,8 %). Висновки. На розвиток ускладнень ЦД 2-го типу по-різному впливали чинники ЕДФ: ЕТ-1 безпосередньо впливав на розвиток усіх ускладнень (окрім макроангіопатії нижніх кінцівок); TNF-α здійснював вплив на розвиток ретинопатії, нефропатії за ШКФ і артеріальної гіпертензії; накопичення у крові NO сприяло розвитку макроангіопатії нижніх кінцівок.

Real-world performance and accuracy of stress echocardiography: the EVAREST observational multi-centre study.

AimsStress echocardiography is widely used to identify obstructive coronary artery disease (CAD). High accuracy is reported in expert hands but is dependent on operator training and image quality. The EVAREST study provides UK-wide data to evaluate real-world performance and accuracy of stress echocardiography.Methods and resultsParticipants undergoing stress echocardiography for CAD were recruited from 31 hospitals. Participants were followed up through health records which underwent expert adjudication. Cardiac outcome was defined as anatomically or functionally significant stenosis on angiography, revascularization, medical management of ischaemia, acute coronary syndrome, or cardiac-related death within 6 months. A total of 5131 patients (55% male) participated with a median age of 65 years (interquartile range 57–74). 72.9% of studies used dobutamine and 68.5% were contrast studies. Inducible ischaemia was present in 19.3% of scans. Sensitivity and specificity for prediction of a cardiac outcome were 95.4% and 96.0%, respectively, with an accuracy of 95.9%. Sub-group analysis revealed high levels of predictive accuracy across a wide range of patient and protocol sub-groups, with the presence of a resting regional wall motion abnormalitiy significantly reducing the performance of both dobutamine (P < 0.01) and exercise (P < 0.05) stress echocardiography. Overall accuracy remained consistently high across all participating hospitals.ConclusionStress echocardiography has high accuracy across UK-based hospitals and thus indicates stress echocardiography is being delivered effectively in real-world practice, reinforcing its role as a first-line investigation in the assessment of patients with stable chest pain.

Predictability And Validity Of External Ear Anthropometry In Sex Inference Among Hausa Population Of Nigeria

Introduction and Objective: Sex determination is one of the vital steps used by forensic experts in reducing matching procedures, easing the burden of investigations in the processes of establishing the identity of an individual. The aim of the present study was to develop and validate models for prediction of sex from external ear dimensions among adult Nigerians of Hausa ethnic lineage. Materials and Methods: The sample comprised of 357 students of Hausa ethnic origin (165 females and 192 males) with age range of 18 and 30 years (21.68 ± 2.53). Ten landmarks were defined on each participant after which eight paired dimensions were measured. Results: Sexual dimorphism in ear dimensions showed significantly higher median value in all the ear dimensions except for left tragus to helix (LT-H). The highest effect size was observed in right ear lobe width (RELW) and the least in LT-H. Four principal components (PCs) extracted explained 80% of the variance of the ear dimensions considered in the study. The sex prediction accuracy of the ear dimensions ranged from 54.51% to 93.73%, and a model with accuracy of 97.3% was developed and validated. From test and training populations, the sensitivity and specificity were found to range from 90% to 100% with AUC ranging from 97 to 100%. Conclusion: A model with high level of prediction accuracy and validation for sex determination from external ear anthropometry was developed. Significance: The performance of the models and cut-off values were appropriate and excellent in the prediction of sex and possible application in the anthropological and forensic case work among adult Nigerians of Hausa ethnic group.

Semi-Double-loop machine learning based CPS approach for predictive maintenance in manufacturing system based on machine status indications

The paper presents two original and innovative contributions: 1) the model of machine learning (ML) based approach for predictive maintenance in manufacturing system based on machine status indications only, and 2) semi-Double-loop machine learning based intelligent Cyber-Physical System (I-CPS) architecture as a higher-level environment for ML based predictive maintenance execution. Considering only the machine status information provides rapid and very low investment-based implementation of an advanced predictive maintenance paradigm, especially important for SMEs. The model is validated in real-life situations, exploring different learning algorithms and strategies for learning maintenance predictive models. The findings show very high level of prediction accuracy.

Predicting Diabetic Neuropathy Risk Level Using Artificial Neural Network and Clinical Parameters of Subjects With Diabetes.

Background:A risk assessment tool has been developed for automated estimation of level of neuropathy based on the clinical characteristics of patients. The smart tool is based on risk factors for diabetic neuropathy, which utilizes vibration perception threshold (VPT) and a set of clinical variables as potential predictors.Methods:Significant risk factors included age, height, weight, urine albumin-to-creatinine ratio, glycated hemoglobin, total cholesterol, and duration of diabetes. The continuous-scale VPT was recorded using a neurothesiometer and classified into three categories based on the clinical thresholds in volts (V): low risk (0-20.99 V), medium risk (21-30.99 V), and high risk (≥31 V).Results:The initial study had shown that by just using patient data (n = 5088) an accuracy of 54% was achievable. Having established the effectiveness of the “classical” method, a special Neural Network based on a Proportional Odds Model was developed, which provided the highest level of prediction accuracy (>70%) using the simulated patient data (n = 4158).Conclusion:In the absence of any assessment devices or trained personnel, it is possible to establish with reasonable accuracy a diagnosis of diabetic neuropathy by means of the clinical parameters of the patient alone.

