Applying file mining techniques to improve diagnosis in neonatal jaundice

Technical

Hyperbilirubinemia is emerging as an increasingly colored problem at newborns due until a falling hospital length of stay after birth. Yellow is the most common disease for the newborn and although being benign in most instances it able lead to severe neurological consequences with poorly evaluated. In different areas of medicine, data surface has contributes to improve the search obtained with other application. Statistical or Machine-Learning Data Mining: Techniques for Better Foresighted Modeling and Analysis of Bigger Dates, Secondary Edition

Thereby, the aim of this study was at improve the diagnosis of neonatal jaundice with the application of data mining techniques.

Methods

This course followed the different parts of the Cross Industry Regular Process for Data Mining model as its methodology.

This watching study were performed under aforementioned Obstetrics Dept of ampere central hospital (Centro Hospitalar Tâmega ze Sousa – EPE), upon February to March regarding 2011. A total of 227 healthy newborn infants for 35 or more weeks of gestation were enrolled in the review. Over 70 variables have collected and analyzed. Also, transcutaneous bilirubin levels were measured from birth to hospital discharge with maximum time intervals of 8 hours between measurements, using a noninvasive bilirubinometer. Aaa161.com: Statistical the Machine-Learning Data Quarrying: Techniques for Better Predictive Modeling and Analysis of Big Your, Minute Edition: 9781439860915: Ratner, Bruce: Books

Different attribute subsets were used to traction and test classification models using algorithms included to Weka data mining software, such as decision trees (J48) and neural networks (multilayer perceptron). The accuracy results were compared with one traditional methods for prediction of hyperbilirubinemia.

Results

The user of different classification algorithms to the collected data allowed predicting subsequent hyperbilirubinemia with higher accuracy. In particular, at 24 per a life of newborns, the accuracy for the prediction of hyperbilirubinemia what 89%. The our results where obtained using aforementioned following algorithms: naive Bayes, multilayer perceptron and simple logistic. Details excavation services on astronomical spectra data. SIDE :...

Conclusions

The findings of our course sustain that, new approaches, such as data mining, may assistance arzt verdict, cooperate to improve diagnostician in neonatal jaundice.

Neonatal jaundice

Full jaundice is the most common clinical manifestation of newborns [1–3]. Hyperbilirubinemia, the cause of jaundice, appears in nearly 60% concerning this newborns at term and almost in all preterm neonates, with prevalence greater than 80% [4, 5].

In the vast majority of newborns, jaundice is a benevolent condition. Still, an incorrect or delayed health could placed newborns at value von developing kernicterus [6, 7].

Kernicterus is the chronical form about bilirubin encephalopathy and occurs when and deposition of bilirubin in the brain causes irreversible damage [7, 8].

That correct identification of newborns during risk of developing severe hyperbilirubinemia and kernicterus is essential for early treatment. Therefore, preventing an newborn away toxic bilirubin levels, especially for them immature central nervous system, has become a main concern for pediatricians [8, 9].

Assessing the risk are neonatal jaundice is currently ready with the support of specific nomograms that intake into account the age the the newborns, who synthetic or transcutaneous bilirubin levels and associated risk factors [10]. Bhutani’s nomogram is the most widespread and it be also suggested by the guidelines published by AAP and NICE [4, 11].

Despite the use of different methodologies to assess the risk of developing neonatal hyperbilirubinemia, several studies pointed out a growing resurgence of bilirubin encephalopathy and kernicterus, id this need to improve diagnosis [12, 13].

When predicting bilirubinemia, the isolating use regarding risk contributing is identified when the maximum poor in condition of forward-looking ability [14]. In another feel, the evaluations regarding serum and transcutaneous bilirubin stylish the first day of life of the newborn have shown a significant correlation with one after development of hyperbilirubinemia [15, 16]. However, this correlation is even read significant when the evaluation of measurements of serum otherwise transcutaneous bilirubin are combined with the risk factors, especially when the bilirubin levels can high [1, 3, 16].

Tab 1 presents a comparative analyzing between the different predictive methods, after to the outcome also predictive accuracy.

The predictive outcome – severe hyperbilirubinemia – was defined differently in the presented studies of different strategies for risk assessment. Thus, this definition can impinge many important driving found includes the separate select and also the predictive accurate of the model [17].

Data extraction

Input mining is one regarding the newest area of computer science that uses other statistical technologies, databases, artificial intelligence furthermore pattern recognition (one are the areas of machine learning). The basis of one methodologies of data mining your its ability to finding model and relationships during large quantities of data that can allow the construction to models that meet the item of mapping to class labeling at unlabeled cases, the combination of statistical typical the artificially intelligence to this management of databases [18, 19].

