Google is launching a set of tools meant to make medical images more interoperable and help organizations develop artificial intelligence and machine learning models around them.

Billions of medical images are scanned each year, and imaging data makes up 90% of all healthcare data, according to research from Cornell University.

The images have been a focus for efforts looking to leverage technology to improve the quality of care delivery, with hopes that AI can mine insights and arrive at diagnoses faster and more accurately than human clinicians, while lowering the workload on radiologists.

Google’s new Medical Imaging Suite includes cloud-based file storage with secure data exchange capabilities, automated image labeling through AI-assisted annotation, and tools to create training datasets for algorithms and accelerate the development of scalable machine learning models with less code, according to the tech giant.

Some Google Cloud clients have already begun using the suite of digital tools, including Hackensack Meridian Health in New Jersey, which is using it to de-identify images in order to build AI algorithms capable of predicting metastasis in patients with prostate cancer.

In addition, medtech company Hologic is using the tools to expand capabilities for its digital cytology platform for labs, which helps cytologists and pathologists identify precancerous lesions and cervical cancer cells in patients, Google said in a release on the product.

Funding for AI in healthcare has exploded over the past few years due to its potential to reshape how healthcare is delivered in the U.S., though potential benefits have largely yet to materialize. The technology faces roadblocks for use, including clinician buy-in, regulatory uncertainty and concerns about health equity and bias.

Radiology is one area of promise for AI, as proponents of the technology argue AI systems can analyze data that’s not visible to the human eye, making it a valuable diagnostic tool.

A number of healthcare organizations are implementing AI to create, analyze and make sense of medical images, including New York University Langone Health, which has partnered with Facebook’s AI team to develop an algorithm that allows an MRI to take just 15 minutes instead of an hour, and also developed a tool to predict the probability of breast cancer in MRI scans that’s capable of reducing unnecessary biopsies by up to 20%.

Dive Brief:

• A proposed rule the HHS announced Monday seeks to strengthen civil rights and nondiscrimination protections for patients and consumers under Section 1557 of the Affordable Care Act.
• A previous version of the rule enacted under the Trump administration limited its scope and power to cover fewer programs and services, officials said on a call with reporters Monday.
• The newly proposed version, which prohibits discrimination on the basis of race, color, national origin, sex, age and disability in healthcare programs, interprets Medicare Part B as federal financial assistance for the first time. 

Dive Insight:

The Obama administration first issued regulations on discrimination, but the Trump administration stripped a number of protections, notably those regarding sexual orientation and gender identity. 

Officials clarified on a call with reporters Monday that the new rule would reinstate such protections, and that it covers both providers and insurers that receive federal financial assistance. That includes telehealth providers and Medicare Part B, which covers some doctors visits, preventive care and outpatient care, among other services.

“Everyone in America should be able to get the care that they need from any health provider in the country, especially if that provider is receiving funding from HHS,” Secretary Xavier Becerra said on the call.

“We want to make sure that Americans are free from discrimination when they try to access the care that they need,” he said.

Other tweaks in the new rule include broadening language accessibility services and requiring covered entities to provide alternate communication formats for patients with disabilities or those who require auxiliary aids. Accommodations also must be made for people who do not speak English.

Another notable change includes language around clinical algorithms, prohibiting discrimination through the use of that technology in healthcare settings. 

Research has found AI algorithms used for decision making in patient care can be racially biased and lawmakers in the past have pushed for greater oversight.

The previous version of the rule under Trump eliminated the prohibition of discrimination based on gender identity, along with specific health insurance coverage protections for transgender individuals, according to an analysis from the Kaiser Family Foundation.

Shortly after that rule was published, the U.S. Supreme Court ruled in Bostock v. Clayton County that employers cannot fire employees on the basis of sexual orientation or gender identity.

The new rule would affirm covered healthcare entities cannot discriminate on the basis of sexual orientation or gender identity, consistent with the court’s decision, according to an HHS release.

After Athenahealth was taken private in 2019, the electronic health record vendor decided its R&D organization needed to focus more on its core business.

