Predictive analysis is an advanced branch of analytics that uses data to make predictions about future events. Predictive analysis analyses data using techniques such as data mining, statistics, modeling, machine learning and artificial intelligence. Predictive analytics also automates complex decisions and trade-off to make recommendations based on predictions and then proactively make changes. Predictive analytics in healthcare is the processes of determining the patients who are at risk of getting or developing certain conditions such as lifetime illnesses, diabetes or asthma. Healthcare organizations incorporate predictive analysis into their system to support their medical decision making. When healthcare organizations adapt predictive analysis for healthcare they should understand and be guided by certain principles. These principles will help guide them and enable them to make the most of their information because information plus context equals knowledge but the information out of context can be disastrous.
Principle 1: The Economy of Prediction
The topic of predicting hospital readmissions is a very important and often talked about one with rapidly growing interest. This topic is particularly popular because being able to accurately predict and prevent hospital readmissions will help healthcare organizations in improving patient care while avoiding financial and reimbursement penalties and costs for hospitals. So how can predictive analytics be used to help control costs and improve patient care?
While evidence-based medicine is a powerful tool that helps minimize treatment variation and unexpected costs, the best-practice guidelines contribute further to the goal of standardized patient outcomes and controlling costs. In order for predictive analytics to be effective, Lean practitioners must truly live the process to best understand the type of data, the actual workflow, the target audience and what action will be prompted by knowing the prediction. Decision makers should not be isolated from the points of decision making. For example, if a manager makes a decision on if a certain process should be changed, the manager should have been a part of the process and if possible be affected by the decision too. At the same token, to best leverage, the data, predictors should also not be used in “isolation,” although in the healthcare industry today, readmission risk profiles are often used as standalone applications. So how do you apply [predictive analysis for healthcare in the most efficient way?
One way is to learn from those that have already figured it out or those with existing expertise. Fortunately, in the healthcare industry, there are numerous existing models from other industries that are very efficient at risk stratification in the realm of population management. Studying recent history will also likely ease some of the potential pains and pitfalls that could accompany healthcare’s adoption of predictive analytics.
Principle 2: Don’t confuse data with insight
Having a technology driven and more generalized prediction model that inputs big data and global features is that the target use or utility is often lost in translation. The prediction that is focused on a specific clinical setting or patient need will always trump a generic predictor in terms of accuracy and utility. This principle is particularly important because of how highly humans regard new technology. However, without having the proper technology framework in place, with context and metadata for meaningful use, new technology is really not very useful. Prediction focused on a specific clinical setting or patient need will always trump a new technology driven generic predictor in terms of accuracy and utility.
Principle 3: Don’t confuse insight with value
Just because you have a better understanding of what’s wrong doesn’t mean you should stop looking as the problem may connect with other different parts of the process. Data that is taken as a whole, will often provide an early warning as a patient begins to fail, where even a careful human observer cannot possibly “connect the dots” between so many unrelated data points simultaneously. One key to the success of the algorithm is first obtaining all of the necessary data. Assessing only part of a picture often yields an incorrect view.