Through years as a practicing cardiologist, Amarasingham developed a strategic five-stage clinical pathway that begins with the patient’s first admission and ends ninety days after discharge. The strategy behind this concept, Amarasingham stated, is “to help healthcare providers shore up the messiness of our modern clinical processes and make use of historical data for comprehensive treatment plans for conditions ranging from heart disease to diabetes.”
To make this happen, Amarasingham believes that a lot of algorithms need to be created.
The first step in creating these algorithms is to build out his five-stage approach for patients:
- Identification. What’s the condition? What factors, including environmental and historical data, may affect treatment?
- Predict. What risks are at stake? What are the circumstances of unknown variables?
- Activate. What resources are needed for treatment?
- Monitoring. What clinical data can be compiled throughout the treatment process?
- Learning. What findings from ninety days after discharge can be applied into algorithms?
With these details factored into clinical workflows, Amarasingham believes predictive algorithms can suggest treatment pathways across an array of conditions for future patients. As for Amarasingham’s solution for the endless list of conditions to treat, he expects ten algorithms per stage will need to be created with the help of machine learning and artificial intelligence.
“Machine learning and interoperability between health systems and EMRs will lend a heavy hand in supplying the data needed for these algorithms,” Amarasingham said, suggesting that true predictive analytics, in this instance, would be virtually useless without a real-world care-coordination application. “It’s truly about how you use the data and its application to clinical pathways for the betterment of the healthcare community.”
