Artificial Intelligence in Health Care

There are several avenues where artificial intelligence can make an impact in health care, and Frost & Sullivan has identified a few important areas that are already benefiting from advanced analytics platforms.

In medical school, students often confront the saying, “When you hear hooves, think horses, not zebras.” This line encapsulates the kind of structured thought process young doctors are encouraged to develop—reach for the simplest, easiest explanation for a diagnosis. Less likely diagnoses should only be considered after eliminating more reasonable diagnoses. This process of thinking is referred to as differential diagnosis, and the term “zebra” often refers to a surprising medical diagnosis. While this school of thought seeks to teach young doctors the importance of utility over accuracy due to their limited time and intellectual energy, artificial intelligence (AI) has no such constraints. The computing power of AI platforms means that they can be trained to specifically look for zebras while the doctors tackle horses.

The term AI has been used so often and indiscriminately that the extent of its capabilities is often lost. Frost & Sullivan categorizes advanced analytics platforms into four levels, in the increasing order of their capabilities and complexities, beginning with big data analytics. This refers to algorithms that can sift through large quantities of data and identify patterns that humans cannot comprehend. Machine learning, a degree higher in sophistication, is the process of teaching machines to create their own algorithms so that they think and act as the situation demands. Finally, there are artificial neural networks—algorithms that are inspired by how the human brain works and can be used by machines to interpret both structured and unstructured data. This is a particularly important feature because medical data comes in many shapes and sizes, such as medical images, laboratory test results, genomic data, prescriptions handwritten by doctors and, increasingly, as information from wearable devices and smartphones.

There are several avenues where AI can make an impact in health care, and Frost & Sullivan has identified a few important areas that are already benefiting from advanced analytics platforms. These include medical diagnoses, clinical decision support, pharmaceutical R&D and population health management. Provided below are brief profiles of companies providing AI-based solutions to some of these areas.

Zebra Medical Vision (Kibbutz Shefayim, Israel)

Elad Benjamin, CEO of Zebra Medical Vision, told Frost & Sullivan that the company was named as such to clearly indicate that it is metaphorically hunting zebras. Zebra’s Radiology Assistant is a product that has the ability to scan medical imaging scans from various modalities. The device then uses its proprietary AI algorithm to scan the images, constantly comparing them with reference images to capture any indication of disease that is not apparent to the human eye. For instance, the tool can scan a patient’s chest X-ray and compare it with hundreds of archived scans and look for signs of emphysema, lung cancer or other lung diseases. It can then quantify the volume of emphysema in comparison with the lung volume and predict the stage of disease. Similarly, it can work with chest CT images to diagnose pulmonary hypertension. Zebra Medical Vision has focused on the detection of liver, lung, brain and heart diseases to showcase the capability of its machine learning platform.

Enlitic (San Francisco, Calif.)

Enlitic has developed an eponymous tool that was designed using the human brain as a model. Just as millions of neurons simultaneously sense and analyze varied sensory signals, Enlitic’s artificial neural networks are designed to analyze information from a diverse range of medical data, including images, patient data, and population health charts. Enlitic is programmed to learn by looking at hundreds of legacy records and it adapts and evolves as new data becomes available, thereby processing emergent and evolving trends as well as hard data. Enlitic is currently partnering with hospital and research groups to address real-world radiology problems. The company is also interested in applying its tools to pharmaceutical development, clinical trial testing and other data-driven health care applications.

NuMedii (Menlo Park, Calif.)

NuMedii’s proprietary Artificial Intelligence for Drug Discovery (AIDD) platform was developed in 2008 at Stanford University. NuMedii leverages big data analytics and artificial intelligence to aid the discovery and development of drug candidates. The technology uses advanced analytics to identify target biomarkers for a particular disease by analyzing millions of clinical, biological and pharmacological data points. NuMedii then translates these predictions into novel, de-risked drug candidates that pharmaceutical companies can act upon for development and commercialization. Numedii has partnered with the intellectual property consultant Thomson Reuters to combine its big data technology with the intelligence provided by Reuters on more than 365,000 compounds in order to systematically identify new applications for existing drug compounds.

The Road Ahead

AI has taken roots in the health care industry, perhaps deeper than in any other industry. Epidemic scares such as the 2014 Ebola outbreak or the 2015 Zika virus epidemic have further underscored the urgency of population health management. AI tools can efficiently analyze large population data sets to design programs for preparedness. AI will also play an important role in designing therapies based on an individual’s specific requests, aligning perfectly with the growing trend of personalized medicine. 

Copyright © 2017 Frost & Sullivan