حيدر خضير عبد الله الفتلاوي

.

.

Abstract Aim: Management of healthcare’s resources contributes to improving the quality of medical services, thereby enhancing the level of health of society in general. This management requires providing prospective information about the need for patients admitting in a hospital, and the necessary medical resources.Prediction techniques represent an effective tool for knowledge discovery in huge and complex datasets in many fields including healthcare. Methods: In this work, we design and implement a prediction model called a Modern Prediction Model for HealthCare Problem (MPM-HCP) which introduces two improvements for Gradient Boosting Machine (GBM) prediction technique. MPM-HCP developed (GBM) by inspiring positive sides of linear regression to replace splitting criterion with a correlation measure in regression tree building. It also reduced the complexity of building boosted model by using a fast method for choosing best split point. Results: MPM-HCP has significant behavior in terms of prediction error and execution time. In comparison with tradition gradient boosting trees, the MPM-HCP has a testing error of 0.468,while original GBM based on sum of squared has error of0.491, and original GBM based on standard deviations has 0.481 error. Training time is also reduced more than 85%. Conclusions: MPM-HCP implementation showed that there were three attributes frequent in binary regression trees building. Those attributes were gender of patient, number of claims to admit hospital, and the medical procedure group, which means those attributes is more correlated with the target of prediction (i.e. number of hospitalization days). MPM-HCP confirms the ability to produce precise prediction result, and the scalability to deal with huge dataset in suitable execution time

Prediction techniques represent an effective tool for knowledge discovery in huge and complex dataset in many fields including healthcare. The problem of healthcare is managing available medical resources and preparing plans for the future needs aiming to enhance medical services. This research provides an integrated prediction model to solve the problem above by analyzing medical data records and predicating the duration of future patient‘s hospitalization. The proposed model consists of three major stages; starting with preprocessing the data; applying prediction algorithm; and ending with evaluating the model based on real data. Our model used Gradient Boosting Machine (GBM) algorithm which reduce training error by building a sequence of decision trees. GBM is characterized by updating values of target feature after the construction of each decision tree. In this research, we tried to discover the effect of reducing the update process on terminal nodes that have lowest percentage of error, the results showed the ineffectiveness of reduction compared to the original. The research tried to determine the best measure for choosing splitter feature during the building of the decision tree, and the results showed that standard deviation is better than mean. The research also studied effect of changing values of GBM algorithm parameters in behavior of training process.

Abstract With the arrival of big data and cloud computing as a computing concept, it is becoming ever more critical to efficiently choose the most optimum machine on which to execute a program, for example in the healthcare environment. This process of choice is also complicated by the fact that numerous machines are available as virtual machines. Hence, predicting the most optimum choice of machine based on a target application is a challenge. Prediction techniques consume large amount of computing resources when operating with multi-dimensional data that can cause long delays compounded by cross validation process in evaluating and choosing the most optimum prediction model. We propose a model of prediction techniques to predict and classify some of the health datasets to retrieve useful knowledge to illustrate how a data miner can choose a suitable machine especially in cloud environment with good accuracy in a timely manner. Our results show that the execution time has an inverse relation with the use of resources of a machine and the accuracy of prediction could be different from one machine to another using the same predicting technique and dataset.

Abstract

برز مصطلح علم البيانات موخرا كمظلة واسعة لعدد من التخصصات المتداخلة مثل الذكاء الاصطناعي و التنقيب في البيانات و معالجة اللغات الطبيعية و علم الاحصاء. السبب الرئيسي لظهور هذا المصطلح هو حاجة الباحث في مجال تحليل البيانات و استخراج المعرفة منها الى الالمام بعدد من المفاهيم الرياضية و الاحصائية و طرق تعلم الماكنة. الازدياد المضطرد للبيانات و تنوع مصادر توليدها يعد من اهم التحديات التي تواجه الباحثين في هذا المجال. حيث يبلغ حجم البيانات التقريبي 20 اكزابايت و يتوقع وصوله الى 40 اكزابايت في 2020. متحسسات الحركة و الاضاءة و الحرارة و اجهزة الشبكة و ملفات الاستخدام log files تشكل جزء كبير من مصادر التنوع لهذا البيانات. اضافة الى ذلك فان سرعة توليد و وصول البيانات تشكل تحديا اخر خصوصا في مجال تحليل تدفق البيانات Data stream analysis ، حيث يتطلب التعامل مع بيانات تصل تباعا و لا يمكن تطبيق الكثير من المباديء الاحصائية عليها مثلا حساب اعلى القيم تكرارا او المعدل . تم النشر بتاريخ 12/5/2018