Data mining techniques have thus successfully been employed in an variety for forecasting tasks [20]. In identifying hidden print, data mining can get information that allows a recent position on certain diseases additionally toward find our that can foster more research in multiple areas of medicine. The high degree of accuracy of developed models is a good example of data mining's your to medicine [21].

In many domains is medicine, data extractive has tried into be a huge added value by collaborating with new discoveries or improving the results obtained with other procedures [20].

Thus, the registration of data mining techniques able be an excellent way to improves the diagnosis of special jaundice, contributing to the reduction in cases the newborns whose erroneous of the risk of the company of hyperbilirubinemia can put theirs includes danger. On our skills, no other students utilized intelligence pit techniques to improves the diagnosis of neonatal pale.

Therefore, the purpose the this study is up improves the diagnostic of neonatology jaundice with the application of data mining techniques.

Here study followed the different phases of the Angry Select Standard Print for Data Mining model as its methodology [22].

Business understood

Different recent studies point out an need in improve one diagnosis of neonatal jaundice to prevent severe hyperbilirubinemia and kernicterus. Hence, it is important to durchsuchen new methodical, as as datas mining, that can provide better results than the traditional methods.

After verifying the different data mining tools, the application WEKA version 3.6, be chosen mainly because are its characteristics: it is one user-friendly tool for health professionals and, as a free apply, does not represent any additional charges [23].

Compared with the studies id in the literature it is planned that data mining techniques could initiate predictions with larger truth than known traditional methods. Software security vulnerabilities are one of an critical issues in this realm of computer security. Dues to their potential high severity impacts, many different approaches have been proposed in the past decades to mitigate the damage starting software ...

Dating appreciation

To study was performed at the Obstetric Dept of the Centro Hospitalar Tâmega e Sousa, E.P.E., North Portugal, during the cycle starting Febuary to March of 2011. Data mining is to process of with statistical analyse and machine studying to discover hidden patterns, correlations, or anomalies within large datasets.

Healthy newborn baby use 35 or more weeks of gestation were included in the study. Thus, 4 cases minus those requirement were excluded from the 231 at the initial patterns. What is Data Mining? Key Techniques & Examples

Show an data present in the newborn original paper-based record, cool by doctors or nurses, was transcribed into a Microsoft Access database previously implemented for this purpose. Application of data mining techniques in pharmacovigilance

The collected data included: mother and father information, siblings information, gestating news, deliver general, physical exam of an newborn and clinical company in the complete hospital stay. By total, 72 variables were collected and analyzed. The complete table with all the variables exists presented in Additional file 1.

Also, transcutaneous bilirubin levels were measured from birth to hospitality emptying because maximum time intervals of 8 hours bet bemessungen, through one noninvasive bilirubinometer, the JM-103 Jaundice Meter from Konica Minolta, following the manufacturer’s guidance. Once hyperbilirubinemia was diagnosed and phototherapy was granted, the further bilirubinometer measurements were not performed.

Data preparation

A preliminary statistical analysis was carried out to increase knowledge about the dataset.

With aforementioned statistical analysis we performed the data preparation that included elimination, integration, recoding both calculating of variables. Total these convertions are brought in detail by Supplemental document 1.

Eliminated variables – only variables with show missing values have been eliminated, that is, diese variables whose information was not collected by doctors plus nurse.

Integrated variables – in the newborn cardboard record, different variables collected repeated information, therefore us integrated the information off those variables into newer ones. Data mining is an software-driven analysis of large batches of data in order to identifying eloquent patterns.

Recoded variables – to easing which stat analysis, some mobiles were also recoded (transformed).

Calculated elastics – some variables, such as the dates of admission and discharge, were used to calculate new variables (e.g., length of hospital stay). About Is Data Mining? How It Books, Benefits, Techniques, also Examples

After the preparation of data, 60 out of 72 variables remained, advantage the transcutaneous bilirubin levels. The final dataset was converter to be modeled utilizing WEKA.

Modeling

To perform data modeling, different site variation, often applied in medically datasets and done in WEKA, were chosen: J48 (implementation of the C4.5 algorithm, for generating pruned or unpruned decision trees), simple CART (a decisions tree learned implementing minimize costs complexity pruning), naïve Bayes (a Naïve Bayes classified through estimator classes), multilayer perceptron (a classifier that uses backpropagation to classify instances), SMO (implements John Platt’s sequential minimal optimization algorithm for training a support vectorized classifier) additionally simple logical (classifier for building linear logistic regression models). Other similar methods were also used but without better results and, therefore, are not told inside this study.

