Mostrar código-fonte de Data-Mining-in-Healthcare

Authors (HEADS-INFORM 2025):
* Henrique Pereira
* Olívia Oliveira

= Introduction =
Data mining refers to the process of discovering meaningful patterns and extracting actionable knowledge from vast amounts of data, often with the aid of computational tools.

The healthcare sector generates massive amounts of data, including electronic health records (EHRs), genomic sequences, medical imaging, wearable device outputs, patient-generated data, and clinical trial data. These data sources provide a fertile ground for applying data mining to improve care delivery, patient outcomes, and operational efficiency (Islam et al., 2018 <ref name="dm-review">Islam MS, Hasan MM, Wang X, Germack HD, Noor-E-Alam M. A Systematic Review on Healthcare Analytics: Application and Theoretical Perspective of Data Mining. Healthcare. 2018 May 23;6(2):54.</ref>).

Importance:
* Enhances clinical decision-making;
* Facilitates the early detection and prediction of diseases;
* Improves operational efficiencies in healthcare systems;
* Supports public health research and epidemiological studies.


The image above is presented in Gordan et al., 2022 <ref name="dm-soa">Gordan M, Sabbagh-Yazdi SR, Ismail Z, Ghaedi K, Carroll P, McCrum D, et al. State-of-the-art review on advancements of data mining in structural health monitoring. Measurement. 2022 Apr 1;193:110939.</ref>, and illustrates a historical timeline of advancements in Structural Health Monitoring (SHM) and its integration with Artificial Intelligence (AI) and Data Mining technologies.

(image-evolution-dm)

The timeline traces significant developments across multiple decades, emphasizing the evolution of data mining techniques in SHM and highlighting their growing relevance from the 1990s onwards. Key milestones in this context include real-time data monitoring in 1992, operational modal analysis in 2001. 

The recent focus, as seen in the "Data mining-based SHM" from 2014 to the present, underscores the role of data mining and AI in optimizing decision-making, predictive analysis, and enhancing the overall reliability of critical infrastructure.

In the context of healthcare, this timeline provides an analogous framework for understanding how data mining and AI can revolutionize decision-making, diagnostics, and operational efficiency by leveraging advancements in real-time monitoring, predictive analytics, and data integration.

= Key Data Mining Techniques in Healthcare <ref name="dm-techniques">Birjandi SM, Khasteh SH. A survey on data mining techniques used in medicine. J Diabetes Metab Disord. 2021 Aug 31;20(2):2055–71.</ref><ref name="dm-models">Wu WT, Li YJ, Feng AZ, Li L, Huang T, Xu AD, et al. Data mining in clinical big data: the frequently used databases, steps, and methodological models. Military Medical Research. 2021 Aug 11;8:44.</ref> =

== Classification ==
Used for categorizing patients or medical conditions based on predefined labels (e.g., predicting disease risks).

'''How it works:''' Algorithms are trained on labeled datasets to learn patterns and apply these insights to new data.

'''Common algorithms:''' Decision Trees, Random Forest, Support Vector Machines (SVM), Neural Networks.

'''Use cases in Healthcare:'''
* Diabetes risk prediction based on patient history and demographics (Islam et al., 2018 <ref name="dm-review">Islam MS, Hasan MM, Wang X, Germack HD, Noor-E-Alam M. A Systematic Review on Healthcare Analytics: Application and Theoretical Perspective of Data Mining. Healthcare. 2018 May 23;6(2):54.</ref>);
* Classification of medical images for cancer detection (e.g. using CT or MRI scans) (Islam et al., 2018 <ref name="dm-review">Islam MS, Hasan MM, Wang X, Germack HD, Noor-E-Alam M. A Systematic Review on Healthcare Analytics: Application and Theoretical Perspective of Data Mining. Healthcare. 2018 May 23;6(2):54.</ref>).

== Regression ==
Focused on identifying relationships between variables to predict continuous outcomes (e.g. mortality rates or healthcare costs).

'''Common algorithms:''' Linear Regression, Logistic Regression, Ridge and Lasso Regression.

'''Use cases in Healthcare:'''
* Estimating the length of hospital stays based on patient conditions;
* Analyzing the impact of cholesterol levels on cardiovascular disease outcomes (Islam et al., 2018 <ref name="dm-review">Islam MS, Hasan MM, Wang X, Germack HD, Noor-E-Alam M. A Systematic Review on Healthcare Analytics: Application and Theoretical Perspective of Data Mining. Healthcare. 2018 May 23;6(2):54.</ref>).

== Clustering ==
Identifies natural groupings within data without predefined labels. This unsupervised technique is valuable for segmenting patients or detecting disease patterns.

'''Common algorithms:''' K-Means, Hierarchical Clustering, DBSCAN (Density-Based Spatial Clustering).

'''Use cases in Healthcare:'''
* Patient segmentation for personalized treatments;
* Identification of patterns in epidemic outbreaks based on geospatial data.

== Anomaly Detection ==
Used to detect outliers or irregular patterns in data; often used for monitoring and quality assurance.

'''Common algorithms:''' Isolation Forest, Autoencoders, Statistical-Based Methods.

'''Use cases in Healthcare:'''
* Identifying fraudulent claims in insurance systems;
* Detecting rare adverse drug reactions in pharmacovigilance studies.

== Association Rules ==
Used to unveil frequent co-occurrences between variables (e.g. analyzing if the presence of condition A increases the likelihood of condition B).

'''Common algorithms:''' Apriori, FP-Growth (Frequent Pattern Growth).

'''Use cases in Healthcare:'''
* Identifying frequently co-prescribed drugs to understand potential interactions;
* Discovering risk factors associated with chronic diseases like hypertension.

