|
|
|
Index - Major Sections
Site Map
Product and Services _______________ Index - Same Level Subject
Index - Child Subjects |
Data Analysis
Data analysis and Healthcare Intelligence (Business intelligence) can be used for baseline calculations, perditions, performance measurements, cost-benefit analysis, inventory management, classifying, fraud detection. Since all information is available from all sources for analysis, relationships among all health care indicators can be determined. Causal relationships between project policies and improved health care can be determined by having the ability to analysis longitudinal data. Uses of Data AnalysisData Analysis should allow the user Some of the uses of data analysis
Quality of InformationData quality can be improved by having internal checks on
the data as it is entered into the system. If data is out of range, or is
inconsistent with other data values, that information would be returned
immediately to the service worker. With detail information available, data will be actuate
and reliable. In many cases, in present day systems, since only summary data
is sent up to the next level, the lower level staff, due to time constrains or
political pressure may “make up
the data.” A Healthcare Information System can promote and support honesty by the Healthcare Professional. This can be done by first using controls on who enters the data and when and secondly by allowing Healthcare Professionals to modify his or her records after initial entry by explaining, reinterpreting and commenting on his or her actions. The Healthcare Professional will be identifiable and the time of his entry will be determinable. Since the "old information" is readily accessible, cross checks can be made against any entry. Errors can be readily identified and corrected at the source without any time delay and face to face with the client. By allowing Healthcare Professionals to modify their information, the Healthcare Professional will not feel that he or she is "bound" by the system. As new information is collected this information can be used to modify the original records. After all, it is correct information that we are after. Prescribed information that is required will now be collected. Before the information that was suppose to be gathered was never collected because no one monitored its collection. Ensures complete documentation of encounters. Algorithms can be developed to correct errors and suggest alternatives. Every piece of data is coded for better data analysis and data mining Data MiningWhile Data Analysis tells you what happen and using trend analysis why tell you where you are going, Data Mining tells you "interesting things" with the intention of uncovering hidden patterns. It tells you the relationship between variables...it tells you what you may not already know. Data Mining uses complex statistical algorithms to extract patterns from data. It is used in profiling practices (what individual is most likely to have a certain illness), marketing (what individuals are most likely to respond to a promotion in nutrition), surveillance, and fraud. Data Mining is the heart and sole of the HIS system (or it should be!). It is medical research that has carried healthcare to gains that we see today. Without research knowing would change. Client Profiling Clients are all individuals but at the same time each has many attributes in common with other individuals. While it is common to say that a physician treats every individual differently...it is not true. Physicians treats patients based on what they have learned in school and what their experiences have taught them....and that they have time to "investigate". This process is intuitive and is less than perfect. Profiling and Segmentation of Clients with Data Mining into groups actually provide better and more personal care to the individual. While a physician may treat a patient based on age and gender, data mining can profile the patient based on age, gender, education, economic factors, previous history, family history, previous compliance with treatment...and the list goes on. Data Mining allows more attributes to be consider and it produces a probability of this individual responding to several alternates. A single healthcare professional cannot know or even remember all the factors that are relevant in creating a treatment plan....Buying a "Standard Guideline" and using it for all patients is not particularly a "cool" thing to do... A primary reason for using data mining is to assist in the analysis of collections of observations of behavior. Prediction The busy healthcare professional (HCP) is use to dealing with many clients and the need to evaluate their risks. If this risk assessment is not made correctly, the consequences may be that the client suffers or that the follow-up is a waste of time, money, and some other person suffers. Using a prediction model, the HCP can better select the clients with whom to follow up. A Prediction tool can also be used in Triage not only to assign scores but also to rate what influence each attribute has on that score. While a older person may have the same (signs and symptoms), the older person may be at greater risk because of their age. Recommendations Data mining tells you what services and products the healthcare organization "should" be offering based on the population profile. Clients that receive the services that they need are more likely to have a stronger relationship with this organization. An example is Amazon.com offering books that you may be interested in based on past behavior. Anomaly Detection Data mining can be used to detect anomalies in data by "flagging" data that does not fit with the rest. It can determine if a particular transaction is valid by comparing the client profile with the symptoms, the results of diagnosis tests, the treatment plan, the orders for that individual and the outcomes. Anomaly Detection can be used to check if the data entered is correct at the point of care. This eliminates major problems of miscommunication. Insurance companies use this to detect fraudulent claims. (see Benefits - Error Detection) Risk Management Data Mining can determine, based on the client profile and behavior, their healthcare risks. Preventive medicine and chronic care procedures can be started much sooner than has been done in the past...when it is too