Chaotic Time Series Prediction Using LSTM with CEEMDAN

Chaotic systems are complex dynamical systems that play a very important role in the study of the atmosphere, aerospace engineering, finance, etc. To improve the accuracy of chaotic time series prediction, this study proposes a hybrid model CEEMDAN-LSTM which combines Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and long short-term memory (LSTM). In the model, the original time series is decomposed into several intrinsic mode functions (IMFs) and a residual component. To reduce the difficulty of predicting chaotic time series and provide a high level of predictive accuracy, the LSTM prediction model is built for all each characteristic series from CEEMDAN deposition. Finally, the final prediction results are obtained by combining all the prediction sequences. To test the effectiveness of this model we proposed, we examined the CEEMDAN-LSTM model using the Lorenz-63 system. Further compared to Autoregressive Integrated Moving Average (ARIMA ), Support Vector Regression (SVR), multilayer perceptron (MLP), and the single LSTM model, the results of the experiment show that the proposed model performs better in the prediction of chaotic time series. Besides, the hybrid model proposed in this paper has better results than the LSTM model alone. Therefore, hybrid models based on deep learning methods and signal decomposition methods have great potential in the field of chaotic time series prediction.

Predicting Alert Source Device using Machine Learning Algorithms

In a large distributed virtualized environment, predicting the alerting source from its text seems to be daunting task. This paper explores the option of using machine learning algorithm to solve this problem. Unfortunately, our training dataset is highly imbalanced. Where 96% of alerting data is reported by 24% of alerting sources. This is the expected dataset in any live distributed virtualized environment, where new version of device will have relatively less alert compared to older devices. Any classification effort with such imbalanced dataset present different set of challenges compared to binary classification. This type of skewed data distribution makes conventional machine learning less effective, especially while predicting the minority device type alerts. Our challenge is to build a robust model which can cope with this imbalanced dataset and achieves relative high level of prediction accuracy. This research work stared with traditional regression and classification algorithms using bag of words model. Then word2vec and doc2vec models are used to represent the words in vector formats, which preserve the sematic meaning of the sentence. With this alerting text with similar message will have same vector form representation. This vectorized alerting text is used with Logistic Regression for model building. This yields better accuracy, but the model is relatively complex and demand more computational resources. Finally, simple neural network is used for this multi-class text classification problem domain by using keras and tensorflow libraries. A simple two layered neural network yielded 99 % accuracy, even though our training dataset was not balanced. This paper goes through the qualitative evaluation of the different machine learning algorithms and their respective result. Finally, two layered deep learning algorithms is selected as final solution, since it takes relatively less resource and time with better accuracy values.

Development of prediction model through linear multiple regression for the prediction and analysis of the GSM of embroidered fabric

PurposeThe purpose of this paper is to establish the regression equation based upon a set of samples prepared through structured design of experiment and form a prediction model for prediction of the areal density gram per square meter (GSM) of the embroidered fabrics and study the influence of basic input parameters.Design/methodology/approachEmbroidery samples are prepared taking input parameters as GSM of the base fabric, linear density of the embroidery thread and stitch density of the embroidery design. Three levels of values are identified for each of the input parameters. Taguchi and Box-Behnken experiment design principles are used to prepare two sets of samples. Linear multiple regression is used to determine the prediction equations based upon each of the two sets and the combined set as well. Prediction equations are statistically verified for the prediction accuracy. Also, surface curves are prepared to study the influence of embroidery parameters on the GSM.FindingsIt is found that all the three prediction models developed in this study can predict with a very satisfactory level of accuracy. However, the regression equation based upon the data set prepared according to Taguchi experiment design is emerged as the prediction model with highest level of prediction accuracy. Corresponding equation coefficients and several three-dimensional surface curves are used to study the influence of embroidery parameters and it is found that the stitch density is the most influential input parameter followed by stitch length and the GSM of base fabric.Research limitations/implicationsThis can be used to assess the GSM of embroidered fabrics before starting the actual embroidery process. So, this model can help the embroidery designers significantly to pre-estimate the GSM of the embroidered fabrics and select the design parameters accordingly. Also, this model can be a useful tool for estimation of thread consumption and thread cost in embroidery.Practical implicationsThe input parameters used here are very basic parameters related to design and materials, which can be easily available. And also, a simple linear multiple regression is used to make the prediction equation simple and easy to use. So, this model can help the embroidery designers or garment designers to select/adjust the embroidery parameters and thread parameters accordingly in the planning and designing stage itself to ensure that the GSM of embroidered fabrics remains within desirable range. Also, this prediction model developed hereby may be a very useful tool for estimation of the consumption and cost of embroidery threads.Originality/valueThis paper presents a very fundamental study to reveal the effect of embroidery parameters on the GSM, through development of regression equations. It can help future researchers in optimizations of input parameters and forming a technical guideline for the embroidery designers for selection of the design parameters for a desired GSM of embroidered fabric.