== Neural Networks and Deep Learning ==
Simulates human neural processes to handle highly complex, high-dimensional and unstructured data (e.g. medical imaging or genomics).

'''Common algorithms:''' Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Autoencoders.

'''Use cases in Healthcare:'''
* Detection of diabetic retinopathy from retinal scans;
* Prediction of clinical outcomes using unstructured EHR data.

= Applications of Data Mining in Healthcare <ref name="dm-review">Islam MS, Hasan MM, Wang X, Germack HD, Noor-E-Alam M. A Systematic Review on Healthcare Analytics: Application and Theoretical Perspective of Data Mining. Healthcare. 2018 May 23;6(2):54.</ref><ref name="dm-hc-tech">Kolling ML, Furstenau LB, Sott MK, Rabaioli B, Ulmi PH, Bragazzi NL, et al. Data Mining in Healthcare: Applying Strategic Intelligence Techniques to Depict 25 Years of Research Development. Int J Environ Res Public Health. 2021 Mar 17;18(6):3099.</ref> =

1. '''Early Diagnosis and Disease Prediction:''' Using patterns in data to predict diseases in early stages (e.g. cancer detection through image processing and analysis).

2. '''Personalized Medicine:''' Leveraging genomic data to design individualized treatment plans.

3. '''Real-Time Health Monitoring:''' Analyzing data from wearable devices for continuous health monitoring (e.g. arrhythmia detection).

4. '''Operational Efficiency in Healthcare providers:''' Predicting demand for resources like beds or optimizing staff allocation.

5. '''Public Health Research:''' Mapping epidemic patterns and identifying high-risk populations.

6. '''Drug Discovery:''' Accelarating drug development by identifying new therapeutic targets or repurposing existing drugs.

7. '''Fraud Detection and Error Prevention:''' Identifying anomalies in billing systems or medical record inaccuracies to prevent financial and clinical risks.

= Challenges in Healthcare Data Mining <ref name="dm-challenges">Qiao H, Chen Y, Qian C, Guo Y. Clinical data mining: challenges, opportunities, and recommendations for translational applications. Journal of Translational Medicine. 2024 Feb 20;22(1):185.</ref> =

1. '''Data Quality and Diversity:''' Handling incomplete, inconsistent or heterogeneous data from various sources.

2. '''Privacy and Security:''' Ensuring compliance with regulations like GDPR and HIPAA to protect sensitive patient information.

3. '''Interoperability of Systems:''' Integrating diverse data formats from EHRs, wearable devices, laboratory systems, and many others.

4. '''Complexity of Medical Domain:''' Bridging the gap between data scientists and clinicians to ensure actionable insights.

5. '''Bias in Data:''' Ensuring equitable outcomes by addressing skewed or imbalanced datasets.

6. '''Result Interpretability:''' Translating complex patterns into clinically meaningful recommendations.

= Future Directions of Data Mining in Healthcare  =
The future of data mining in healthcare is marked by advancements and opportunities, driven by emerging technologies and the growing availability of complex datasets. These developments promise to revolutionize how healthcare is delivered, monitored, and optimized for better patient outcomes.

* '''Integration with Artificial Intelligence (AI):''' The convergence of data mining techniques with AI, particularly deep learning, is poised to enhance predictive modeling and decision-making. By combining the strengths of traditional data mining approaches with advanced AI algorithms, healthcare systems can achieve greater accuracy in predicting disease risks, treatment outcomes and patient pathways, and generally improve elements of the care delivery process (Lin et al., 2021 <ref name="dm-ehr">Lin AL, Chen WC, Hong JC. Chapter 8 - Electronic health record data mining for artificial intelligence healthcare. In: Xing L, Giger ML, Min JK, editors. Artificial Intelligence in Medicine [Internet]. Academic Press; 2021 [cited 2024 Dec 17]. p. 133–50. Available from: https://www.sciencedirect.com/science/article/pii/B9780128212592000089</ref>).

* '''Real-Time Data Utilization:''' The proliferation of Internet of Things (IoT) devices and wearable technologies enables the collection of real-time health data. This capability supports timely health interventions, such as monitoring chronic conditions, detecting anomalies and providing personalized feedback to patients, ultimately improving care management and outcomes (Ashok, 2007 <ref name="dm-applications">Ashok PR. Data Mining in Healthcare: Applications, Benefits, and Future Directions. 2007;5(4).</ref>).

* '''Value-Based Healthcare Models:''' Data mining is becoming instrumental in transitioning to value-based healthcare models, where patient outcomes take precedence over the volume of services provided. By analyzing diverse datasets, healthcare organizations can identify key drivers of positive outcomes, optimize resource allocation, and tailor care delivery to maximize value for patients and providers alike (Ashok, 2007 <ref name="dm-applications">Ashok PR. Data Mining in Healthcare: Applications, Benefits, and Future Directions. 2007;5(4).</ref>).

* '''Expansion of Omics Data Analysis:''' The integration of genomic, proteomic and metabolomic data into healthcare analysis is paving the way for personalized medicine. Data mining techniques enable the identification of patterns and correlations within these large-scale datasets, supporting tailored treatment strategies that align with an individual’s unique biological profile (Bensmail et al., 2005 <ref name="dm-proteomics">Bensmail H, Haoudi A. Data Mining in Genomics and Proteomics. J Biomed Biotechnol. 2005;2005(2):63–4.</ref>).

As these advancements unfold, data mining will continue to play a pivotal role in shaping the future of healthcare, fostering a system that is more precise, efficient, and patient-centered.


= References =
<references />