Prediction Techniques In Healthcare Field
Adaptable vital available tools for exploration of data is a basic necessity in today's world. Such tools can have tremendous impact for future medical diagnostic system. Miscellaneous machine learning and data mining techniques such as association rules, decision trees are utilised for clinical decision making. We use various statistical methods such as Bayesian approach, probabilistic approach, logistic approach and regression models to target specific disease outcomes. Survival based techniques deployed to measure the survivability rate of patient is also used. For Example, if the patient is suffering from last stages of breast cancer, what all drugs can be given to increase the survivability among the patient case studied integrated with age and gender.
Time Series Data Mining
The Time series data is also known as temporal data which is studied with respect to time. It is the process of discovery of hidden and inexplicit information from temporal databases. Further, algorithm that processes the temporal data is known as temporal algorithm. Generally it is restrained with analytics for finding patterns and irregularities among the sets of temporal data. Also it allows automatic exploration of hidden facts from data regardless of dimensionality and complexity. Given the type of healthcare databases which include EHR (Electronic Health records) or Sensor data the temporal data mining are utilised in respect to nature of data present. There also exist several temporal data mining pattern methods which include temporal classification, temporal association rules, temporal clustering, temporal prediction and trend analysis. These techniques when implemented can be utilised for discovery of hidden facts and information from big data for future medical diagnosis and prediction.
Visual Analytics for Big-Data
The big data visual analytical provides a platform where multiple data resources are integrated to provide better visualisation of data. It promotes better decision making of the clinical workflow because of the discovery of identified patterns. The visual analytics technique facilitates the multiple integration of interactive interfaces for exploration of complex databases.
Due to swift increase in healthcare databases, the visual analytics can help to build an efficient and effective interpretation of data in visualisation form which can forfeit patients for future medical diagnostics and benefit healthcare databases by providing insights of the data which are not easily interpreted by human capabilities alone.
Hybrid Case Based Reasoning Model
In a Web Prediction Framework for Disease Prediction based on Hybrid Case Based Reasoning Model. The CBR method observes the similarities with the previously solved problem in order to solve the new problem. If it finds any case that is similar to the case that is being looked upon, it will show the result of the previous case. For non satisfactory results, it revises the selected cases and finds a new solution. Then, it adds the revised case to the cases library and, thus, expands its knowledge. The web framework for the CBR model is illustrated in figure. It consists of majorly 4 parts. Data pre-processing is the first layer which does data cleaning, feature extraction, transformation, etc, thus, making the data suitable for the data mining process. If there are no irregularities in the input data, then the framework ranks the features by using techniques like Information Gain, Gain Ratio and Correlation-based Feature Selection. After this, the data clustering module which uses k-means technique. K-means is a cluster solving algorithm that tries to minimise the sum of the squared error over all K clusters.