The vendor, which was struggling in an increasingly competitive health IT sector, elected to reorient its attention on the ambulatory space and pivot away from elbowing for market share in the hospital market with bigger rivals like Epic and Cerner.

As a result, management reached out to their clients to take stock of a day in the life of an ambulatory independent physician practice, and made a targeted series of investment bets to support what doctors said they needed. Athenahealth repurposed its Austin office into an R&D center focused on delivering a better patient and physician experience, zeroing in on any pain points in its existing tools.

That investment is starting to bear fruit this year, as Athenahealth plans to release "a bunch" of new products, including the Massachusetts-based vendor's first patient-facing app, chief product officer Paul Brient told Healthcare Dive on the sidelines of the annual HIMSS conference in Orlando on Tuesday.

Brient also shared his thoughts on the government's push toward interoperability and how it's affecting doctors on the ground, along with how 25-year-old Athenahealth plans to modernize its record to mitigate the burnout that's reaching a fever point among U.S. medical workers.

1. Consumerism in the driver's seat

   "Oddly, when I joined, there wasn't a lot of talk about patients. They're kind of important in the healthcare system, from my experience anyway," Brient, who's been at Athenahealth for about three years, said. "If you go to just about any healthcare system, or healthcare provider, it's kind of like going back in time. You have to call them up to get an appointment, you leave, they give you a printout of stuff, you show up, they give you a clipboard with a bunch of paper on it. I'm like, why are we doing this? It's 2022."

   Athenahealth plans to release an app for patients focused on patient engagement this year, aimed at extending the care experience beyond the clinical encounter. Currently, Athenahealth's patient portal works on a phone, but the vendor wants to move beyond that with a more targeted and comprehensive tool.

   "How do we make patient engagement a 365-day-a-year thing? Having an app will obviously help us do that," Brient said.

   The patient app will include secure messaging, appointment scheduling and management and financial bill pay services.

   But with increased interoperability spurred by the 21st Century Cures Act and recent regulations from HHS pushing the whole clinical record to the patient — something Brient noted is controversial among doctors, as patients can access notes and results without context before the clinician — Athenahealth wanted to create a more expansive app, beyond just the classic patient portal.

   The app will also let patients view and collect their information from connected devices, and will dabble in the care management space as well.

   The areas of digital patient engagement, coaching and management have taken off during the pandemic, as COVID-19 restricted access to a physical care team.

   The hope is that the app will be a platform for holistic engagement — a more modern, available version than Athenahealth's care management app AthenaWell, which has struggled with uptake.

   "One of the things we'll be doing this year is launching a care management service. We'll actually have nurse care managers helping our practices manage the patients," Brient said. "You can imagine coupling that with some AI chatbot for interaction, where patients get a question from a chatbot in the morning. 'Hey, how are you feeling today? '... 'Oh, I'm feeling terrible this morning.' And then the next thing that happens is a care manager calls you."

2. Split on info blocking regulations

   HHS in early 2020 released two sweeping regulations meant to push the industry toward free, unfettered health data sharing across different entities. The buildup to their publication was mired by controversy over patient access versus the security of medical data, along with how to best push entrenched IT operations toward a more open architecture.

   In April, the brunt of the regulations came into effect, requiring hospitals and certified health IT vendors to share data with patients and with each other. But physicians still have some worries about how they were enacted — as does Brient, the longtime software executive told Healthcare Dive.

   "There's certainly a camp that says just giving patients the lab results, the imaging studies, the notes that I write about them without any context or frankly me, as a doctor, having a chance to review them first, is going to create a lot of anxiety that's unnecessary," Brient said. "The counter is, I can go right now to a LabCorp and get any test I want, just about. Consumerism is happening in all kinds of different ways. If I wanted to go get an HbA1c test, I can go myself, order it and interpret it myself and Google it and figure it out. So I think there's just a tension and a change."

   This anxiety could also manifest itself in additional work for medical staff, which are reporting record-high levels of burnout as the pandemic enters its third year. Brient pointed to small-scale studies that suggest open notes seem to increase calls into the practice as people follow up with questions stemming from their data.