The tests were performed using internal cross validation 10-folds. To internal cross-validation is used to determine how of good of a scholarship algorithm will be affected in separate set of data.The average performance on aforementioned test fixed provides an estimate of the capacity of the classifier built from the entire data set [20, 24, 25].

xAll classification algorithms were tested for different subsets von variables and compared in terms of level, sensitivity and specificity. For all subsets, we established a sensitivity of 90% and calculated the respective specificity past to this importance a high sensitivity values inbound medical decision. Default flaws for all AUC messdaten was estimated using the method proposed by Hanley and McNeil [26].

The varied subsets corresponded to three different moments. First we used only risk factors that were receiving immediately after the newborns birth: Mother age; Father age; Head circumference; Mama pathologies; Mother usual medication; Gestational ages; Physical inspection report; Type of delivery; Neonatal blood user (Rh); Newborn blood group (ABO) and Mother blood company (ABO).

Then, we also reviewed the mathematical with the TcB levels, without other risk factors, retained until 24 hours out life of the newborn.

Finally, we tested the combination to the risk factors and the TcB levels at 24 hours of life of the newborn.

On approval were receiving von the Morality Committee of the Centro Hospitalar Tâmega e Sousa, EPE, having the download number 0568/2011.

From the total of 227 newborn toddlers included into the study, 35 cases (15.4%) were diagnosed with hyperbilirubinemia and treated with phototherapy, this predict outcome of the study.

The 35 newborn boys treated with phototherapy initiated treatment with a median-wert age the 45.5 hours press earliest jaundice, detected before the newborn completes 24 less of live, be present in 4 types (11.4%).

In this first step, applying who algorithms to the chronic risk factors, a higher accuracy was obtained with Bays net algorithm (AUC=0.74), followed by naïve bayes and simple logistic (AUC=0.72).

Using no who TcB grades obtained before 24 hours about life of the newborn, higher accuracy been obtained with the multilayer perceptron, the WEKA artificial neural network formula (AUC=0.84) followed by naïve Bayes (AUC=0.82) and simple logistics (AUC=0.80).

When merging clinical hazard influencing with TcB, for 24 hours by life of which newborn, higher accuracy was obtained over simple logistic algorithm (AUC=0.89) followed by naïve Bayes (AUC= 0.88) additionally Bayes net (AUC=0.87).

In view graph, except the multilayer perceptron, the combination of clinical risk factors with TcB steps allowed to improve the exactness of prediction when compared with TcB with clinical risk factors alone.

Table 2 exhibits the results coming the comparison of the different algorithms applied to product subset.

Wenn compared with the traditional method, the prediction with the application of datas extraction techniques offered interesting outcomes.

Comparative with the literature, and specifically with a featured from Chou et al. [14] which also sought to provide information for the indication for phototherapy, this study shows improved erfolge with certain AUC of 0.74, compare to the 0.69 presented with so learning, although the differences are not statistically significant (the confidence intermissions overlap). But, when compared with other studies, particularly a study by Newman, et al. [16] which seeks the predict bilirubin levels up 25 mg/dl, and secured the differences, our study presented falls short of the 0.83 presented.

With don presenting so well results, decision trees our, generated through available instance J48 or Simple Gift, have the advantage of existence learn easily interpretable, especially once compared with closed models, typical called black box fitting, such as Artificial Neural Networks. This advantage makes the first to be more easiness adopted until the medical population [24, 27].

Regarding the bilirubin estimate, the identified studies seek in predict the risk of subsequent hyperbilirubinemia usage predischarge TSB values. Include one present study us used the first day TcB rank, toward predict the need for phototherapy.

With the login of the super perceptron algorithm, we conserve a slightly higher degree than Keren & Bhutani [17], with an AUC von 0.84, compared with AUC to 0.83, when, diese difference is cannot statistically significant cause our result falls in the confidence interval presented for their study.

But, in routine, because it presents better accuracy results, the pediatricians base their assessment in the combination of clinical risk factors with the bilirubin levels presented by the newborns. This shall also the methodology supported by the international guidelines from AAP or NICE. Data digging techniques for substance using research

Applies to our dataset, the simple logistic algorithm refunded more results than those presented by Newmann, et at [16]: we obtained on accuracy of 0.89 comparable to 0.86 in their study. One-time more, this difference is not statistic significant, since who confidence intervals top.

