Unit
5 Written Assessment
Name
Institution
Affiliation
Introduction
A
formal approach is more suitable to validate the problem. The outlook offers a
valuable tool that helps to reveal omissions, inconsistencies, and ambiguities
in health records. Traditional approaches can't detect uncertainties and
several other issues in health information system. Also, it enhances the
efficacy and applicability of data in any health settings. However, a formal
approach allows and complements exclusive validation procedures. Patient data
efficiently used has excellent capability to improve and even save human lives.
The need for the transparent information system is of great importance to
creating a reliable service design and healthcare delivery system (Sammon,
O'Connor & Leo, 2009).
The
hospital collects demographic data of its patients. It includes analytical
expression of patients' socioeconomic status such as education level, sex,
marital status, religion, age, death rate, income rate, birth rate, family
size, and occupation (Yanamadala et al, 2016). The data is stored in structured
form. Structured information is accessible and readily searchable using
straightforward algorithms. On the other hand, unstructured data is the
opposite of being readily available.
According
to Sun and Reddy (2013), data mining process uses several essential techniques
that define data recovery and mining operation. Techniques such as prediction,
classification, long-term processing,
clustering, association, decision trees, sequential patterns, and
combinations are applicable in both structured and unstructured data sets. For
this assessment, classification and association techniques are suitable to
extract the required patient data. Preferably, I will use the Naïve Bayes (NB)
mining algorithm. It is a supervised algorithm that uses Bayes' Theorem to make
predictions within a data set. By doing so, it facilitates the detection of the
issue from the relevant evidence as presented in the patient data (Sun & Reddy,
2013). To arrive at an evidence-based answer, approximately 75 percent of the
patient need to be successfully extracted.
In
most cases, researchers undermine the data pre-processing step in data
extraction exercise. “Garbage In, Garbage Out” is a significant phrase in data
mining process. Methods used to gather the data need to be controlled to avoid
ambiguities and irrelevant or redundant data combinations such as (Gender:
Male, Pregnant: Yes). Also, to make sure that all essential values are
available to avoid misleading conclusions. Therefore, quality of data and
accurate representation is of importance before analysis. Data pre-processing
improves the quality of the data by ensuring that the data isn't noisy, doesn't
have inconsistencies, and missing values (Sammon, O'Connor & Leo, 2009).
Also, it makes the mining process easy and efficient. This critical step in
data extraction prepares and transforms the initial dataset. Data cleaning
includes the following categories: Data cleaning, Data integration, Data
transformation, and Data Reduction.
The
10-fold cross-validation technique is the best evaluation technique to evaluate
this kind of dataset. It's a method that analyses predictive sets by diving the
initial data sample into a test set and training set. The researcher repeats
the cross-validation procedure for ten times and uses each of the ten sub-sets
once to represent the validation data. Then the ten folds produce an average
that results to a particular estimation (Burrows, 2017). The method is advantageous
because all extracted sets make up the validation and training samples, and
uses each fold once. Also, the samples contain equal proportions. The
supervised approach is recommendable for learning purposes. Considering am a
single mother and as well as a strong Christian, health information records
require transparency. A formal report containing all the results from the
analysis will be the best way to transform and create a transparent and
reliable health sector that aims at improving and saving patients’ lives.
References
Burrows,
S.C. (2017). The Importance of Internships: Ensuring Informatics Students’
Future Success. Journal of Health Informatics & Management 1:2.
Sammon,
D., O’Connor, K.A., and Leo, J. (2009). The Patient Data Analysis Information
System: Addressing Data and Information Quality Issues. Electronic Journal Information Systems Evaluation Volume 12 Issue 1
2009 (95-108).
Sun,
J. and Reddy, C.K. (2013). Big Data Analytics for Healthcare. SIAM
International Conference on Data Mining, Austin, Texas. IBM.
No comments:
Post a Comment