RSNA18: Radiologists embrace their fate, choose to engage/encourage AI vendors rather than fear/flee them

Key Takeaways

• Fear of AI has given way to exuberant optimism/anticipation/impatience.
• Many vendors are trying to pivot their legacy products as next-generation AI platforms/marketplaces.
• Confidence in AI performance will only come with independent validation.
• Cloud has faded to the background.
• Not everything about AI is good news, adversarial attacks could bring about a new era in cyber-security concerns for providers.

Over the course of the last 18 months artificial intelligence (AI) has matured to the point where there are several viable vendor options for nearly every use case.

AI dominated every aspect of the annual gathering of the Radiological Society of North America (RSNA18) in Chicago. Self-described ‘machine learning’ vendors with a presence on the conference floor more than doubled from 49 in 2017 to over 100 in 2018, 25 of which were first-time presenters.

I moderated a panel hosted by Life Image on practical uses cases of imaging AI and was blown away by the conversation that ensued, particularly what I learned about how the veteran radiologists feel about being “replaced.” During the question period, a senior radiologist approached the microphone to address a comment made by a more junior radiologist on the panel which he interpreted to be too pessimistic about the potential for AI. To paraphrase the elder, “Listen here sonny, you are too young to fully appreciate what you don’t know, and you don’t know how many mistakes you are truly making on a day to day basis. 1-2 percent error rate due to fatigue alone. WE NEED AI to save us from ourselves.”

Not all old school radiologists are so optimistic: “When you’re going up the ride, you get excited,” noted University of Chicago radiologist Paul Chang said during his workshop on AI. “But then right at the top, before you are about to go down, you have that moment of clarity—‘What am I getting myself into?’—and that’s where we are now. We are upon that crest of magical hype and we are about to get the trench of disillusionment… It is worth the rollercoaster of hype. But I’m here to tell you that it’s going to take longer than you think.”

The cloud fades to the background

Last year, the major cloud vendors each had a significant footprint at RSNA, but this year the two largest, Amazon and Microsoft, were nowhere to be found. Only Google Cloud had a significant, if smaller than last year’s, presence. Donny Cheung, one of the Google Cloud team leaders, was on the panel I moderated and his message to the imaging community could be boiled down to two words: storage and compute. No dashboards or toolkits or tensorflowing, just storage and compute, a smart and refreshing strategy amidst the obvious feature creep many other vendors suffer from.

Over the course of the last 18 months artificial intelligence has matured to the point where there are several viable vendor options for nearly every use case.

What’s Facebook doing here??

While it was surprising that Amazon had no noticeable presence, it was even more surprising to find Facebook making news on the conference floor. Facebook AI Research (FAIR) has partnered with the Center for Advanced Imaging Innovation and Research (CAI2R) in the Department of Radiology at NYU School of Medicine and NYU Langone Health to release the fastMRI, an open source dataset for training and testing machine learning algorithms to reconstruct MRI images.

This offering is roughly equivalent to similar X-Ray and CT datasets released by NIH. Given that algorithms ALWAYS significantly outperform on all metrics against the data used to train them versus new data, the industry needs independent validation of AI claims so it is unlikely that Facebook moves the needle with this offering.

Platforms abound

PACS vendors want to get in on the AI action by positioning their existing products as AI marketplaces or platforms (Philips HealthSuite Insights, PureWeb, LifeImage, GE Edison, FujiFilm REiLI, Nuance AI Marketplace, Blackford Analysis). Nuance has shown there is a viable market for these platforms, counting 40 startups and health systems among user groups for its marketplace. There is no shortage of startups taking this approach (MDW, Envoy.ai, Medimsight, Lify, Fovia). Imaging hardware vendors refused to be left out too, with many partnering with AI vendors to embed their algorithms on the “edge.”

AI is global

International AI startups, particularly from Israel, China, and South Korea, stood out from the crowd in terms of their approach to product design, but only the companies from Israel have been able to break into the US market so far. One Korean company voiced frustration with the FDA, saying it couldn’t understand what was wrong with their application. I wonder if it underestimates the importance of using data from US patients to validate their algorithms?

