Data mining: health care must look to adopt predictive analytics

That’s according to the results from an August survey of 136 hospital and health system executives by Health Catalyst, which would seem to confirm that widespread adoption of analytics is here, or close to it. Yet the same survey revealed that only 31 percent of hospitals have used the technology for more than a year, although 38 percent said they plan to adopt predictive analytics within the next three years. What’s preventing these organizations from using a powerful tool, right now, that can help curb preventable readmissions, develop precision treatments for diseases such as cancer, and renegotiate contracts with insurers that pay less than the cost of providing care, just to name some of the longstanding challenges that predictive analytics can tackle? It turns out that while there’s certainly a demand for predictive analytics, the necessary infrastructure and ready access to data are lacking. That’s the verdict of 32 percent of respondents in the above-cited survey, and it’s not a surprising one. Producing reliable predictions of future probabilities or trends does require an accessible, trusted source of data, aggregated from multiple IT systems such as the hospital’s EHR and financial systems. When asked to list the top sources of data for making predictions, respondents cited clinical data from the EHR, claims data and patient outcomes data, financial data and non-medical patient demographics and patient satisfaction data.