   "It's going to generate more work. Like everything, it will evolve, and we'll figure out how to make it effective," Brient said. "Personally, I think it's not good. I don't think it was particularly well thought out by the regulators."

   Brient noted that giving patients access to their records is paramount, but enacting the regulations as codified necessitates a distinct shift in how physicians write and disseminate their notes.

   It could result in the creation of some sort of private, protected, not-medical-record communication method between doctors, the executive suggested.

   "I wouldn't be 100% surprised if we invent a new concept that's not note, that's kind of like the private physician thing that physicians write other physicians," Brient said. "There's some things that if you're providing care to somebody in the hospital that you need to talk about, like 'Hey, the mother in law's crazy.' You don't write those things down, but they're really important things to know if you're the next person on shift. So I wouldn't be surprised if something like that gets developed."

3. Mitigating click fatigue

   As Athenahealth works to revamp its product around its ambulatory clients, the vendor is also taking steps to ameliorate burnout by personalizing records to specific physicians.

   EHRs swept onto the scene in healthcare in force a little under a decade ago, when the government mandated their adoption. Despite streamlining continuity of care, billing and administrative functions, doctors continue to say EHRs are a major pain.

   Athenahealth's EHR is popular among physicians, but "it's a little bit like the one they hate the least," Brient said. The goal is to make it like the iPhone. "You love that thing. You probably have a few things you don't like about it. But our goal is to make the EHR that way, and it's totally doable."

   Historically, EHRs have a one-size-fits-all workflow that physicians need to be trained on, which can be expensive both in terms of dollars and time. To try and fix this, Athenahealth is investing in the configurability of the EHR, specialty-specific workflows and even personal preference workflows, Brient said.

   "There's fundamental, different ways about the way doctors work with an EHR. One of the important things is, can you type? Some doctors type super fast, and they love to type. Some doctors type like 10 words a minute, and they hate to type. So that's a really important thing for the EHR to know," Brient said.

   Other factors include variability in what devices physicians prefer, whether they like voice-to-text functionality and whether they want the EHR accessible in the exam room or not.

   "Over time — and this is not a thing today — but over time, when we go to implement a physician, our goal is that we will ask them five or six questions like that. And the system will automatically configure itself for physicians who work like that. And we won't have to train them, Brient said.

   Athenahealth is working to roll out this customizable EHR for some specialties as soon as the next three or four years, Brient said.

   In the few years leading up to the pandemic, adoption of tools like voice-to-text and artificial intelligence-enabled clinical documentation was growing in healthcare to try and streamline notetaking. Athenahealth is also interested in that space, Brient said, calling the use of AI key to cracking the code of making the EHR experience delightful for physicians.

   Athenahealth now has a division devoted to applying AI to the clinical workflow, when just three years ago, the company was only using AI on the back office, claims processing, and financial side of the business.

   On the clinical side, Athenahealth is exploring using AI to pare down work like data reconciliation, inbox management and workflow experience.

   "If we know how you work, and know what you do ... and we know what you look at, well good. We'll put that on one screen and the things you want to do, we'll put right there," Brient said.

After helping establish the FDA's Digital Health Center of Excellence in 2020, Bakul Patel was recently promoted to a new role.

As chief digital health officer of global strategy and innovation, Patel's tasked with helping the Center for Devices and Radiological Health meet its 2022-2025 strategic priorities that it shared last week, through the lens of advancing digital health. 

CDRH Director Jeff Shuren in announcing Patel's promotion noted that under his leadership the center's digital health capabilities and services "grew significantly" and that in the new role he will continue to grow and advance the work of the digital health center, working with stakeholders both nationally and internationally. 

At the same time, digital health has a prominent role on the FDA's list of planned guidances for 2022, including how to handle changes in algorithms as well as long-awaited guidance on clinical decision support tools. The FDA last year also issued its first action plan for AI- and machine learning-based devices, including how to make these systems transparent to users, and how to support good machine learning practices.

MedTech Dive caught up with Patel about his new role and what to watch for on the horizon as these discussions culminate into guidance.

