The integration of big data analytics into healthcare has transformed the way patient care is delivered, particularly in
predictive nursing and early disease detection. Traditionally, disease identification in healthcare relied on manual
patient monitoring, standardized diagnostic procedures, and physicians’ experience [1, 5]. However, these methods
often lead to delays in diagnosis, increased hospital readmissions, and inefficient allocation of medical resources [2, 3].
The explosion of big data in healthcare has created an opportunity to leverage advanced computational techniques such
as machine learning (ML) and artificial intelligence (AI) to improve the accuracy and efficiency of disease detection [4,
17, 19]. Big data in healthcare refers to large, complex datasets collected from various sources [7, 9]. These sources are
outlined in the table below:
Table 1 The Sources of data collection and the type of data collected
S/n Data collection sources
Data collected
1
Electronic
(EHRs)
Health
Records
Comprehensive patient histories, clinical notes, laboratory test
results, and medication prescriptions [10,21].
* Corresponding author: David Thomas Omoregie
Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0.
2
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
Wearable and IoT Devices
Real-time monitoring data from smart watches, ECG monitors,
and glucose sensors [1, 4, 22].
3
Genomic Data
DNA sequencing information for precision medicine and
disease risk assessment [8, 13, 21].
4
Medical Imaging
Data from X-rays, MRIs, CT scans, and ultrasounds analyzed
using AI-powered tools [8, 14].
5
Social Determinants of Health
(SDoH)
External
factors
such as socioeconomic conditions,
environmental exposure, and lifestyle habits [8].
By utilizing big data analytics, predictive nursing allows for early disease detection, risk stratification, and proactive
patient management [13]. Machine learning models can process large datasets to identify subtle patterns, correlations,
and risk factors associated with various diseases [14]. Incorporating predictive analytics into nursing care enhances
clinical decision-making, improves patient safety, and optimizes healthcare costs [1, 18]. For example, AI-driven
predictive models have demonstrated the ability to detect sepsis hours before symptoms become critical, thereby
reducing mortality rates and improving recovery outcomes [16, 19]. This study aims to explore how big data analytics
and machine learning can be leveraged in predictive nursing to enhance early disease detection and intervention.
1.1.1. Definition of Predictive Nursing
Predictive nursing involves using data-driven models to forecast health conditions, detect early disease symptoms, and
improve patient outcomes [3, 5] . It shifts the focus from reactive to proactive care, enabling nurses to intervene before
conditions worsen [4].
Predictive analytics helps identify high-risk patients, optimize resource allocation, and reduce hospital readmissions.
One notable area of improvement is sepsis prediction. AI models analyze real-time vitals to detect early sepsis onset,
reducing mortality rates [12, 19, 21]. It is also key in assessment of risk of diabetes. ML models predict pre-diabetes
progression, allowing for timely lifestyle modifications [10]. It also seeks to improve fall detection and prevention of
such occurrences. AI-powered smart beds and motion sensors prevent falls in elderly patients [19]. Predictive tools also
help to identify patients at risk of readmission, improving discharge planning.
1.1.2. Overview of Machine Learning Techniques in Predictive nursing
Machine learning (ML) techniques enable computers to learn from data, identify patterns, and make predictions. In
healthcare, ML models process large patient datasets to detect disease trends and optimize clinical decisions [12]. This
is done using supervised learning algorithms. Supervised learning uses labeled datasets to train models for classification
and regression tasks. Common ML models include: Logistic Regression, Decision Trees & Random Forests and Support
Vector Machines (SVMs) [14, 17].
1.2. Research Objectives
Early detection of diseases is critical in healthcare, as it significantly improves patient survival rates, reduces
complications, and enhances the efficiency of treatment [1]. However, conventional diagnostic methods have several
limitations such as delayed diagnosis, overburdened healthcare systems and lack of personalized risk assessment.
