Johns Hopkins Research Finds that Data Mining of Health Records useful to Reduce Physician and Treatment Mistakes Diagnostic errors are one of the biggest patient safety issues we face in healthcare and these very often lead to medical errors. It is often caused by a diagnosis that is missed, wrong or delayed, as detected by some later definitive test or finding. These costly errors can result in delay or failure to treat a condition, or to provide treatment for a condition that doesnâ€™t actually exist. On evaluating 25 years of U.S. malpractice claim payouts, Johns Hopkins University researchers found that diagnostic errors are the reason for death or permanent damage for around 160,000 patients every year. These errors are also the leading reason for malpractice claims that are paid to physicians. The notable thing is that diagnostic errors are more easily preventable than any other medical mistakes. Automation is a practical solution that can address this problem. Computers can be used to check medical records and identify possible errors, and also to prompt doctors to follow up on risky test results. Helpful online services that can assist doctors with diagnoses and tests/devices that can help them identify conditions/illnesses more accurately are other solutions. Doctors are being made aware of the risk involved in holding on to one diagnosis and not looking further. They need to keep an open mind in cases that appear confusing with conflicting evidence. The new healthcare law that lays emphasis on coordinated care is expected to improve diagnosis while also ensuring that patients consult specialists when they are required to do so. Effort is on to develop techniques that can identify and measure diagnostic errors. Data mining from electronic records can help identify information such as lab results that may have escaped notice. Data mining is one of the powerful techniques available for accurate disease diagnosis. When a large amount of medical data is available, more powerful data analysis tools can be used to mine useful information. For example, researchers are employing statistical and data mining tools to assist healthcare providers diagnose heart disease accurately. Data mining techniques are also employed for the prevention of diseases such as cancer, stroke, cardiac arrest, and diabetes. It helps in the prevention of hospital errors, in early detection and prevention of diseases, and in the detection of fraudulent insurance claims.
Data mining techniques for diagnosis varies with disease. For example, the most frequently used techniques for the diagnosis of heart disease are naïve bayes, decision tree, and neural network. Kernel density, automatically defined groups, bagging algorithm, and support vector machine are some other techniques used in heart disease diagnosis. Studies reveal that diagnostic errors occur usually due to problems in ordering diagnostic tests, history taking, examination, and referrals. About 14% of immediate deaths, 19% of serious permanent damage, and 16% of serious harm are some of the end results. Such a situation can be prevented using data mining which includes the following steps:
Problem Definition (Identifying goals) Data Exploration (Analyzing Quality of Data) Data Preparation (Cleaning Data) Modeling (Applying data mining algorithm) Evaluation and Deployment (Extracting Information)
Although applying data mining in disease diagnosis is beneficial, not much research has been conducted in this area to identify treatment plans. Anyway, researchers claim that incorporating these techniques can improve the efficiency of the physician or practitioner. As data mining consumes a lot of time and effort, whether to identify missed information or to make an accurate diagnosis, it is better to outsource this job to a reliable company with expertise in data mining, rather than setting up a team in-house. Relying on professional data mining services will help to complete the task accurately and within quick turnaround time. Avoiding mistakes in the evaluation of health records could reduce diagnostic errors and thereby mortality rate to a considerable extent.