This interview has been edited for clarity and brevity.

MEDTECH DIVE: You recently were promoted to chief digital health officer of global strategy and innovation. How do you view your new role and what are your priorities stepping into it? 

BAKUL PATEL: This new role was primarily created for enhancing the implementation of one of the strategic priorities in our 2022 - 2025 strategic priorities that we published last week. This role allows me to focus on that priority and on advancing health equity. As you know, digital health technologies are suited nicely to meet people where they are, and I think that's what excites me about this new role.

I also feel like my passion toward advancing health equity aligns really well with the strategic priorities. I'm excited about taking what I've learned about digital health and how that promise of digital health can reach people where they are, and can actually be something that can be implemented. I'm looking forward to seeing where we can lean forward and now establish expectations from research and really advance this field.

On that note, I noticed that health equity was something you had commented on in your action plan last year. How are you thinking about how to approach that, and where the FDA can have a role?

PATEL: It's exciting to see that, in the last few years, digital health and technologies that are under the umbrella of digital health — it's either an app, it's a wearable, etc. —  are now reaching people that we normally did not reach. Now, we're also seeing those technologies being used in care, and we want to accelerate that further. So, that's one aspect. 

The second aspect of it is last year we published guidance on technology used in clinical trials. That's another great example of how digital health technologies can help reduce not just the burden on clinical trials, but also reach people that normally would not be part of trials. 

Mental health care can now reach people where they are through the use of connectivity, software and other means of technology. I think that's sort of exciting.

It's too early to say exactly how we're going to frame this, but I think those are some initial threads that we can pull on. 

We've been following the pre-certification pilot program, and Pear Therapeutics, Apple, and a couple of other companies have gone through that. Do you have any updates on where that is right now or what you've learned from that process so far?

PATEL: We set out to explore a new paradigm for technologies that are evolving at a much quicker rate than other technologies are. Theoretically, digital health technologies and software involve going further at different rates, and when you layer on top of it AI and machine learning, they also evolve very quickly. So, we set out with this program and we put out a few concepts.

We have found two fundamental things. One is this concept about having a really nimble, tailorable paradigm; absolutely that's number one. Number two, we also learned things that we would like to do or feel that balances the need for FDA independent oversight, that a standard of safety and effectiveness can still be maintained with the paradigm that we explored. However, those concepts that we explored cannot be achieved in our current regulatory paradigm. We don't have the authority to do that. And I think that's sort of the next step of the program.

You probably have heard folks from the FDA talk about how we are writing a report on the learnings from the program.

So, essentially, the ability to adjust for how much software or an app might change over time?

PATEL: One of the key tenants of the program was we wanted to move away from this concept of an episodic way of looking at products and putting them in the market and waiting for signals to come back, and come up with a paradigm that's much more fluid much more continuous across the entire lifecycle of that product. So, that becomes more real when you have solutions and products that are evolving on a much more frequent basis.

Let's talk a little bit more about some of the plans that you all shared last year related to AI and software as a medical device. I noticed that you plan to issue a draft guidance about how to handle changes in the future, the change in control plan. How is that coming together? 

PATEL: The guidance is one of the guidances on our priority list to publish. The team is working really hard on publishing that draft guidance. The idea emerged from this concept of — for a changing product — how can we as a regulator assure it's changing within the confines or the boundaries of what we feel addresses the risks to patients?

From a safety perspective, and then from an effectiveness perspective, you can imagine changes could happen that could go from indicated for condition one and you know, over time, the changes could be for condition two. We wanted to draw those two parallels, two dimensions, and say if manufacturers or makers of this product can predefine some of those changes to anticipate as much as they could anticipate — while we are authorizing the product to go to market — we could allow them to sort of make those changes with no special controls in place, right? 

As the name says, pitch to me a change in control plan where the manufacturer identifies the change and engages with FDA in that review of the product to say, "this is how we're going to control those changes to maintain the safety and effectiveness boundaries that you approved it for." And those are the concepts that evolved when we put out this discussion paper in 2019 and got a bunch of feedback. People loved the idea and we also tried it with a couple of de novos during our process of reviewing them, and we found that it could be applicable for other areas and this is why we are publishing the draft guidance. 