Big data analytics and machine learning offer data-driven approaches to address these challenges by identifying high
risk patients [4, 6]. This involves using AI models to analyze thousands of patient records to detect subtle precursors of
diseases that may not be noticeable through traditional methods. ML algorithms continuously learn from new patient
data, improving predictive precision and reducing misdiagnosis rates. By analyzing historical patient data and lifestyle
patterns, predictive analytics enables personalized treatment plans and targeted interventions.
This study is guided by the following objectives:
• To examine the role of big data analytics in enhancing predictive nursing.
• To identify and evaluate machine learning models used for early disease detection in healthcare.
• To analyze the effectiveness of predictive analytics in improving nursing interventions and patient outcomes.
• To assess the ethical, privacy, and implementation challenges associated with big data in predictive nursing.
• To provide recommendations for integrating big data analytics and machine learning into clinical nursing
practice.
1053
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
1.3. Significance of the Study
The findings of this research will have significant implications for healthcare professionals, policymakers, and
technology developers.
Table 2 Key Areas of focus and how each area is affected
S/n Key areas of
focus
How each area is affected
1
Contribution to
Nursing
Practice
Enhances clinical decision-making by equipping nurses with AI-powered predictive
tools.
Enables early identification of disease progression, reducing hospitalization rates and
mortality.
Improves nurse efficiency and workload management, allowing for more proactive
patient care.
2
Contribution to
Healthcare
Systems
Optimizes resource allocation by identifying high-risk patients who require
immediate attention.
Reduces healthcare costs by preventing late-stage disease treatment expenses.
Enhances hospital management and emergency response preparedness.
3
Contribution to
AI
and Data
Science
Advances machine learning applications in clinical practice.
Highlights the need for explainable AI (XAI) models in healthcare to ensure
transparency.
Promotes the integration of federated learning for decentralized, privacy-preserving
data sharing.
1.4. Challenges of Big Data in Nursing
There are many challenges involved in using big data in the nursing field. These challenges must be overcome for a
successful implementation of this study. These challenges include the following:
• Data Integration Issues: Different hospitals and healthcare providers use varied data formats and storage
systems, making integration difficult [14]
• Privacy and Security Concerns: Sensitive patient information is vulnerable to cyber threats and unauthorized
access [12].
• Algorithmic Bias: ML models trained on biased datasets may produce discriminatory healthcare predictions
[20]
1.5. Research Questions
This study seeks to answer the following research questions.
1054
Figure 1 Research Questions
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
2. Materials and Methodology
The paper analyses the effectiveness of machine learning models in early disease detection. The approach used to get
the findings for this work is a data driven one such that large-scale healthcare data that is publicly sourced is used. The
models analysed are then evaluated based on some key performance index (KPI). The results are then compared with
traditional methods of predicting diseases and the effectiveness determined.
2.1. Types Of Data Collected
Datasets is usually collected from publicly available sources such as MIMIC-III for critical care data (Johnson et al., 2016).
The collection of this data is one that must be carried out in line with ethical considerations so as not to trample upon
and misuse information that is gotten. The types of data that is collected is shown in the figure below:
Figure 2 The types of data collected
2.2. Data quality, preprocessing and model selection
In order to be able to use the data in machine learning models, it must be preprocessed and cleaned up. This means
missing values and inconsistency must be addressed using ethical standards. There is also a need for normalization and
transformation of the dataset. This means that all variables contribute equally to the analysis. This is particularly true
when using distance-based machine learning algorithms.
The study will experiment with several machine learning models to determine the most effective approach for
predictive nursing. The figure below outlines the models that were analysed in the course of the project. Each of them
are key in their identification of set parameters. For example, the logistic regression is for binary classification tasks
such as disease presence or absence.
1055
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
Figure 3 Different types of machine learning models
2.3. Model training and validation
The training process involves data being split into training, validation and the testing of the models. This is to ensure
that performance is not overfitted. Then, there is cross validation which is employed to enhance model robustness.
Standard metrics will be used to assess model performance. Confusion matrices will provide additional insight into the
model’s error rates. This includes false positives and negatives.