late. Education and Treatment Compliance Based on Client profiling the client can be present with a education program or treatment that they are more likely to use. There is no need to spend a lot of money on any type of promotion until the targeted clients are profiled to determine what they will response to. Targeted Marketing Same as previous topic. Instead of blanking a whole town with advertisement, with data mining a organization can personalize their advertisements to the targeted client that they want to reach. It is no use sending out "no smoking" literature to people that do not smoke or people that would not respond to this type of literature. This is happening now when you access a web site. Advertisement will be displayed depending on what other sites you have "clicked" on. Basket Analysis and Forecasting The results tells the organization that if someone buys one product, what is the next most likely product they will buy and what items are in the basket. This can be used to great advantage in healthcare to project the resources needed to care for an individual. If the patient is pregnant what are the services most likely needed and when. Basket Analysis is used to a great extent in grocery stores to order displays in their stores. Churn Analysis Which Clients are most likely to switch to a different type of organization? Churn Analysis indicates when a client is not happy with the existing services offered by the organization. For better and continuous care of the individual, it is better to retain the client as long as possible. Churn analysis can help managers identify the clients who are likely to leave and why. Knowing this management can take actions to improve their client relationship. Research First, Searches can easily be performed to select individuals for clinical trials. Once these individuals are selected, a Data Mining procedure can be initiated that can "discover" all relationship between this individuals. Who is more likely to get an illness and what are the attributes of these individuals that make them more likely. Data is reusable and kept over long periods of time. Comparisons can be compared over time and as tools get better, so will the old research. Research requires data and if this data can be saved, better use can be made of it. Low participation in clinical trials is a significant problem in clinical and translational research. Why should it not be...in "gold standard" clinical trials, one person gets the good stuff, and the other gets the not so good. Would you want to participate in a clinical trial? By using massive data bases and Retrospective analysis there is no need to perform this "not-so-fair" treatment protocol. It should not be too hard to compare the outcomes of what exists vs. the outcomes of a new treatment. (see Benefits to Research) Decision Support Operational business Intelligence (OBI) can be used to make decisions in real time. As an example, when a HCP is examining a client, as more information is obtained, the original perception can be modified with the updated data. OBI requires masses amounts of reference data along with the operating data in order to enable real time decision making. Integration of Health Care SystemsThis integrated system makes it possible for every stakeholder to access and use the same information, thereby increasing the ability to coordinate their activities. Increases the priority of preventive measures in health care. One of the reasons that preventive health is not made a priority in health care by governments is that it is difficult to measure both the implementations that must be made and the results of those implementations. For example, it is extremely easy to measure hospital visits and the number of patients "cured," but is it very difficult to measure the success of "exercise." With an Information System that provides for longitudinal data analysis, the success of exercise or other risk factors can be measured. Insurance Programs can be implemented and privatized. One reason that it is so difficult to introduce insurance into a developing country is due to the lack of risk analysis. The introduction of Insurance programs into a country can have a synergetic effect on health care. As the Insurance programs gather more data, this in term can provide for better care. Not only clinical data, but all data such as accounting, human recourses, physical recourses and even external data such as special studies and research from other NGO's can be integrated into the data warehouse thereby providing the ability to get the "whole picture." By having the ability to integrate all the data, consistency checks between the outputs of one unit with the inputs of another (inventory, for example). SurveillanceAll data is available at the detail level for every clinical contact. Data is transferred within a few hours to a central database where research and epidemiological studies may be carried out. There is no requirement that the local staff must be trained in epidemiological or environmental surveillance (where it is generally found to be ineffective). Data can be made available to all (to those with proper authorization) for “competitive” analysis. Decisions can be based on better analysis that is provided at a lower cost. (see Benefits Surveillance) Data WarehouseThe Data Warehouse can become a storage system for the
“industry.” It can function as a industry clearinghouse of data. When
other research is performed, the data in the Warehouse can be made available
to all. A readily available source of data will attract competition among organizations that perform data analysis services. It is possible for the Data Clearinghouse to act as a
“coordinator” of several funded projects. For example, two or more funded
projects may share their research information in a way that produces better
results with less cost. This would enforce standards on the “industry.”
Other organizations will want to contribute their data to this data
warehouse or they will themselves not be able to access its database. They
will be left out and at a disadvantage. Reports
Reporting possibilities for the clinician are virtually limitless and should facilitate data acquisition in activities ranging from clinical trials to outcome analysis of routine patient care.Information can be easily disseminated to all interested stakeholders. Links
|
|
|