Some cautionary tales were told…

Not everything we learned about AI at RSNA was positive. A paper presented at the conference showed that neural networks could be used to insert malignant features into mammograms giving a false positives, and then reverse the alterations without detection. Even scarier, it took about 680 images to train the algorithm that executed the adversarial attack. Cyberattacks have been increasing in healthcare over the last couple years, but mostly just for taking data hostage and demanding ransom to get it unencrypted. This type of attack would represent a frightening new paradigm in cyber-vulnerability, and it is certainly not difficult to imagine ways this could be exploited to make money. It could be used for a different sort of ransom, with every image appearing to show cancer until a ransom is paid and the adversarial attack is reversed. Another conceivable way this type of attack could be exploited would be falsifying data for clinical trials.

This post was originally published on the Chilmark Research blog.

2016 ONC Annual Report to Congress

Below you can read the full report recently delivered to Congress by the Office of the National Coordinator for Health Information Technology (ONC).

I think it is well worth spending 20 minutes quickly catching up on what the government has done to date to support the adoption and development of electronic medical records by the health care system, especially in light of the coming transition to a new administration with new priorities. My biggest complaint is the continued emphasis of "interoperability" as a goal when in reality it is too abstract to be tangibly realized. While I certainly see value for physicians and providers in the ability to view a patient's longitudinal medical record, I fail to see why it is necessary or even optimal to demand EMR vendors integrate this feature into their platforms.

 

Best of Goldman Sachs digital health market forecast

Below are a select few of the best insights 

Commercially, we forecast a near-term revenue total addressable market (TAM) of $32.4 billion with 45% coming from remote patient monitoring; 37% from telehealth; and 18% from behavioral modification. Chronic disease management falls entirely in remote patient monitoring and accounts for the majority of our TAM projections. At present, we approximate that total revenue is in the several hundred millions with a wide range of company performance, ranging from as low as $5 million to just under $100 million. We acknowledge that the current profit pool is insignificant; however, advancements in technology and sweeping changes to the US healthcare operating environment make us confident in the path to adoption and future revenues. 

Implications for incumbents in a digital health ecosystem. 

Beyond the revenue opportunity associated with digital health technologies, potential systems savings represent a major incentive to adoption – we peg the total savings opportunity (TSO) at $305 billion. When discussing these savings, we are referencing the dollars that we have calculated can be saved through digital health adoption. Taking this one step further, we have sought out to identify the amount of unnecessary/repetitive care or waste in annual US healthcare spending. As is the case with revenue, the vast majority of the opportunity resides in chronic disease management, where we estimate the potential savings could amount to $200 billion, or 66% of our TSO.

In sum we quantify the outstanding “Total Savings Opportunity” (TSO) brought about by digital health to be ~$305 billion (at 2015 chronic disease incidence levels). We expect the penetration of this number to ramp slowly given the early stage nature of these technologies, but have conviction that universal adoption of these systems can cause a significant reversal of trend in healthcare spending growth over time. This number excludes “behavior modification” platforms which stand to eliminate many chronic diseases in general, but on an extremely long term basis.

We build our “Total Addressable Market” (TAM) projection of $32.4 billion in digital health sales as a sum of the most viable near-term opportunity sets (i.e., from the bottom-up). This TAM is comprised of $15 billion in remote patient monitoring, $12 billion in telehealth, and $6 billion in behavioral modification. An important element of these assumptions is that we have isolated our analysis to the most prevalent and highest cost chronic diseases. The logic for this market segmentation rests in the fact that average selling prices (ASPs) are generally in the range of “several hundred dollars.” Therefore, the size of the target patient populations need to be large in order to support the meaningful upfront capital investments needed to validate clinical outcomes and justify high-risk VC commitments. Continuing on that tact, we size telehealth (comprised of standard visits and psychological counseling services) as a $12+ billion opportunity and behavior modification (centered on obesity) as a $6 billion market. In the case study section later in the report, we detail the assumptions behind each component of our TAM build.

It is also important to note that our market sizing estimates represent what we view as a tangible and realistic near-term opportunity for the HC IoT (as we have defined it) and not a “blue sky” or bull case scenario. Framed quite simply, we have sized the market for products that have a targeted aim, a wide addressable market, an operable platform, and drive meaningful and clinically measurable patient outcomes. To caveat our market sizing exercise, we recognize that we exclude a substantial portion of the offerings that comprise the whole of “digital health” (namely consumer wearables, IT solutions, and data-analytics platforms). Each of the aforementioned, individually, could create multi-billion-dollar markets (Healthcare IT, broadly speaking, already has).

Click to view larger image. 

