KenSci is fighting death with data science by taking advantage of the Microsoft cloud and data platform

This story was first published on Microsoft’s customer portal. You can read the whole story here

KenSci is the first machine learning–powered risk prediction platform that uses electronic medical record (EMR), claims, psychosocial, operational, and patient-generated data to align payers and providers around value-based care initiatives. The KenSci Risk Prediction platform, built on Microsoft cloud and data technologies, is powered by more than 150 prebuilt machine learning models, and it predicts care and cost risks for 17 million patients today.

As we move toward value-based care, predicting clinical and cost risk is becoming central to the digital transformation of US healthcare.

Sunny Neogi: Chief Growth Officer

KenSci

The limitations of heuristics in healthcare

KenSci uses machine learning to power predictive risk management in healthcare. We’re developing solutions that help payers, doctors, nurses, and others across the care continuum identify patients who will get sick, when they’ll get sick, how sick they’ll get, and what can be done to improve their health or help them avoid getting sick in the first place.

We come from a strong academic and research background, having worked within the University of Washington, collaborating with Microsoft Research for more than five years. Our team consists largely of doctors and data scientists. Our sole focus is the application of advanced analytics to improve healthcare and help people enjoy longer and healthier lives.

For example, if a patient is older than 45, overweight, hypertensive, and has poor blood test results, a doctor will recognize a high risk of a heart attack or stroke with relative confidence. But the doctor’s prediction still suffers from a serious limitation: precision. It lacks mathematical rigor and can have a statistically significant margin of error. Worse, we often seek medical help after we fall sick, constraining our health system’s ability to prevent health issues and overburdening it in turn. Caregivers are often pressed for time and lack tools to incorporate social and psychographic determinants that have outsized influence on health outcomes. Some patients teeter on the border between obvious health and visible risk. (These borderline patients often end up costing the most to treat.)

Data-driven assistance for healthcare professionals

KenSci tackles these problems head-on with a data-driven approach in the purest sense. We develop mathematical models that account for hundreds of variables — and the connections among them — ranging from genetics, demographics, income, and psychosocial factors to living situation, childhood illnesses, and even the patient’s postal code. Health-monitoring devices also provide a stream of data for analysis. That’s where machine learning can help. By using Microsoft Azure Machine Learning, we can build our models by recognizing patterns in both static and real-time medical data, and the results so far have been impressive. We work with some of the largest health systems in the United States and Asia, and we help more than 17 million patients reduce their risk and cost of healthcare.

The right cloud choice for the right job

We chose Azure as our cloud service for several reasons. The most important was security. When we are dealing with our customers’ most personal data, security is non-negotiable. Azure helped us manage healthcare compliance and simplify the security experience better than all other cloud platforms we evaluated.