There was some discussion about best practices for developing algorithms and making them transparent. That was something that was discussed a little bit as well last year. Where do you see that going in the future? 

PATEL: We had a workshop last year on transparency. The team is still distilling some of the feedback we have gotten. 

I would say that there's going to be more than one thing that's going to come out of it. Either it's a research project, which we may be engaged in, or perhaps even an activity around just understanding how transparency for different people would mean. 

I think there are lots of concepts being explored, like the model card methods and the nutrition labels concept brought up at the workshop, which were really interesting. And before we actually go out and publish a guidance, we probably want to do some homework and understand which ones will yield the best outcomes. So, I think there's a little bit more work to be done on that aspect. 

Digital health has changed a lot over the last couple of years. It seems like the area has become a lot more robust in terms of the number of companies working in this space and also the stage that they're at. From your perspective, what have you seen change and how are you thinking about that?

PATEL: I'll just give you my personal opinion. I've seen the industry really embracing the science that goes behind evidence generation to really prove some of these technologies and the solutions they're bringing to actually help patients. And I think evidence-based digital health technologies is really sort of where I've seen it mature over the last few years. I think there are lots of people working in general wellness, and I've seen that same level of energy now being applied to solving real problems. So, I'm really excited about where this will go next.

Dive Brief:

• IBM is selling the data and analytics assets of its Watson Health business to a private equity firm as it looks to refocus on its core cloud business.
• The sale, which is expected to close in the second quarter this year has been anticipated for quite some time, and comes following the limited success of Watson Health, despite a spate of high-profile acquisitions of health information companies to bolster the enterprise.
• Financial terms of the deal were not disclosed. IBM's stock drooped slightly in Friday's trade following the news.

Dive Insight:

Despite the computing giant's big aspirations for the potential of AI applications in medical research and clinical decisionmaking — and spending millions to acquire health IT players including Truven, Merge Healthcare and Explorys — Watson Health fell short, leading IBM to begin exploring a sale of the beleagured healthcare analytics segment.

Tom Rosamilia, senior vice president of IBM Software, said in a statement the deal is a next step allowing IBM to focus more intensely on its platform-based hybrid cloud and AI strategy, though New York-based IBM remains committed to supporting its clients in health IT.

IBM never gained as much traction in the health IT market as some of its chief competitors, like Google and Microsoft, according to Paddy Padmanabhan, CEO of Damo Consulting.

"In the current competitive landscape, IBM would not be considered a significant player in healthcare. Selling off the data assets essentially means an end to the Watson Health experiment," Padmanabhan said, though it could allow IBM to develop a new approach to healthcare.

The spoils of the divisional carveout are going to PE firm Francisco Partners, which has invested in more than 400 technology companies since launching more than 20 years ago, including healthcare players like GoodRx and Zocdoc.

The agreement appears to hinge largely on Watson Health's products and datasets, which Francisco will likely look for ways to monetize.

Watson Health was originally launched in 2015 with the goal of applying IBM's algorithms to derive advanced insights with real-world applications. However, the division struggled to bring that vision to fruition.

In one early misstep, IBM's marketing for Watson for Oncology, a project using AI to generate cancer treatment advice, greatly oversold the product's capabilities, which weren't backed up with clinical evidence. The move was met with complaints from doctors and clients after finding Watson for Oncology in some cases was issuing unsafe and incorrect treatment recommendations, and was followed by flagging sales, layoffs and management turnover.

The Watson Health sale hints at the challenges of disrupting U.S. healthcare, a problem faced by other tech giants like Amazon and Google as they elbow into the industry. Amazon, for example, has reported challenges scaling its Amazon Care business, while Google faced a firestorm of controversy in 2019 over using medical data from millions of Americans to develop new product lines without their consent.

