With the growth in electronic health records (EHRs), more and more facilities are gathering huge amounts of digitized patient data. Much of the responsibility for patient data input has been taken on by nurses who previously recorded patient information in paper-based formats. Although accurate documentation is essential for patient care, computerized patient data also enhances quality for the entire healthcare system.
Healthcare providers can use data mining to uncover previously unknown patterns from vast data stores and then use this information to build predictive models. Since the 1990s, businesses have been using data mining for activities such as credit scoring, fraud detection, and maintenance scheduling. Now, healthcare organizations are also seeing value in data mining.
Because EHRs contain so much data, they provide a rich source for healthcare organizations to tap for information about how to improve patient care and reduce costs. Moreover, if their data mining proves useful, these facilities can be eligible for government funding through President Obama’s $14.6 billion program to promote EHR use.
Because the success of healthcare data mining depends on accuracy, healthcare organizations are investigating the best ways to gather data. In many cases, they have turned to nurses or scribes rather than physicians to input this information. By collecting data fully, accurately, and in a standard vocabulary, nurses can contribute greatly to data mining efforts that will ultimately reduce healthcare costs and increase the quality of patient care.
How Data Mining Works
By examining and analyzing stored patient data, expert data miners can uncover important trends. For example, when a Washington, D.C., hospital wanted to know why its patients were getting sick soon after discharge, data mining revealed that patients who had stayed in the same hospital room later developed the same infection. This is just one example of how data mining can help identify and solve problems in health care.
Data miners commonly use the Cross-Industry Standard Process for Data Mining (CRISP-DM) to study the data. This process involves six steps:
- Business understanding—identify the project’s objectives and requirements from a business perspective and define the data mining problem.
- Data understanding—collect the initial data, become familiar with it, and look for any data quality problems.
- Data preparation—build the final dataset from the raw data.
- Modeling—use data mining software to analyze.
- Evaluation—evaluate the achievement of the project’s objectives by comparing data mining models and their results using a common yardstick.
- Deployment—implement the data mining results.
Benefits of Healthcare Data Mining
Data mining in health care has become increasingly popular because it offers benefits to care providers, patients, healthcare organizations, researchers, and insurers.
Care providers can use data analysis to identify effective treatments and best practices. By comparing causes, symptoms, treatments, and their adverse effects, data mining can analyze which courses of action are most effective for specific patient groups. It can also identify clinical best practices to help develop guidelines and standards of care.
Patients can receive better, more affordable healthcare services. This is especially true when healthcare managers use data mining applications to identify and track chronic diseases and high-risk patients, design appropriate interventions, and reduce the number of hospital admissions and claims.
Healthcare organizations can use data mining to make better patient-related decisions. For instance, it provides information to guide patient interactions by determining patient preferences, usage patterns, and current and future needs—all of which helps to improve patient satisfaction. With healthcare organizations under increasing financial pressure, data mining can also influence revenues, costs, and operating efficiency while maintaining high-quality care.
Insurers can detect medical insurance fraud and abuse through data mining by establishing norms and then identifying unusual claims patterns. For example, data mining can pinpoint inappropriate prescriptions or referrals and fraudulent insurance and medical claims. Insurers can use this information to reduce their losses—and the costs of health care.
Nurses have shouldered a large portion of the responsibility for recording patient data in the EHR, but their efforts will contribute to a significant potential benefit to patients and to the health delivery system. As more data becomes available to data miners, nurses will also benefit by having more opportunities to provide appropriate, well planned, and cost-effective patient care.