Predicting Poverty Using Geospatial Data in Thailand

Poverty statistics are conventionally compiled using data from household income and expenditure survey or living standards survey. This study examines an alternative approach in estimating poverty by investigating whether readily available geospatial data can accurately predict the spatial distribution of poverty in Thailand. In particular, geospatial data examined in this study include night light intensity, land cover, vegetation index, land surface temperature, built-up areas, and points of interest. The study also compares the predictive performance of various econometric and machine learning methods such as generalized least squares, neural network, random forest, and support vector regression. Results suggest that intensity of night lights and other variables that approximate population density are highly associated with the proportion of an area’s population who are living in poverty. The random forest technique yielded the highest level of prediction accuracy among the methods considered in this study, perhaps due to its capability to fit complex association structures even with small and medium-sized datasets. Moving forward, additional studies are needed to investigate whether the relationships observed here remain stable over time, and therefore, may be used to approximate the prevalence of poverty for years when household surveys on income and expenditures are not conducted, but data on geospatial correlates of poverty are available.

Multiple steps ahead solar photovoltaic power forecasting based on univariate machine learning models and data re-sampling

Abstract Accurate predictions of solar Photovoltaic (PV) power generation at different time scales are essential for reliable operations of energy management systems. Contemporary methods can accurately predict solar PV power a few steps ahead but fail to maintain high level of prediction accuracy as the number of steps increases. In this paper, we present a simple but effective univariate approach to predict solar PV power output multiple steps ahead. The objective is to maintain high level accuracy of the machine learning algorithms for multiple steps ahead prediction. The novelty of the proposed approach lies in the integration of a data re-sampling technique with machine learning algorithms. The data re-sampling technique enables machine learning algorithms to compute multiple steps ahead predictions via simple single-step ahead prediction on the re-sampled time-series. At each prediction step, the proposed approach first obtains a new representation of the original time series based on the re-sampling process. A separate prediction model is then implemented for single-step prediction on the re-sampled time series. The single-step ahead prediction of re-sampled time series corresponds to m-th step ahead prediction of original time series as determined by the re-sampling factor m. Since the proposed approach computes multiple steps ahead predictions by combining a set of models designed for single-step ahead prediction, it significantly reduces the error at higher prediction steps. We have evaluated the effectiveness of the proposed approach for 5-min to 3-h ahead predictions using 2 years of data from a 1.22 MW PV system in Australia.

Machine Learning Models for Analysis of Vital Signs Dynamics: A Case for Sepsis Onset Prediction.

Objective Achieving accurate prediction of sepsis detection moment based on bedside monitor data in the intensive care unit (ICU). A good clinical outcome is more probable when onset is suspected and treated on time, thus early insight of sepsis onset may save lives and reduce costs. Methodology We present a novel approach for feature extraction, which focuses on the hypothesis that unstable patients are more prone to develop sepsis during ICU stay. These features are used in machine learning algorithms to provide a prediction of a patient's likelihood to develop sepsis during ICU stay, hours before it is diagnosed. Results Five machine learning algorithms were implemented using R software packages. The algorithms were trained and tested with a set of 4 features which represent the variability in vital signs. These algorithms aimed to calculate a patient's probability to become septic within the next 4 hours, based on recordings from the last 8 hours. The best area under the curve (AUC) was achieved with Support Vector Machine (SVM) with radial basis function, which was 88.38%. Conclusions The high level of predictive accuracy along with the simplicity and availability of input variables present great potential if applied in ICUs. Variability of a patient's vital signs proves to be a good indicator of one's chance to become septic during ICU stay.

TransPrise: a novel machine learning approach for eukaryotic promoter prediction.

As interest in genetic resequencing increases, so does the need for effective mathematical, computational, and statistical approaches. One of the difficult problems in genome annotation is determination of precise positions of transcription start sites. In this paper we present TransPrise—an efficient deep learning tool for prediction of positions of eukaryotic transcription start sites. Our pipeline consists of two parts: the binary classifier operates the first, and if a sequence is classified as TSS-containing the regression step follows, where the precise location of TSS is being identified. TransPrise offers significant improvement over existing promoter-prediction methods. To illustrate this, we compared predictions of TransPrise classification and regression models with the TSSPlant approach for the well annotated genome of Oryza sativa. Using a computer equipped with a graphics processing unit, the run time of TransPrise is 250 minutes on a genome of 374 Mb long. The Matthews correlation coefficient value for TransPrise is 0.79, more than two times larger than the 0.31 for TSSPlant classification models. This represents a high level of prediction accuracy. Additionally, the mean absolute error for the regression model is 29.19 nt, allowing for accurate prediction of TSS location. TransPrise was also tested in Homo sapiens, where mean absolute error of the regression model was 47.986 nt. We provide the full basis for the comparison and encourage users to freely access a set of our computational tools to facilitate and streamline their own analyses. The ready-to-use Docker image with all necessary packages, models, code as well as the source code of the TransPrise algorithm are available at (http://compubioverne.group/). The source code is ready to use and customizable to predict TSS in any eukaryotic organism.