CLOSING THE GREAT HEALTH CARE DIVIDE WITH PATTERN RECOGNITION AND DATA - MINING TECHNOLOGIES
In today's world, people, cultures, and nations are distinguished by which side of the " great divides" they are on. There are the financial and economic divides, which differentiate between the haves and have-nots with respect to wealth, income, and prosperity. There are the educational divides that distinguish among people on the basis of their access to education and according to whether a college degree is achieved.
There is a "great digital divide," on either side of which are those who have or do not have access to technology. Worldwide-and certainly in the United States-is a health care divide between those who can and cannot afford good health care. With health care costs spiraling out of control, the health care divide is widening and the population with access to affordable health care is shrinking.
Therefore, many health care providers and technology providers such as IBM, the Mayo Clinic, and the Cleveland Clinic are collaborating on IT-enabled strategies to reduce the cost of health care while providing better health care than ever before. IBM and the Mayo Clinic are collaborating on costefficient ways to provide customized medical treatments to individual patients, such as choosing the right chemotherapy treatment for a specific patient given his or her unique genetic makeup, by applying pattern recognition and data-mining technologies to the electronic medical records of some 4.4 million patients.
In the first phase of the project, IBM and the Mayo Clinic worked together to consolidate into one database all patients' medical records from numerous separate systems-including digitized (electronic) patient files, lab results, X-rays, and electrocardiograms-and from all of Mayo's hospitals in Arizona, Florida, and Minnesota.
In the current, second, phase, IBM and the Mayo Clinic are applying pattern recognition, datamining technologies, and other decision support tools to help health care providers determine the best therapies for each patient. The goal of the decision support system is to find patterns relating to how well different types of patients respond to certain types of therapies. The system sifts through patient information concerning age, medical history, and genetics to find patterns that suggest the best treatments.
For example, the Mayo Clinic recently began using a new chemotherapy treatment for lung cancer that produced disappointing results, despite all the positive research findings. The decision support system, however, was able to determine that a small percentage of patients, all with the same genetic pattern, did respond positively to the therapy, which is now targeted for only those individuals matching that genetic profi le. Similarly, IBM and the Cleveland Clinic are working on a decision support system to identify patients most susceptible to abdominal aortic aneurysms. The presence of aneurysms often shows no symptoms until the aneurysm is about to rupture, and the survival rate of the rupture is less than 50 percent.
According to Dr. Kenneth Ouriel, chairman of the Division of Surgery and the Department of Vascular Surgery at the Cleveland Clinic, "If we can predict what patients are at risk for these ruptures, we can identify the patients who need surgery or other treatments." To identify such patients, the system evaluates a wealth of information including lab, genetic, imaging, and drug history and how different factors contribute to risks as well as to treatment outcomes.
Questions
1. In the case study, we referred to the systems being developed and used as decision support systems. However, we also identified various artificial intelligence (AI) technologies. How can a decision support system incorporate and use AI technologies such as pattern recognition?
2. At the Mayo Clinic, patients are given opt-in and opt-out rights concerning whether or not their information is used in the system that determines the most appropriate therapies given the specific patient profile. So far, 95 percent of the patients have opted to have their information included in the system. (This is the notion of opting in. ) Why do you believe that 5 percent of the patients have opted out? Would you opt in or opt out in this case? Please provide your reasoning.
3. In this case, demographic information such as ethnicity, gender, and age greatly impacts the quality of the decision support and analysis. The same could be argued for the predictive analytics system used by the Richmond police in this chapter's first closing case study. Why would some people find it acceptable to use such demographic data in this case (for medical purposes) and not in the first case (for predicting crime, its location, and its timing)?
4. One of the most popular and widely used application areas for expert systems is medicine. What role could an expert system play in helping the Cleveland Clinic identify patients susceptible to abdominal aortic aneurysms? What sort of rules would the expert system include?
5. How might a monitoring-and-surveillance agent be used for patients in a medical environment? How can health care providers take advantage of the capabilities of an information agent to stay abreast of the latest medical trends and treatments?