Descriptive, Predictive, and Prescriptive Analytics

Descriptive, Predictive, and Prescriptive Analytics

Big data is crucial to the continued success of any big organization. The accumulation of information relating to systems, processes, supply channels and a myriad of other sectors particular to any field predicts future trends and helps organizations plot strategic decisions that could uplift the overall vision and mission of that organization.

In the field of healthcare, this takes on the added role of saving lives, prioritizing patients, and increasing the efficiency of healthcare systems. The three big trends of analytics that are most prevalent in working towards improvement of outcomes include descriptive, predictive and prescriptive analytics. Key Performance Indicators (KPI’s) using these systems are more likely to form systematic improvements and help organizations become innovators in their field.

The Three Major Types Of Analytics:

Descriptive analytics utilizes raw data from the past to accumulate information and present statistics on how previous interactions within the industrial chain led to certain results. By using basic mathematical principles, it is possible to create summations of raw data. In healthcare, these important insights serve as models for future patterns of organizational behavior.

Predictive analytics plays into utilizing big data to predict potential outcomes for the future. The processes used could be a risk assessment, simulations, and even geographic information systems that streamline population management. Using predictive analytics in the clinical workflow could not only help in the reduction of waste but also streamline the operations. Let’s take the case of one of the most commonly used predictive analytics in healthcare: LACE index predicts readmission or death of a patient within 30 days of discharge. The independent variables that are used to predict this outcome are Length of Stay, acuity of the admission, co-morbidities, and the number of ED visits that the patient had within the last six months. This helps to potentially prioritize the patients that have a higher risk of 30-day readmission or death over those that are not at the risk of imminent danger.

Predictive analytics does not serve as an infallible predictor of future outcomes but serves to be an improvement on existing processes which, under a streamlined system, use analytics to determine the most useful course of action to take.

Prescriptive analytics also looks at future outcomes, but unlike the predictive analytics, this model provides information on recommended action that could be taken. Another differentiating factor is that this model not only provides information on what and when it may happen but also why it may happen. In general, the prescriptive models are built by ingesting both structured and unstructured data (text, documents, images, videos, etc.). With this level of information, the organizations and the clinicians have an opportunity to take full advantage of the situation or even mitigate a future risk.

Conclusion:

Real success comes from using all these three forms of analytics as improvement indicators in the healthcare system. Descriptive analytics by itself serves only to create raw data that has no use for future models of organization. Charting the past can provide insights, but without working in tandem with predictive and prescriptive analytics, that data may never see any actual use in improving the lives of patients and better-streamlining healthcare operations. Data-driven organizations can even chart the prevalence of diseases and predict the potential of an outbreak based on that data, giving them the chance to pre-emptively prepare medication and vaccinations. Without a complementary interaction of these three systems of analytics, KPI’s would be limited to just statistical data.

Tags: , , , , , , , , , , , , ,