However, the AI in healthcare landscape is still very attractive to big tech, as companies look to harness the industry's massive datasets to move the needle on care quality and streamline administration, while nabbing a share of the profits. AI was cited as one of the top technology investments for healthcare executives in a survey conducted last year.

And some cloud companies are still looking to scoop up data, even as IBM steps aside: Oracle acquired EHR giant Cerner in December for more than $28 billion.

After a year of historic startup funding across every investment phase, from seed to late stage, some market watchers expect VC investment as a whole to moderate in 2022. Digital health could be no exception amid rising interest rates and as startups flush with cash eye next moves, including M&A or exits.

However, some expect funding levels to only continue rising, cheered by factors such as a healthy mix between first-time and existing investors marrying an infusion of new interest with continuing bets from digital health veterans, and moves from outside players — including deep-pocketed tech firms like Amazon and Microsoft — to invest and acquire in digital health.

In addition, the adoption of digital health products and services seems to be stabilizing, with sustained patient demand and no widespread rollbacks in virtual services from providers. That's ameliorating concerns that demand for such solutions would plummet once the brunt of COVID-19 subsides.

Digital health startups capitalizing on the need to manage costly, pervasive conditions, free up physicians and enable care at scale, along with the shift to individualized, high-touch medicine, are expected to bring in the lion's share of funding this year, experts say.

Barely a week into 2022, U.S. venture capital firms raised almost $13 billion, according to PitchBook data — about equal to what the firms raised in an entire year about a decade ago.

That money has got to go somewhere, and it's a good bet the healthcare sector will be a major recipient. The industry raised the second-highest amount of global venture funding in 2021, trailing financial services, according to Crunchbase data.

Health tech grabbed a serious share of the attention.

U.S.-based digital health startups brought in almost $30 billion in 2021, almost doubling the total investment the year prior. Investors and market watchers are wary on making firm forecasts that 2022 will break 2021's record, but most are eyeing the year through rose-colored glasses.

"I expect this positive momentum in deal flow to continue in 2022," said Ravi Kumar, a partner at professional services firm the Connor Group.

Market watchers see digital therapeutics, personalized medicine, provider-focused infrastructure and mental health and chronic condition management as particularly ripe for continued investment as the pandemic moves into its third year.

"It's hard to see what could stop it," said Marc Albanese, senior director of research for healthcare and emerging tech at market intelligence firm CB Insights.

Digital therapeutics and personalized medicine

Digital therapeutics saw significant funding growth in 2021, with startups delivering medical interventions using evidence-based software to manage and treat a broad spectrum of diseases reporting surging investments. It's a continuing trend: Investments in U.S. digital therapeutics companies have increased an average of 40% a year over the past seven years, according to McKinsey.

The goal of digital therapeutics is to treat chronic conditions at scale, alleviating some of the biggest stressors on the medical system right now, such as physician shortages, Albanese said. Digital therapeutics allows care delivery to disconnect, in a way, from the supply side, enabling patients to receive care asynchronously and remotely, without doctors having to constantly monitor the apps.

A number of factors are converging to spur investments in digital therapeutics. Coverage pathways for prescription digital therapeutics widened last year, allowing regulatory approvals to escalate. Startups are growing their body of clinical evidence to back up their technologies. There's also increasing buy-in from healthcare stakeholders, including payers and providers, around integrating digital therapeutics and building payment models around them.

Digital therapeutics were a huge highlight in 2021, but "when we think about the space, I think we're still in the early innings here," said Amanda DiTrolio, a healthcare intelligence analyst at CB Insights. A number of startups are still developing their therapies, and will likely search out additional infusions of cash to do so.

Two salient examples are virtual reality therapeutics and cognitive behavioral therapy apps, according to DiTrolio.

AppliedVR, a VR therapeutic for pain management, received breakthrough designation from the Food and Drug Administration early last year for its flagship product for treating fibromyalgia and chronic intractable lower back pain. It was the first FDA designation for a VR prescription therapeutic, and for many market watchers signals growing regulatory acceptance of the space.

"There's definitely excitement around that," DiTrolio said.

