Editor’s note: Kanupriya Agarwal, MD, is Digital Health Physician, Icahn School of Medicine at Mount Sinai. This blog originally appeared on OliverWyman.com
Although healthcare’s first use case of artificial intelligence was nearly 70 years ago, its real focus and impact has only recently being actualized. Quantifying artificial intelligence’s direct impact through both key performance metrics and via the explosion of data analytics within the last decade has advanced implementation of machine learning and artificial intelligence solutions. But pinpointing artificial intelligence’s potential to shape the health market is yet to be determined.
Here at New York City’s Mount Sinai Health System, Deep Patient, an artificial intelligence-based electronic health record (EHR) data mining tool for at-risk patient identification and readmission prevention and reduction, is demonstrating interesting results.
Over on the West Coast, Silicon Valley artificial intelligence healthcare startups are chock-a-block of venture capital dollars, precisely due to the coalescence of this presence in this particular area.
With corporate giants like Google and Amazon foraying into health artificial intelligence, applications like Alexa and Google Home are likely to become commonplace in daily healthcare tasks and decision-making. For instance, Google is building end-to-end artificial intelligence solutions across areas including data generation, disease detection, and disease / lifestyle management, initially targeting specific diseases like heart disease, eye disease, diabetes, Parkinson’s disease, and Multiple Sclerosis, with others to be considered later. Although Google has patents in these areas, it has been able to show in research that their algorithms are at par with ophthalmologists in areas like diabetic retinopathy detection.
Regarding other examples of new artificial intelligence advancements propelling the industry forward, last year, the FDA announced pre-certification for digital health companies, and now it has come out with extended support for the use of artificial intelligence in medicine and drug development. A new FDA incubator called INFORMED (Information Exchange and Data Transformation) is being developed, which will allow the FDA to initially focus on devices that work on forwarding cancer treatment and drug development. This will likely help speed up discoveries in the pharma sector, which previously took decades and was a costly affair.
So, progress and advancements in artificial intelligence and machine learning are everywhere around us. But how much traction have we actually gained? Below is an analysis of where opportunities to scale and close the gap exist.
“There is a specific ongoing rumor that artificial intelligence will replace physicians such as myself over time. The fact is technology is only supplementary and can augment the work of healthcare providers, and free them up from menial and repetitive tasks with more quality time to spend with the patient.”
Artificial intelligence in healthcare is booming
According to a report released last year from Accenture, the artificial intelligence market will spike, estimated to grow 40% annually through 2021. Due to maximal cost savings potential, the hottest top five health artificial intelligence applications are:
Regarding these estimated dollar amounts (note that EHR Clinical Decision Support is still naive), Accenture reported combined artificial intelligence key clinical applications can potentially save $150 billion in healthcare savings by 2026, in the US alone. And artificial intelligence may fill 20 percent of unmet clinical need through 2026 nationwide.
Consider also the recent explosion of partnership deals from those looking to be “in it to win it.” In just over five years (Q1 2013 – Q1 2018), 481 deals in healthcare artificial intelligence startups have garnered $3.6 billion in investor capital. Of that, just Q1 of 2018 has seen a record number (nearly 70) deals, with about $500 million in funding already invested. Besides, venture capital in artificial intelligence digital health companies has more than doubled between 2015 and 2017, and is predicted to continue on that positive upslope.
Regarding competition for overall artificial intelligence on a global scale, European venture capital firm Asgard notes the US is the global market leader with 40 percent market share, followed by China second, and Israel third, the latter two with an equal 11 percent market share. The top cities worldwide (in this order) are: Greater Silicon Valley area, London, Tel Aviv, New York (second natiowide), Beijing, Boston (third nationwide), and so forth.
The artificial intelligence mindset is shifting
Conversations about artificial intelligence usually include some reference of robots taking over the world. There is a specific ongoing rumor that artificial intelligence will replace physicians such as myself over time. The fact is technology is only supplementary and can augment the work of healthcare providers, and free them up from menial and repetitive tasks with more quality time to spend with the patient. It may rather replace only those physicians who don’t adapt or learn how to work with artificial intelligence over time.
“A new era in medicine is awakening, and we are at a seismic shift with artificial intelligence penetrating every aspect of life on the globe at rapid pace and scale, just like with the birth of the World Wide Web.”
In addition, there has been a valid fear of bias in the algorithmic-clinical decision making, leading to wrong outcomes which could cost a patient’s life. Due to inadequate data and errant algorithmic pathways, bias in decision-making is a common outcome where datasets are not large enough. Therefore, the larger the dataset, the more accurate the artificial intelligence algorithm. Adequate testing for bias during the R&D phase is crucial, such that after it leaves the lab, the implications on society are not harmful and more ethical, in addition to being valuable. Besides, AI keeps learning with time as the data grows and biases are auto-eliminated.
Some are taking charge in this arena. For instance, Accenture’s Citizen AI concept of advocating for and promoting a responsible artificial intelligence mindset, which can morally contribute back to society – thereby benefitting humanity, in addition to advancing entrepreneurship productivity and proving patient outcomes – is an important one to remove the risk of deeming the technology as irresponsible and non-transparent.
Despite this boom, executives lack expertise to leverage artificial intelligence efficiently
There’s a popular argument from many venture capitalists who claim artificial intelligence founders don’t correctly differentiate between artificial intelligence, machine learning and deep learning in their products or don’t actually have deep artificial intelligence technology insight applied into their product.
In a poll of over 50 artificial intelligence healthcare executives from across the world, nearly 50 percent of those interviews agreed artificial intelligence adoption across the healthcare industry needs to be stronger linked to ROI from an artificial intelligence / machine learning investments perspective. This was concluded to be the number one challenge to artificial intelligence adoption in the US, far ahead of the international artificial intelligence healthcare market. Secondly, in both the US and international artificial intelligence health markets, healthcare providers said although they understand the value of artificial intelligence, they lack the technical expertise to make the product viable.
Adding fuel to the fire? A lack of female decision makers. According to Rock Health’s analysis, gender parity in healthcare reflects poor participation of women as Chief Executive Officers – only at 8% in artificial intelligence / machine learning digital health companies. But legend has it more investors are looking to support and connect with diverse and female founders. In fact, the majority of venture capital firms (61%) investing in digital health themselves have zero female partners!
These limiting mindsets arguably affect trust and make investing difficult. For instance, a Frost and Sullivan analysis shows 50% cost reduction with artificial intelligence implementation in healthcare and more than 50% improvement in patient outcomes. Executives need to be more aware of the cost advantages that come with taking a different approach to artificial intelligence. Leaving money on the table is no longer an option.
A new era in medicine is awakening, and we are at a seismic shift with artificial intelligence penetrating every aspect of life on the globe at rapid pace and scale, just like with the birth of the World Wide Web. With multiplying acquisitions and attractive value propositions, it only goes to show that it has been easier and cheaper for giant corporates to acquire smaller companies/startups in healthcare, than to build a whole new vertical and go through the development cycle versus leveraging the already available expertise within a short span of time. For health artificial intelligence startups and venture funds alike, the opportunity orchard is blooming and the time is ripe to step in to build a lasting foundation which will create a legacy for the changing landscape of medicine in the subsequent decades. We’ve heard of concepts like the cloud hospital, the digital doctor, and the virtual patient. The question is, are you in?
Keep up with latest coverage on digital healthcare by signing up for Internet Health Management News today.