Diving Deep Into HealthCare Data
By: Sanveer Dhanju & Andrew Truong
The Ivey Business Review is a student publication conceived, designed and managed by Honors Business Administration students at the Ivey Business School.
Nearly $100B is spent nationwide on medical imaging each year, according to the United States Government Accountability Office. Radiology is the largest contributor to US hospital profit margins, totaling 37% of profits - 3x greater than cardiology, the next largest contributor. With advanced imaging becoming increasingly important in patient diagnosis, there has been explosive growth in the amount of imaging performed since 2002. With the average American hospital operating on profit margins of just 3%, reductions in costs and increases in quality of care translate into significant improvements for their bottom line.
Since the Affordable Care Act was passed under the Obama administration, hospitals are subject to Medicare reimbursement penalties of up to 3% for excessive readmissions of patients with certain conditions. In context of hospitals’ low profit margins, this penalty is spurring healthcare providers across the country to improve the accuracy of their diagnoses. GE Healthcare, a world leader in diagnostic technology equipment manufacturing, is strategically positioned to take advantage of advances in deep-learning analytics technologies to help hospitals better diagnose patients and reduce readmissions. To do this, GE Healthcare should acquire San Francisco based deep learning analytics company Enlitic.
Deep Learning with Enlitic
It is estimated that in today’s hospitals, over 90% of all medical information available is in the form of medical images. The availability image format data could radically change how a diagnosis is determined, and the use of artificial intelligence in healthcare has the ability to advance the field of evidence-based medicine. Tools that are currently available to help physicians with diagnoses are rudimentary and rely on broad, oversimplified assumptions. However, deep learning, a form of artificial intelligence, is at the forefront of changing the way radiologic data is analyzed. Deep learning is the use of technology to train artificial neural networks on large quantities of data and getting them to provide inferences about new data. New technology startup, Enlitic, is making breakthroughs in this field by using algorithms that can analyze radiology images to diagnose conditions with unprecedented speed and accuracy. Their technology has the potential to reduce the number of misdiagnoses in the US, which currently stands at 12M adults annually. Enlitic’s algorithms seamlessly integrate into a radiologist’s workflow by running in the background as images are taken. These algorithms are ‘trained’ to recognize anomalies and confirm the initial diagnosis of the radiologist using symptoms, patient history, lab test results, prior images, and any comparable patient cases. Anomalies between a physician’s diagnosis and Enlitic’s analysis can be quickly cross- analyzed to verify if the initial diagnosis was correct.
GE Healthcare
GE Healthcare is one of the largest global healthcare technology providers, with sales of over $18B in fiscal 2014 and profits of over $3B in fiscal 2014. Despite the success of GE Healthcare since its inception, its sales and profits have stagnated since 2012 due to a decline in healthcare spending and the strength of the US dollar. Management has made it an objective to lead in technology innovation, focusing on technologies that allow health facilities to improve productivity. For example, GE acquired API Healthcare in 2014, a healthcare-specific workforce management software company, for $300M in an effort to incorporate API’s staffing offerings into their core product line. GE should continue its expansion into healthcare technology and acquire Enlitic in order to inherit its deep learning technology.
Dr. Watson is on the Move
Enlitic is not the only company with deep learning technology capabilities. The entrance of IBM into healthcare analytics, with Watson Health, poses a large threat to Enlitic. IBM’s advantage is the amount of data they have access to, since the amount of data is determinant of the value of deep-learning algorithms. IBM’s Watson Health Unit has been on an acquisition spree, including a $1B buyout of Merge Healthcare in October 2015 to access Merge’s 30 billion pieces of radiology data. This acquisition foreshadows IBM’s entrance into the radiology space, and its ability to become a formidable direct competitor to Enlitic. Using Merge’s data, Watson Health can increase their effectiveness at interpreting future radiological data in real time and providing correct diagnoses. Merge Healthcare is one of the largest providers of picture archiving and communications system (PACS) technology. It is currently used in over 7,500 different healthcare sites in the US to manage, store and distribute the workflow associated with all forms of diagnostic techniques such as MRI, computed tomography (CT), X-rays and ultrasounds. Not surprisingly, Merge’s biggest competitor in the PACS space is GE. GE is in a strategic position to use the radiology images it has collected to develop a deep learning analytics technology. The global demand for PACS systems is expected to grow at a compound annual rate of 10% from 2014 to 2019, showing increasing demand for new PACS systems in hospitals and clinics. GE’s ability to introduce greater analytics ability to its PACS systems will help it capture more market share in this growing industry.