COPD/Asthma Medication Adherence and Disease Management

Propeller Health: inhaler-based sensor to track puffs and prevent episodes

 Invasive Health Failure Monitoring

St Jude Medical's CardioMEMS: wirelessly tracking pulmonary artery pressure

 Non-invasive Heart Failure Monitoring

Vivify Health: Intermountain Health Pilot

Smartphone Enabled Continuous Glucose Monitoring (CGM) 

DexCom SHARE: keeping tabs on diabetes, as easy as checking your email

Remote Cardiac Arrhythmia Diagnosis 

iRhythm ZIO: the wearable patch that provides a diagnosis, faster

Smartphone Enabled ECG for Monitoring and Arrhythmia Diagnostics  

AliveCor: transforming your mobile device into a medical grade ECG monitor

Telehealth 

Doctor on Demand: bringing the physician into the living room

Behavior Modification: Digital Weight-loss Management and Diabetes Prevention 

Omada Health: Diabetes

Cancer Moonshot 2020 is our best hope yet to conquer the Emperor of all Maladies

Dr. Soon-Shiong in his LA office where the walls are literally filled with his grandiose vision for the future of medicine.

In the most significant proposal of President Obama's final State of the Union he announced Patrick Soon-Shiong has teamed up Vice President Joe Biden and the largest pharmaceutical companies in the world to launch Cancer Moonshot 2020 in a unified effort to cure cancer once and for all.  Vice President Biden published his vision for the Moonshot project today on Medium.

"There are unique times in history when events and advancements in technology converge to elicit a quantum leap in medical care. This is not only a unique time, but also a unique inflection point in the history of cancer,” said Dr. Patrick Soon-Shiong, M.D., Founder and Chief Executive Officer of NantWorks and the Chan Soon-Shiong Institute of Molecular Medicine. “The era of immunotherapy has taken the oncology world by storm. For the first time in 40 years there is a glimmer that we may be able to win this war against cancer. Large pharma and biotech companies are developing dozens of agents to activate the immune system. The problem is that while these drugs are being developed individually in silos by each entity, they need to act together when it comes to activating the immune system. If we follow the current path of drug development, it may take 40 or 50 years before we have worked out the right cocktail combination and countless lives will be lost as a result of this inefficiency.”

• Nation’s first insurance coverage of next generation whole genome sequencing and proteomic diagnostic platform in cancer patients‎. 
• Next generation sequencing ‎and precision medicine evolving from research to the clinical trial and cancer care setting
• Coalition to design, initiate and complete randomized clinical trials at all stages of cancer in up to 20 tumor types in as many as 20,000 patients in multiple phase 1 to 3 trials by year 2020‎
• Beneficiaries and patients will undergo next-generation molecular sequencing and gain access to over 60 novel and approved molecules to be tested as immunotherapy combinations in 20,000 cancer patients with cancer across all tumor types: The QUILT Program

The QUILT (QUantitative Integrative Lifelong Trial) program is designed to harness and orchestrate all the elements of the immune system (including dendritic cell, T cell and NK cell therapies) by testing novel combinations of vaccines, cell-based immunotherapy, metronomic chemotherapy, low dose radiotherapy and immunomodulators -- including check point inhibitors-- in patients who have undergone next generation whole genome, transcriptome and quantitative proteomic analysis, with the goal of achieving durable, long-lasting remission for patients with cancer.

The multiple Phase 1 and 2 protocol designs will be a collaboration between academia, pharma, and clinical scientific experts in immunotherapy in accordance with the recent published U.S. Food and Drug Administration guidance of “Co-development of Two or More New Investigational Drugs for Use in Combination.” Multiple companies are currently exploring first-in-human clinical trials as part of agreements between government agencies, including Immunology Branches of NCI, academia and community oncologists. Multiple randomized Phase 2 trials testing genomically and proteomically informed novel combinations of immunotherapy agents, will pave the way to identifying cancer therapy combinations with the lowest toxicity and the highest quality of life.

Both academic cancer centers and community oncologists will participate in the QUILT Program to enroll 20,000 patients by 2020.  The QUILT Program will be stratified across multiple Phase 1 and Phase 2 trials, addressing up to 20 tumor types including breast, lung, prostate, ovarian, brain, head and neck, multiple myeloma, sarcoma, pancreatic cancer, among others. Pharmaceutical and biotechnology partners have made an unprecedented commitment to make more than 60 novel immunotherapy, targeted therapy and chemotherapeutic agents available to be combined across multiple tumor types.

Below are videos of the press release at the JP Morgan Conference in San Francisco and a description of the project.