The analyst also noted that as digital therapeutics launch more clinical trials and expand into more indications beyond chronic care, like gastroenterology, IBS, women's health and neurodegenerative diseases, "that's going to require large sums of capital in the coming year."

Growing interest in providing more targeted care also is likely to help startups in the nascent personalized medicine space, according to Pitchbook emerging tech analyst Kaia Colban.

Colban predicts that personalized medicine startups, which use AI- and ML-based data platforms to glean insights into patient care and create personalized treatment plans, will receive a record level of VC investment in 2022.

It's a huge total addressable market with strong market drivers, including a confluence of technologies like remote patient monitoring devices, electronic health records, genetic databases and cloud computing, that are helping advance the state of such applications, according to Colban.

Mental health

When it comes to clinical indications, mental health is likely to continue raking in the bulk of funding.

Digital health startups offering mental healthcare raised $5.1 billion in 2021, almost double what they brought in the year prior, according to Rock Health. Investors will continue looking to fund tools enabling the delivery of mental healthcare at scale, CB Insights' Albanese said.

COVID-19 stressors on mental health are unfortunately unlikely to abate anytime soon as the virus becomes endemic. The pandemic's exacerbation of conditions like anxiety and depression has interlaced with a shortage of U.S. health professionals to create historic levels of demand for therapy and other treatments. Payments have also never been more generous, as employers and payers increasingly see the downstream value of proactively addressing their members' mental health conditions.

That's all spliced to result in a pandemic wave of new virtual care options for intensive mental and behavioral health needs. Well-funded startups in the space, such as Lyra Health, which raised $387 million last year alone, often pair traditional therapy options with wearables, targeted therapeutics, chatbots and other features to try to meet consumer expectations for preventive care, mental health coaching, therapy and medication.

One early-stage startup in the space to watch is Woebot, according to Pitchbook's Colban, a company that develops AI-based chatbot technology delivering mental healthcare through a digital therapeutic. Woebot has raised almost $130 million in venture funding to date.

Additionally, DiTrolio expects more funding to flow to startups targeting specific underserved demographics, including Folx Health, which offers virtual care for the LGBTQ community; Equip Health, which treats people with eating disorders; and Eleanor Health, a value-based startup targeting opioid abuse.

"The theme of targeting underserved populations really took hold in 2021, and there's plenty of room for more investment there," DiTrolio said.

Chronic condition management

Investors also are likely to continue boosting funding for chronic condition management startups that look to prevent downstream health complications, while recouping a share of any savings.

Rock Health has called out diabetes and musculoskeletal care as key funding arenas, as they're high-spend conditions that increasingly are managed virtually. Growth in startups managing such conditions could help lift some weight off the overtaxed physical health delivery infrastructure, as January COVID-19 hospitalizations break all-time records.

MSK funding, for example, grew sixfold between 2020 and 2021, when it reached $1.4 billion. Virtual MSK clinics like Hinge Health and Sword Health each completed multiple raises throughout the year, and that pace is unlikely to let up, analysts said.

Hinge alone has seen its valuation double in the past 12 months, making it the top-valued VC-backed retail healthtech startup — and a likely IPO candidate for 2022, according to Colban.

The focus on specialty care in general is also going to increase because of growing risk assumption, predicted Jacob Effron, a principal at VC firm Redpoint Ventures. As more healthcare companies take on risk, they'll need partners to help manage patients with more niche and high-cost conditions, according to the investor.

"I feel like I meet a different entrepreneur every week that's building in a different specialty area," Effron said.

Behind the scenes: Infrastructure plays

Investment in back-end infrastructure and interoperability startups has been consistently strong in past years, but appears to be accelerating. Such provider operations companies raked in more than a third of total seed and series A funding in the U.S. and Europe last year, up from a fifth in 2021, according to Silicon Valley Bank data.

VCs are likely to continue investing heavily in companies that help providers modernize processes like billing, coordinated care and research, following rising pressure to make systems more user friendly. That will benefit startups that can weave their tools into existing IT systems to streamline the experience of administrators, doctors and patients, reducing friction and cost, experts say.