3. Results and discussion
This chapter presents the findings of the study and discusses their implications in relation to existing research on Big
Data Analytics in Predictive Nursing. The results are analyzed using appropriate statistical techniques, visualized where
necessary, and interpreted to draw meaningful conclusions about the effectiveness of machine learning for early disease
detection.
The primary objective of this study was to evaluate the effectiveness of machine learning models in leveraging big data
for early disease detection in nursing.
3.1. Dataset composition
The dataset used in this study consists of electronic health records (EHRs), IoT-based real-time patient data, and medical
imaging data. The key attributes of the dataset include:
• Number of Patients: 1500 records from various hospitals and medical institutions
• Age Distribution: Mean = 55.3 years, Standard Deviation = 13.7
• Gender: 53.2% Female, 46.8% Male
• Common Diseases Analyzed: Hypertension, Diabetes, Cardiovascular Diseases, Chronic Kidney Disease, and
Pneumonia
• Missing Data: 3.7% (handled using imputation techniques)
After normalization and handling missing values, data completeness increased to 98.4%, improving model reliability.
1056
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
3.2. Model comparison based on predictive accuracy
This section presents the predictive performance of different machine learning models, evaluated using standard
performance metrics.
The models tested include Logistic Regression, Random Forest, Support Vector Machines (SVM), XGBoost, and Deep
Neural Networks (DNN).
Table 3 Model Testing and efficiency
Model
Accuracy Precision Recall F1-Score AUC-ROC
Logistic Regression
81.2%
79.5%
Random Forest
76.8% 78.1% 0.85
88.4%
87.3%
SVM
86.2% 86.7% 0.92
85.9%
85.0%
XGBoost
84.2% 84.6% 0.90
91.5%
90.8%
Deep Neural Networks (DNN) 90.1%
89.7% 90.2% 0.94
91.2%
3.3. Observations of results
90.5% 90.8% 0.93
• XGBoost achieved the highest predictive accuracy (91.5%) and the best overall performance.
• Deep Neural Networks (DNNs) showed strong results but require significant computational power and a large
dataset for training.
• Random Forest performed well with an AUC-ROC of 0.92, making it an excellent interpretable model.
• Logistic Regression had the lowest accuracy (81.2%) but remained useful for basic risk stratification.
The results showed that vital signs such as blood pressure, heart rate are highly predictive of early disease onset. It also
revealed that real-time IoT device data significantly enhances disease detection. The most important revelation was
that age remains a critical predictor, emphasizing the role of age-related health risks.
3.4. Comparison with Traditional Predictive Methods
From research it showed that traditional predictive tools performed less effectively than machine learning models.
These models displayed higher sensitivity and specificity. This significantly reduces false negatives in early disease
detection. It also displayed a better scalability which means it enables real time predictions from live patient data
sources. Unlike the traditional models that rely on predefined thresholds, the model’s ability to constantly learn and
adapt is crucial in prevention of diseases.
3.5. Future Research Recommendations
To further improve the use of Big Data Analytics in Predictive Nursing, the following recommendations are proposed:
• Enhance Data Integration: Develop standardized data-sharing protocols between hospitals to improve
machine learning model training.
• Improve Model Interpretability: Encourage the adoption of explainable AI (XAI) techniques to increase
clinician trust.
• Expand Real-Time AI Monitoring: Deploy IoT-based early warning systems to improve real-time predictive
nursing care.
• Address Bias in AI Models: Implement fairness-aware machine learning techniques to reduce disparities in
healthcare predictions.
4. Conclusion
This research presented and analyzed the results obtained from testing machine learning models on big data for
predictive nursing applications. The findings demonstrate that XGBoost and deep learning models outperform
traditional methods in early disease detection. However, challenges such as data privacy, model interpretability, and
computational requirements must be addressed for real-world adoption. The key issue now lies with the application of
this models in real life. Take for instance the recent Covid-19 outbreak. Developing countries suffered more because of
1057
International Journal of Science and Research Archive, 2025, 15(01), 1052-1059
the lack of technology to predict where the virus would likely spread faster. This technology will aim at predicting these
kind of diseases so as to ensure that they do not repeat themselves in the future.