Press Announcement - Cancer MoonShot 2020 from NANTHEALTH LLC on Vimeo.

Animation - Cancer MoonShot 2020 from NANTHEALTH LLC on Vimeo.

Digital Health News and Notes

Bill Gates, Jeff Bezos and Illumina are among the investors in the $150mm seed round of a new cancer diagnostics startup subtly named GRAIL. 

Once promising telemedicine startup HealthSpot (an Ohio-based startup that I first covered in 2012 while I was a full-time blogger at iMedicalApps.com) has squandered almost $30mm in venture capital and partnerships with Cleveland Clinic and RiteAid announcing it will enter the Digital Health Deadpool. To add insult to injury, HealthSpot's leading competitor Chicago-based Higi landed $40mm in new funding this week.

On the investor-side of the table former-Senator Bill Frist has unveiled a new private equity fund based in Chicago, Orbimed announces a new $950mm fund and Partners HealthCare expands it’s life sciences innovation fund to $100mm. The American Medical Association has invested $15mm in a new for-profit startup incubator based in San Francisco called Healthy2047. Not to be outdone, the Governor of Massachusetts has launched an almost cliche effort to make his state the "digital health hub of the world".

Bill Frist

On the personalized medicine front Flatiron Health closed a massive $175mm funding round led by Roche and NantHealth has acquired NaviNet and founder Patrick Soon-Shiong makes it clear he now has all of the pieces in place to move swiftly toward IPO.

According to data from both Rock Health and Startup Health venture funding was down during 2015 in digital health for the first time since data on investments in the sector have been tracked.

StartUp Health Insights Report: 2015 Year End from StartUp Health

Epic Systems is the startup Dragon that eats Unicorns for breakfast

It occurred to me the other day that I have never seen Verona, WI-based Epic Systems mentioned on any of the lists of Unicorns (Cruncbase; CB Insights; Fortune). This is puzzling considering Epic has dominated the electronic medical record market for more than a decade, pulling in well over $1B in annual revenue and growing at breakneck speed creating a quarter of all new jobs in Dane County over the last decade.

Epic is particularly popular amongst the prestigious academic medical centers like Mayo Clinic, Vanderbilt University and Partners Healthcare, as well as with mega-systems like Kaiser Permanente and Providence Health. They may have lost the coveted DoD contract to archrival Cerner, but they won nearly every other high-profile contract up for bid.

Anybody who has visited Epic’s corporate headquarters recently can speak to the enormity of what Judy Faulkner and her right hand man Carl Dvorak have built just outside of Madison, WI. When roaming the campus it’s easy to forget you are in the middle of Midwestern farm country because the company has built a culture teeming with a youthful free-spirited ambition that feels more like Silicon Valley. The selected images below show just a small taste of the fantastical environment that is the Epic campus.

If the only prerequisite to joining the Unicorn club were a valuation of $1B+ Epic qualified back when the number of members could still be counted on two hands. But clearly that is not enough otherwise Epic would be counted. It’s hard for me to imagine they have simply been overlooked by the analysts.

I wonder if the unspoken requirement to join the Unicorn club is bending over to get screwed by the venture capitalists? After all, the term was originally coined by venture capitalist Aileen Lee of Cowboy Ventures, and the manipulative genius of the term is that somehow VC have been able to convince the new generation of “tech bros” in Silicon Valley that achieving Unicorn status is the new IPO.

Judy Faulkner has never taken a penny from investors. Instead she has built a product capable of supporting itself with revenue from day one, and personally spearheading all deals.

Is there another example of a software company growing to the scale Epic has achieved without taking any outside investment? I can’t think of one.

Most entrepreneurs and VC would tell you what Epic has achieved is impossible and the only way to achieve scale and sustain growth is to take venture capital and spend it aggressively.

No venture capital and no sales team!! Epic is truly such a unique company it would be demeaning to call it a Unicorn. I think it belongs in a category by itself — the first and only startup Dragon.

---

Survey Shows Consumers Unlikely to Share Data with Tech Companies

The American public does not trust technology companies with personal health data, according to a survey from Rock Health, a venture capital firm focused on digital health. The survey, of 4,017 people, found that only 8 percent said they would share health data like medical records and lab results with “a technology company.” There was a huge gap between that figure and the number who said they would hand their health history over to a research institution (36 percent) or to their own doctor (86 percent). When asked whom they would share their DNA data with, the responses were similar.

from Pocket
via IFTTT