By managing your information you can keep everything under control. Information is the key to successful life and business. Information analysis – or data analysis – opens as many doors as you can imagine.
As the Healthcare industry picks up momentum and interacts with a wide variety of software technologies and services, the data mining possibilities should not be overlooked. I propose we become acquainted with the data mining possibilities available to the Healthcare industry. And you can believe me; this is only the tip of the iceberg.
Medication Administration Incident Data Measurement Behavioural research Predicting Survival and clinical data mining Patient procedure recommendation
Doctor & nurse note analysis
Cancer Classification Using Gene Expression Data
Identification of Lead Compounds in Pharmaceutical Data Matching of molecular pairs to cluster compounds Product yield analysis
Drug discovery & interaction
Post-marketing drug safety surveillance
Let us clarify the most interesting and, as I think, on a cutting edge, points.
Say NO to Paperwork and Medical Errors
Medication Administration Incident Data Measurement
As a patient, how can you choose a doctor you can rely on? And, as an investor or an insurance company manager, how could you get the objective truth about a clinic’s effectiveness? All the areas of clinical work, that previously have never been measured and put to statistics, now fall under the jurisdiction of data mining.
All patients’ complaints, prescriptions, operation results, information about rehabilitation period are fixed in reports and saved in databases. Was the first doctor’s diagnostics right or not? What prescriptions were made? Was an operation necessary? How many hours did it last? How many days did a patient spend recovering? What results of any therapy were achieved/not achieved? Any discrepancies could then be subjected to a manual analysis - to examine each particular case individually.
But in general, Data mining systemizes all reports and generates a big overview of the clinic’s KPIs with medical error percentages and other gaps that have appeared during the process of delivering medical service being shown in it.
Use of data mining in medical studies relating to human unhealthy behaviour patterns has already become a popular and efficient practice. The daily routines of a group of volunteers are tracked with the help of a mobile application, where people should mark and note some certain information (eg. How much water they drink every day, or the number of cigarettes they’ve smoked, etc.). Together with other biometric indicators - height, weight, age, sport activities, and so on – it can give a dynamic of unhealthy behaviour patterns with their impact on a human’s organism. In this case, the Data Mining technique allows you to pull information from the mobile application data and use it according to your research aim.
Doctor & nurse note analysis, prescriptions and recommendations
Resources, time, and money – progress has taught us that we need to be careful with all of them. Automation accompanied with data mining helps to get rid of medical "paperwork". Using audio steganography, we can significantly reduce the time a doctor spends on writing diagnosis and prescriptions by hand. All prescriptions and recommendations can be also pulled from databases, following the doctor’s or nurse’s voice instructions. This way unstructured clinical notes are transformed into an anonymous patient–feature matrix encoded using medical terminologies. This information can be used for detecting drug–adverse event associations and adverse events associated with drug–drug interactions. Data mining also reveals all information about previously prescribed medications, existing allergies, chronic diseases, and other medical history. With the smart use of data mining techniques we are ready for a rapid analysis of suspected adverse event risk.
Also, the technique allows us to analyse the in-patient nursing records within electronic medical records. Several reasons make data mining valuable here:
Such rare events, as side e?ect from medication and arrhythmia, can be highlighted to improve understanding of their occurrences.
Decisions about success or failure can be obtained regarding courses of treatment for a certain patient. Of course, it must be supported by health care professional supervision in order to identify the symptoms for better understanding of the associated risks.
Data Mining at the Molecular Level
Identification of Lead Compounds in Pharmaceutical Data. Matching of molecular pairs to cluster compounds
Data mining is a part of medical clinical tests or trials, when chemical compounds pharmacokinetics, physical-chemical properties, are pulled from the database in order to use them for further analysis. Using analytics we can measure the result of various chemical combinations in real time and define the best values to use for our study/clinical trial. Virtual analysis reduces production costs and is faster; it can search for the most proper combinations of certain chemical compounds, and automatically maintains all necessary conditions. So called “target discovery” is recognized as one of the most promising key steps in the biomarker and drug discovery pipeline for diagnosing and ?ghting a variety of diseases. Due to the complex structure of different sources, to which information about biological and chemical entities belongs, data mining analysis provides the possibility to map all of this information according to our purposes. Basically, by using a smart navigated search we can make a smart molecule chemical compound that will meet all the necessary criteria to be able to treat a specific disease. With data mining we can see what molecule produces the desired healing result, what elements do not have any functions in the reaction, what conditions lead to the best results, by comparing the results with the help of the latest technologies for data mining and analytics.
For example, it is required to conduct some pharmaceutical research with the aim of reducing the cost of medicine A. The main goal of such research is to find what compounds in our medicine are not useful so that we can remove them, and thus cut the cost. For example, we can identify the combination of several compounds that makes our medicine A so successful in treatment and remove the unnecessary ones. Further using these results we can find affordable substitutes for our main compounds. Based on this we can cut the cost of the medicine itself. Of course, this example is very basic. We understand that Pharmacy is much more complex, but in order to get a basic understanding of the possible results from data mining and further analytics this is good enough.
Another option relates to testing two composition-similar products - sweet A and bitter B - for identifying the effective element. Data mining at a molecular level can show whether the sweetness of the element A is a consequence of fructose or from required treatment components that just happen to have a pleasant taste.
Data mining in the pharmaceutical industry can not only help discover new medicines, but identify patterns which serve best during discovery and delivery phases. The pharmaceutical industry has a lot of information about actual marketing processes, and data mining with strong analytics can help to analyse the perception of the drug’s demand. This can be done by identifying the opinion of the end-user about the medicine, not only through official channels of company itself, but by using social media and various kinds of drug & pharma apps.
Another important issue is dose-response analysis, which is the analysis of different kinds of data (cohort, longitudinal). This analysis provides the procedure to identify key factors that influence patient response by showing different aspects and point of view for the decision-makers. For example, a benefit-risk analysis can be used to see the main benefits over risk of the treatment, and decide whether it is worthwhile to perform a clinical trial or not.
Certainly, data mining itself can’t give the answer to all questions relating to making Healthcare better. But it surely helps while making a patient’s engagement more informative, the decision process more accurate, and the gaps and risks more visible to all players in the Healthcare sector.