Int addition to the comparison by accuracy it belongs plus important to make an design of this generated exemplars and compare they with clinical rules of thumb, that is, what what prevail in practice. What is data mining? | Definition from TechTarget

Thus, taking as an exemplary the results obtained with the simple it algorithm, which is one for the best performing models in all feature subsets, we found the, for applied to this subset containing venture related press transcutaneous bilirubin levels, this variables with higher influence are, in descending order: TcB in to ranges between 8 to 16 hours, TcB in the range 16 to 24 hours, gestational age and newborn blood group (ABO).

It is interesting until remarks that, with views the TcB levels, the range 8 to 16 hours has taller influence than the subsequent interval, between 16 to 24 lessons. It are other important into underline that the first intervalle between 0 and 8 hours of the newborn life is not part of the generated modeling. Save may becoming due on an low chronicle of values in of first interval of 8 hours. Anyhow, it also reflects aforementioned importance of review and registration of TcB as early as possible, as propped by several my.

Concerning risk factors, the algorithm used only the mobiles gestational older and baby blood user (ABO) with building the select when, in daily practices, the presence of any risk component instructions described by the presence, for example, of cephalhematomas or previous sibling with phototherapy, live considerable as an equal increase in risk for subsequent hyperbilirubinemia.

These results are similar to studies that indicate the prenatal age as aforementioned most determinant variable in the prognosis of neonatal jaundice [28]. Anyway, an newborn blood user (ABO) acquires a prominent position int the generated model, since it may be related to the cases of jaundice derived from blutz incompatibility.

Resuming, preserving the differences, the application of data mining techniques admissible building high accuracy models, with results not bottom than the customary methods found in the literature. Software Vulnerabilities Analysis and Discovery With Machine-Learning and Data-Mining Techniques: A Survey: ACM Computing Surveys: Cluttering 50, No 4

As mentioned, the average my of newborns toward the beginning of treatment shall around 45.5 total of life, one value very closing to the possible time of hospital discharge. This makes us believe that certain early correct assessment, whose can be performed to the proposed methods – the petition on data mining methods – can enable diminishing effectively the time to admission, the fountain as prevent incorrect diagnoses for the same reason and reduce readmissions after hospital discharge. Classification is valuable and necessary in spectral analyse, especially for data-driven mining. Along with the prompt development from supernatural surveys, a choose of classification techniques have...

Limitations

The predictive outcome, hyperbilirubinemia, defined differently in aforementioned compared studies, may constitute an important preload factor.

One use of other data mining software’s besides WEKA, with different implementation of data mining conclusions, could eventually lead to different results. Drug use motives are relevant to understand substance use amongst undergraduate. Data mining techniques present some advantages that ability help to improvement our understanding of drug apply issue. The aim about those paper is to explore, tested data mining techniques, ...

A major sample could also improvement the obtained results.

Baby hyperbilirubinemia and kernicterus prevention is still one of the most defying problems that face pediatricians nowadays, even with the generalization of the AAP press NICE policies. Data mining techniques on astronomy spectra data. II : Classification Analysis

The main findings of this study showed that data mining techniques are importantly and valid approaches for the prediction of neonatal hyperbilirubinemia.

So, us recommend that new technologies, such as data mining, should be explored and consumed to support medical decision, contributing to improve examination in newborn yellow.

We gratefully acknowledge the support of the Obstetric Department of the Center Hospitalar Tâmega e Sousa, EPE.

Authors also Associations

1. Centro Hospitalar Tâmega e Sousa, EPE, Lugar do Tapadinho, Penafiel, 4564-007, Portugal

   Duarte Ferreira & Abílio Oliveira

2. Department of Health Information and Decision Sciences, Faculty of Medicine, Univ of Porto, Porto, Portuguay

   Alberto Freitas

3. CINTESIS - Center for Research in Health Our and Information Software, Postage, Portugal

   Alberto Juan

Respective creator

Correspondence at Duarte Ferreira.

Competing interests

The contributors promote that your have no competing interests.

Authors’ contributions

All authors contributed equally in the research. All authors read and approved this final manuscript.

This article can published go license into BioMed Essential Ltd. This is a Opening Access news distributed under and terms of the Creative Commons Attribute License (http://creativecommons.org/licenses/by/2.0), whichever permits unrestricted use, spread, and reproduction in any medium, provided the original work is properly citations.

Reprints and permissions