When we have a huge data set on which we would like to perform predictive analysis or pattern recognition, machine learning is the way to go. Machine Learning (ML) is the fastest rising arena in computer science, and health informatics is of extreme challenge. The aim of Machine Learning is to develop algorithms which can learn and progress over time and can be used for predictions. Machine Learning practices are widely used in various fields and primarily health care industry has been benefitted a lot through machine learning prediction techniques. It offers a variety of alerting and risk management decision support tools, targeted at improving patients' safety and healthcare quality. With the need to reduce healthcare costs and the movement towards personalized healthcare, the healthcare industry faces challenges in the essential areas like, electronic record management, data integration, and computer aided diagnoses and disease predictions. Machine Learning offers a wide range of tools, techniques, and frameworks to address these challenges. This paper depicts the study on various prediction techniques and tools for Machine Learning in practice. A glimpse on the applications of Machine Learning in various domains are also discussed here by highlighting on its prominence role in health care industry.

Predictive analytics in health care using machine learning tools and techniques

Data preprocessing for heart disease classification: A systematic literature review.

The heart disease has been one of the major causes of death worldwide. The heart disease diagnosis has been expensive nowadays, thus it is necessary to predict the risk of getting heart disease with selected features. The feature selection methods could be used as valuable techniques to reduce the cost of diagnosis by selecting the important attributes. The objectives of this study are to predict the classification model, and to know which selected features play a key role in the prediction of heart disease by using Cleveland and statlog project heart datasets. The accuracy of random forest algorithm both in classification and feature selection model has been observed to be 90â€“95% based on three different percentage splits. The 8 and 6 selected features seem to be the minimum feature requirements to build a better performance model. Whereby, further dropping of the 8 or 6 selected features may not lead to better performance for the prediction model.

/pdf/classification-and-feature-selection-approaches-by-machine-4nlhgf29k2.pdf

Classification and Feature Selection Approaches by Machine Learning Techniques: Heart Disease Prediction

In classification and prediction of different types of medical disorders the neuro-fuzzy systems (NFS) are playing vital and significant role. To avoid false diagnosis the NFS assists medical practitioners to a greater extent in automating the domain dealing with medical disorders. With the passage of time the NFS approach has become apparent to enhance accuracy in dealing with a wide range of complicated research problems in the field of medical diagnosis. In this paper the author presents the literature review of the research done in implementing NFS in the field of medical diagnosis for current decade. Total of 100 publications in chronological advancement and up-gradation in models are considered for the time period of 10 years. A detailed study of each disease is carried out to discuss how NFS methodologies have been applied for classification and prediction in the diagnosis of different types of medical disorders. Ten (10) most severe medical disorders i.e. cancer, cardiovascular, depression and anxiety, diabetes, communicable, kidney, liver, neuro-degenerative, respiratory and thyroid has been undertaken for the study. Based on the study carried out it has been observed that NFS found to be effective as compared to the application of other AI techniques in medical diagnosis. Study reveals that effectiveness of NFS increases significantly when integrated with other AI approaches. This review adds into the knowledge of different researchers working in the field of medical diagnosis and will also give the comprehensive view of the effectiveness of the NFS techniques being used in medical diagnosis. The paper also incorporates a few research publications that were submitted in 2019 to incorporate the latest advances in medical science implementation of NFS.

Usage and implementation of neuro-fuzzy systems for classification and prediction in the diagnosis of different types of medical disorders: a decade review

Transit-time flow measurement (TTFM) is a reliable method to check the graft function intraoperatively in coronary surgery. The given parameters are: Mean Graft Flow (MGF); Pulsatility Index (PI) and Insufficiency Ratio (%BF). Some cutoffs of these parameters have been identified as predictors for unfair 1-y clinical outcome: mean graft flow (MGF) less than 20 ml/min and high pulsatility index greater than 5. Other cutoffs have been found as related to postoperative angiography: MGF 15 ml/min or less and pulsatility index at least 3 (sensitivity 94%; specificity 61%); MGF less than 15 ml/min and pulsatility index greater than 3 for left coronary artery or pulsatility index greater than 5 for right coronary artery (sensitivity 96%; specificity 77%); MGF 15 ml/min or less and pulsatility index at least 5.1 left coronary artery (sensitivity 98%; specificity 26%). Hence, with the need to improve the diagnostic accuracy of TTFM, high-resolution epicardic coronary ultrasound module has been added to graft flow evaluation providing 2D ultrasound imaging (either in short-axis or long-axis) and color-flow mapping, allowing an accurate morphological evaluation of body graft and anastomosis. An intraoperative method aimed to verify coronary grafts should be easy to handle, not time consuming, minimally invasive, easily meaningful and relatively cheap; in addition, it should offer objective parameters more than qualitative criteria. We herein report the results of our experience with intraoperative graft verification with TTFM and high-resolution imaging along with a systematic review of the literature in this field with the aim to provide a road map to be followed.

Intraoperative graft verification in coronary surgery.

The idea of medical data mining is to extract hidden knowledge in medical field using data mining techniques. One of the positive aspects is to discover the important patterns. It is possible to identify patterns even if we do not have fully understood the casual mechanisms behind those patterns. In this case, data mining prepares the ability of research and discovery that may not have been evident. This paper analyzes the effectiveness of SVM, the most popular classification techniques in classifying medical datasets. This paper analyses the performance of the Naive Bayes classifier, RBF network and SVM Classifier. The performance of predictive model is analysed with different medical datasets in predicting diseases is recorded and compared. The datasets were of binary class and each dataset had different number of attributes. The datasets include heart datasets, cancer and diabetes datasets. It is observed that SVM classifier produces better percentage of accuracy in classification. The work has been implemented in WEKA environment and obtained results show that SVM is the most robust and effective classifier for medical data sets.

/pdf/effectiveness-of-support-vector-machines-in-medical-data-21gdle4lbm.pdf

L Heena

Papers

Effectiveness of Support Vector Machines in Medical Data mining