The Use of Data Mining to Determine Ventilator Associated Pneumonia (Vap) - Paper Example

The study investigates the use of data mining to determine Ventilator Associated Pneumonia (VAP) a disease that affects patients with external tube - end tracheal. The study proposes data mining from the Huntington Hospital in New York where there are higher numbers of Pneumonia.

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Order now

Methods

Three data mining procedures was used, which primarily include Logistic Regression, Artificial Neural Networks and Decision Tree. Data from twenty-five patients presented with a HRI code of 1, or 2 was used explored. Each technique has its own unique formation; however, RapidMiner was used on the three. In the report, a parallel theoretical analysis was used to mine data hence enabling the stability of any of the three results.

Results

Data was divided into two primarily groups, which include HRI Code, 1 or 2, which were to be fed on the miner. Results of the three procedures were compared, where the miner proceeded with one methodology. The ANN and DT were presented jointly with calculation and graphical background, while the LR was presented predominantly with assumption. Results show that DT had a higher probability of results where the Data Mining was used to Predict VAP. The probability of results based on DT findings was noted to be 2.57 times greater, whereby patients under external tube - end tracheal had a chance of 257 % to contract VAP.

Conclusions

Data mining presents an approach to determine how ailments emerge in hospitals and even determine how communicable diseases grow in the society. Based on the results from data mining, it is possible to construct a policy to counter the growth of diseases.

2. Introduction

Laws, for instance, Health HIPAA, or PPACA, have overseen the need to protect human from lack of reasonable medical attention, or probably unethical medical response. The first law is more concerned with the insurance of a patient treatment and health records. The second one protects a patient against lack funding of medical bills. With time, the need to a patient's control data and financial statements gained popular attention to hospital administrators. Such an approach has achieved popular attention because nowadays, a patient medical history is frequently recorded, while most patients want to treat their health history as confidential. In addition, doctors want to record patient information as a way to keep progress of the patient information. However, retrieving such information is becoming a daunting task if combined. Data can be recovered for the need of verification attention and assessment.

Data mining is a traditional approach that has gained significant attention to the administrator, if mined; determining the progress of ailments is much easier. So far, early detection and treatment of patients have demonstrated outstanding the importance of record keeping. In any case, preventing life threatening requires determining the results of particular clinical problems.

Other approaches for instance EWS are developed to assess the risk facing patients who have their medical records with individual systems. Data mining helps authorities in charge to retrieve such information and use it to develop policy systems. Hence, data mining appears an emerging field that requires significant attention.

Although there have been efforts to use data mining in such areas, there lacks substantial evidence showing that data mining has been used to predict the progress of VAP. Remarkably, there have been efforts to introduce data mining in Universities as a course, but however it essence in exploring data mining in predicting VAP.

There is a range of techniques used for data mining, however for the purpose of this paper, the three techniques to be used include ANN, LR and DT. However, the three techniques have considerate differences with each other. The goal of the research is to establish the most viable technique used, while using findings from this technique to investigate the use of data mining to predict various hospital related infections.

2.1 The Ventilator-Associated Pneumonia (VAP)

Ventilator-Associated Pneumonia is pneumonia that develops shortly after ventilation is provided using an external tube - end tracheal. Hospital Acquired Pneumonia (HAP) might develop with two days or even after the longer admission. The medical condition affects when

When a patient is hospitalized more than two days.

When a patient is facing breathing difficulties

When a patient is hospitalized near cold positions and is being aided in breathing.

When a patient faces hemodialysis

The medical condition noted, in this case, indicates that there is a need to keep tracking of patient data because it is a communicable disease. Different patients have different respiratory systems altogether.

2.2 Study Area

The study will mean Huntington Hospital in New York. The hospital ranks highly in an adult specialty. It experiences close to 15, 869 admissions annually that are reported close to the 8,682 outpatient surgeries. The hospital ranks highly across 5,000 centers through various clinical specialties. It also ranks highly in the COPD treatment. This means that there is a high rate of adults being treated with breathing complications problems in the hospital. Hence, there is a subsequent high rate of VAP in the hospital. As a result, it is necessary to mine data because such data will help to develop of encountering problems inside the hospital and provide other hospitals with vital information that will lead to minimizing VAP related complications, as a way to boost the fight of other ailments altogether.

2.3 Study Design

All data was used generated from the hospital database. The data is contained in the EWS systems design and is provided through the general wards representation. The primary task will mine data from the central database, retrieve, clean it, integrate it, compile it and analyze it. The framework for the research involves directing significant attention to the irregularity as well as the multi-scaling through measuring the limits and the features space.

2.4 List of Acronyms

HIPAA Health Insurance Portability and Accountability Act

PPACA Patient Protection Affordable Care Act

VAP Ventilator Associated Pneumonia

EWS Early Warning Systems

COPD - Chronic Obstructive Pulmonary Disease

ANN Artificial Neural Networks

LR- Logistic Regression

DT Decision Tree

3. Literature Review

There is concurrent covering the few known literature reviews covering the chosen set of publications. Most publications naturally point to the predictive diagnosis associated with data mining. The release related to predictive diagnosis related to data mining is limited to the few time frame. To the knowledge of this research, there lack no systematic literature reviews showing whether hospitals can use data mining to prevent the hospital sharing of PAV. As noted, data mining is part of the knowledge discovery process, used for various data analysis methods, for instance, statistics, machine learning and artificial intelligence as used to finding different predictions values that are hidden patterns and dependencies. Various arguments show that Data mining applications help in exploring significantly large amount of heterogeneous data in search of consistent information. Assorted empirical learning methods have indicated the need to optimize the prediction of the unseen data through the accurate estimate of the generalization error, creates a paramours importance on the influence of data mining application. Diverse theoretical developments have verified the need for mining in health, as way to determine the patient progress of medical condition. To begin with, data mining exploits clustering methods that arrange significant amounts of evidence into an appropriately structured representation that provides for authenticity in data management. Further relative research from Alamri and Tyler Wood (2016), show that some empirical learning methods aim to optimize the prediction of unseen data while ensuring that there is an original formulation that is based on Vapnik-Chervonenkis dimension. The review was used showing that indeed, there is a practical significance of several various aspects of clustered-based approaches of data mining applications.

3.1 Data Mining a Concept of Theory or Practice

The study by Taras (2010, p. 538) combined different quantified findings that analyzed the results of a tertiary hospital and survey on the practical DM usage. The distribution of scientific publications related to the DM in medicine concurrently. Other findings related to the data mining show that only tertiary hospitals have attempted to install clinical software systems that enable the collection of clinical and demographic patient data that is required for DM applications. The presentation of certain differences between the various interests and possible adoption of the DM in health care facilities shows that hospitals are in the process of either adopting such software or probably using different practical usage and potential possibilities of the DM applications.

3.2 Data Mining

Data mining is defined as part Data Intelligence Group involves the extraction of hidden of predictive information from large databases. Accordingly, data mining tools occur through a series of connected databases where there are hidden patterns and findings show that there is practice information of experts that lies outside through different expectations. Bayati (2010, p. 409) investigates how algorithms are used for data mining, specifically noting on the importance of such algorithms in creating operations for procedures that oversee the use of data mining methods to achieve the required goals. According to much-related evidence, it is clear that the field of data mining has not yet been used because the organization has not adopted the right systems. In any case, hospitals have not used supercomputers.

The use of high-performance computing has been in use in various other departments but not necessary used to retrieve information. The highly calculation-intensive tasks help in problem-solving...