Readmission and Misdiagnosis Don’t Go Unnoticed
Less than half of medical decisions in the US comply with evidence-based standards. In fact, 1 in 5 hospitalized patients on Medicare in 2013 were readmitted within 30 days of leaving the hospital. This costs Medicare $26B, 65% of which could have been avoided with better care and diagnostics. Large insurers are bearing the majority of the cost associated with misdiagnosis and non-optimal treatment, which is why there has been a push by the US government to reduce the incidences of readmission. An initiative the government has launched is charging reimbursement penalties to hospitals with too many incidences of readmission. Recently, only 2,665 out of a total of 3,464 hospitals, or 77%, were charged a Medicare reimbursement penalty. The government has also required hospitals to switch to Electronic Health Record technology by the year 2017, which has historically brought an average cost saving of 9.7% for hospitals. This move is pushing almost all hospitals in the US to electronically store their healthcare data, many of whom are opting for the cloud, making their data readily assessable and prime for integration with healthcare analytics systems.
GE Healthcare, the Catalyst for Enlitic Technology
The more data Enlitic’s algorithms have to train with, the more they will learn, and the smarter they will become. Currently, Enlitic is clinically ready to diagnose 12,000 unique conditions via imaging, but CEO Jeremy Howard envisions Enlitic applications in various other forms of personalized medicine. Conversely, GE Healthcare has access to one of the largest databases of medical images in the world, something that Enlitic needs to become proficient in. Enlitic recently entered into a partnership with Capitol Health in order to test its algorithms, which have been shown to perform better than top radiologist during a small-scale trial, but have never been tested in a large- scale trial. Capitol Health is a series of 51 clinics in Australia that provides services such as MRI, X-rays, CT scans and ultrasounds, etc. This is a good start for Enlitic, however it will not be nearly enough for Enlitic to compete with Watson Health and its 30B pieces of radiology data from Merge Healthcare.
The Time is Now
The time for GE to acquire Enlitic is now. With proven technology on a small scale, Enlitic has raised $15M in funding to date. To contrast, IBM has spent over $1B to develop Watson and more than a $1B acquiring companies such as Merge and Explorys in order to access large sets of relevant clinical data. There will be no costs for GE to acquire the datasets Enlitic needs to grow. GE has been collecting data since the launch of its PACS system in 1992, while Merge’s system only launched in 2001. This means GE has a bigger set of data at its disposal for Enlitic to learn from and improve. GE has an opportunity to acquire Enlitic’s technology at a discount compared to other deep-learning companies that have much higher valuations but only offer solutions tailored to radiology data. GE can help Enlitic build a solution similar to Watson Healthcare. This type of investment aligns well GE’s goal of investing $2B in data analytics in 2013 over a period of five years.
If GE does not act while Watson Health is still in development, it is likely that Watson will become the industry standard. Down the road, GE will either have to license IBM’s technology for a steep fee or play catch-up in an industry in which time is critical. GE could also lose market share to Merge Healthcare and its PACS technology, which will no doubt become attractive to healthcare sites with Watson analytics capabilities over the next few years.
Market and Markets predicts the global healthcare analytics market to grow from $5.8B in 2015 to $18.7B in 2020; a compound annual growth rate of over 26%. There is no doubt that this is a market GE wants to be a part of. GE Healthcare has the ability to inject a significant amount of value into Enlitic by providing it with the data it already has, as well as access to a large resource base from its parent conglomerate to make further investments into Enlitic. Based on Enlitic’s most recent $10M series B round of funding in October 2015 and the $15M raised to date, the company could be acquired by GE for between $50M and $200M. With the infusion of GE Healthcare’s data, Enlitic’s technology could be developed to very accurately diagnose patients based on radiology and other data creating immense value for healthcare facilities. The value added to the company could be grown between 2 and 10x its current day value within five years. The could generate compound annual returns between 15% and 58%, far exceeding the high end of GE’s weighted average cost of capital of 10%. There is a risk that the Enlitic technology will not be able to meet its growth expectations resulting in a value drop. Considering the market growth potential and the $2B GE has allocated solely for data analytics R&D, the benefits of an investment in Enlitic far outweigh the potential downside. This acquisition has the potential to create shareholder value of between $7.5M and $117M in the first year following the acquisition and between $13M and $535M in the fifth year of the venture with total value created of between $50M and $1.8B over five years depending contingent on the acquisition price and valuation after five years.
Once GE has developed the Enlitic technology, it can choose to keep it in-house to gain a competitive advantage or to license the technology similar to what IBM has planned for Watson Health. Alternatively, GE could spin out Enlitic entirely at a significantly higher valuation once the technology has been refined for a large one-time profit.
The Future is Here
The idea of a computer being able to take patient data, come up with a diagnosis and suggest a treatment is now within reach. GE Healthcare can position itself to be a major player in this fast growing market and deliver significant value to healthcare systems around the world simply by unleashing the power of the data it already has. GE will not only have multiple ways to profit from the acquisition, but it can be a leader in the development of healthcare artificial intelligence. The purchase of Enlitic will allow GE to be one step closer to its mission to “invent the next industrial era, to build, move, power and cure the world”.