Sign up for our newsletter and get the latest big data news and analysis.

From Safari to the Stratified Medicine: How Machine Learning Connects the Dots to Create a Future Health Trajectory

As wildlife trackers piece together clues to locate an animal, machine learning uses indicators to identify the likelihood of disease.

Several years ago, I set out with my family on an African Safari adventure, engaging a professional animal tracker to lead us to see all sorts of wildlife – lion, elephants, rhinoceros, giraffes, and more. We were all amazed by the tracker’s ability to describe animal activity and lead us to some of these animals simply by interpreting small pieces of information from the surrounding area and skillfully piecing them together. The trajectory of a simple footprint, a broken branch, or even a scratch on a tree enabled the tracker to lead us to wonderful discoveries.

As a health technology executive, I recognized a clear parallel between the work of the professional animal tracker and the use of machine learning in healthcare.  Both share the ability to harmonize data to tell a potentially meaningful story. Trackers are trained to connect the dots between small, scattered pieces of information to reach a probability of an animal being in a specific location at a certain time. Through both experience and inherited learning, they know how to identify and connect the dots. Just as they pick up hints and relay insights which, to the average person, may be imperceptible or seem insignificant by themselves, so does machine learning.

Like animals, we leave footprints signifying what happens to us over the years. These traces may be found in our health lab results, medical events, and treatment histories. However, these small and often disparate pieces of information are hard enough to process when standing on their own, even more so when trying to connect multiple elements and determine a future trajectory.

With a simple set of rules, processing and inferring insights is possible for humans. But with multiple details and factors, even an individual patient’s data becomes big data. Machine learning can process millions of data points without fatigue or external distraction, essentially connecting the dots for us.

In today’s healthcare environment, we are hindered by overdiagnosis and the challenge of properly allocating resources to achieve the best possible outcomes. Using machine learning to connect the dots of a patient’s health story can gain invaluable insights, including rate and ratio of changes, trends, effects of interventions and their impact on other measurements and signals. To providers, these insights are immensely advantageous. Using technology to process data for multitudes of people provides a bird’s-eye view of a population and the ability to single out subpopulations of clinical significance. This enables providers to focus resources on the right subgroups of people, select treatments that are most likely to succeed, and measure the effectiveness of interventions. Effective triage, stratification and selection of people can lead to significant savings for providers by offering fewer interventions and exposing fewer people to unnecessary risk.

Similar to how the animal tracker looks for the subtle clues which can lead to a specific animal, machine learning analyzes specific indicators to identify the likelihood of a specific disease or the probability of rising risk for a clinical condition. A fitting example of this is diabetes, a global health issue effecting approximately 422 million people and costing some $825 billion annually.

In this case, the identifying data are routine electronic health records (EHRs) that can be used to identify prediabetic and diabetic patients who are most likely to develop full diabetes or diabetes-related conditions. Using diabetes-focused algorithms to scan enormous amounts of routine EHR data, physicians and healthcare organizations can identify high-risk prediabetic and diabetic patients, prioritizing timely intervention to enhance care delivery which may prevent or delay downstream complications, and reap financial benefits by keeping chronically ill patients healthier.

Instead of embarking on the safari with our own, untrained eyes we went with a tracker who could accurately detect a trajectory which would lead us to identify clearer traces of animals as well as discovering those close by, resulting in a truly impactful experience. The same can be said about use of machine learning for health data. This critical platform of tools can provide the ability to discover hidden gems that the human eye cannot discern. Something to consider when planning your next adventure.

About the Author

Ori Geva is the Co-Founder and CEO of Medial EarlySign. The company’s advanced AI-based algorithm platform helps healthcare organizations accurately stratify populations to optimize care for individuals and prevent or delay serious health conditions, by leveraging routine blood test results, and common labs and EHR data. Medial EarlySign creates actionable opportunities for better clinical decision making and early intervention to improve patient outcomes, focus financial resources, and reduce overall costs.

 

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: