The Need for Predictive Analytics in Improving Health Outcomes

Where can the ability to predict future events from the past and present data prove invaluable? Pretty much every industry there is, you would imagine. Though data analytics is an emerging field, a number of industries are tapping into its potential.

Data analytics in various industries

Take the retail industry. Historical sales data can help predict future demand in larger regions but fail to uncover underlying signals, especially when there isn’t much data to analyze. Anonymized and aggregated web search, which is a big data solution, can do a far better job of predicting sales at a hyperlocal level with a lower forecast error.

The airline industry is leveraging business intelligence suites to predict mechanical failures in advance to determine the possibility of aircrafts being canceled or delayed. Machine learning solutions analyzing maintenance history and flight route information are helping airlines initiate maintenance actions when aircrafts are being serviced, to counter the risk of cancellations or delays.

The scope of predictive analytics for the agriculture industry is immense. Farmers can leverage it to predict variables such as pesticide quantities, feed rationing for livestock, and crop prices to reduce waste, raise healthy crops and animals, and plan proactively. AgTech solutions company The Yield’s microsensing system can predict salinity levels in water and predict when oyster harvesters need to pull out oysters. The system has been able to reduce unnecessary harvest closures by 30 percent, helping NSW and Tasmanian farmers save thousands in lost revenue per day that can result from harvest closures during peak seasons.

Healthcare potential

Predictive analytics in healthcare can improve outcomes in several ways. After all, data drives the entire health journey, from the public domain to the individual and practice levels, to calculating value-based payments. Here is a look at a few applications of predictive analytics in healthcare.

1. Though less than 0.5% of all newborns have their infections verified by blood culture, 11% of them receive antibiotics. Kaiser Permanente of North California has leveraged predictive analytics to cut down on this overuse. Their researchers have developed an algorithm that predicts the risk of neonatal infection based on the mother’s clinical data and the baby’s condition soon after birth. It is helping doctors figure out which babies need antibiotics, thereby lowering medication costs and side-effects among newborns who may otherwise end up receiving unnecessary antibiotics.

2. Four hospitals that together constitute the Assistance Publique-Hôpitaux de Paris (AP-HP) use external and internal data, including a decade’s worth of admission records, to predict the number of patients arriving each day and each hour. The predictive analytics platform uses time series analysis techniques to predict admission rates at different times.

3. Within the field of genomics, predictive analytics can enable primary care physicians to identify at-risk patients and recommend suitable lifestyle changes. When such proactive changes occur on a larger level, disease patterns may change dramatically and lead to significant savings in medical costs.

4. There are predictive analytics platforms that help couples planning a family to understand inherent health risks that can be passed on from them to their children. Such platforms can provide probabilities on hundreds of health conditions after investigating parents’ DNA. Use cases include disease prevention, early intervention, and proper treatment selection.

5. On an average, the United States records 6.1 infant deaths per 1,000 live births. At 7.7, Indiana’s infant death mortality is higher than the national average. In 2014, the state launched a predictive data analytics platform connecting state agencies (approximately 800 individual systems) that uses historical data to gain a deeper, more comprehensive understanding of the causes of infant mortality.

6. Evidence shows that early administration of antibiotics reduces sepsis-related mortality. Predictive analytics tools can be very useful in lowering or preventing sepsis by connecting to existing hospital information and analyzing patient data 24/7 to support early sepsis detection and provide clinical alerts to smartphones/tablets.

7. The number of central line days is a key predictor of hospital-acquired infection. Long-term lines arising from missed line maintenance activities and failure to complete CLABSI-prevention bundles can be more easily tracked with data visualization on predictive analytics dashboards. Unit nurse managers and bedside nurses can identify which patients might have missed these opportunities and act quickly.

8. When leveraged for inventory management and procurement, predictive analytics can help in maintaining reorder levels for all essential items and reduce purchase cycle times. This can be invaluable in ensuring that the hospital is never out of stock during critical situations.

Conclusion

Predictive analytics is allowing organizations to utilize their time in value-generating activities across all aspects of patient care and internal operations. Predictive algorithms have become more sophisticated, lowering cognitive errors and offering clinicians just-in-time information to make real-time decisions. Healthcare organizations will do well to evaluate areas where big data can unlock meaningful value in everyday clinical care.