"We're continuing to see strong, strong funding there. I would not expect that to go down," Albanese said.

Investments will increasingly be spurred by federal regulations mandating free electronic information sharing between disparate software systems and increased regulatory pressure for price transparency. The resulting emphasis on data interoperability and portability will cause more growth in digital health startups in that arena, Connor Group's Kumar said.

Other macro trends galvanizing investment include the emergence of more direct-to-consumer healthcare delivery options, putting health systems at risk of losing their most lucrative patients — the commercially insured. That trend incentivizes them to digitize their platforms and provide a better consumer experience across the board, whether it's scheduling, patient intake, communication or billing.

Growing interest in risk assumption is also causing providers to look increasingly for digital health solutions to help them take on and manage risk, which will cause funding to startups in that area to increase as well, according to Redpoint Venture's Effron.

"I think the shift to risk is probably the biggest macro trend overall. And I think the opportunity to build supporting tech and services around that to help physicians succeed in that new environment is probably the top area I'd expect provider tech investment in 2022," Effron said.

In addition, COVID-19 has exacerbated labor shortages and provider burnout, sending hospitals and health systems scrambling to improve workflow stressors on their employees, experts say.

It's a big opportunity for startups that received funding last year, including companies that integrate data streams into unified API ecosystems like Innovaccer, Redox and Ribbon Health; enable cross-company data analysis like TripleBlind and Truveta; streamline data structure and labeling like ScienceIO and Centaur Labs; provide building blocks to help others implement digital health tools like Commure, TruePill and Wheel; and offer healthcare-focused robotic process automation technologies like Olive.

Fintech

Billing also is an area to watch, as unsustainably increasing prices for healthcare services have dented the pocketbooks of many Americans even before the coronavirus. Hospitals already were running more financial assistance transactions before the pandemic, but COVID-19 accelerated the trend, boosting the financial pressure on providers and patients struggling with higher costs, according to credit bureau TransUnion.

Financial services was the leading sector for venture investment last year. Such startups operating in the healthcare space are likely to continue enjoying investor interest as consumers look for ways to pay for needed healthcare services without breaking the bank.

Startups offering technology to smooth filing and billing processes, as well as those attempting to ease the burden of payment on cash-strapped consumers, have reported breakneck growth in adoption and funding.

Buy now, pay later platforms, for example, allow consumers to make purchases on payment plans. San Francisco-based PayZen is one such startup operating in the healthcare space. PayZen, which raised $15 million in a Series A round in 2021 to expand its model, uses artificial intelligence to assess patient's ability to pay fees after insurance, then automatically creates a tailored payment plan for the patient and an automated system for the hospital to manage billing.

That leads to hospitals recouping a higher proportion of costs, without having to send patients to collections, PayZen says.

But whether you're a fledgling company combating pain points in medical billing or any other sort of digital health player, "there's never been a better time to start a digital health company," Effron said. "I think there will be a lot of interesting opportunities for investors in 2022."

Proponents of artificial intelligence and machine learning have touted for years how algorithms hold significant promise to streamline healthcare administration and delivery, freeing up nurses and physicians to practice at the top of their licenses.

But COVID-19 has shifted the priorities for that technology, speakers said at HIMSS21 in Las Vegas this week.

Before the pandemic, Rush Medical College's major goals were training staff and students, while promoting health equity. Of its artificial intelligence pilots, "many didn't go anywhere," said Bala Hota, Rush's chief analytics officer, on Tuesday.

But as COVID-19 hit, "for the first time, our organization craved the numbers from the top to the bottom … the liquidity of data that we wanted became the No. 1 thing," Hota said.

Speaking on the same panel, ​Tanuj Gupta, VP of Cerner Intelligence, noted that the EHR vendor's clients want it to do the same tasks —​ but want more AI getting the job done.

EHR companies help with five main jobs, Gupta said: capturing healthcare information, quickly summarizing that information, coordinating care, securing reimbursement and making critical decisions.

Whether five or 50 years from now, those jobs will still need doing, but "people are asking for different technology to do those jobs,” Gupta said, whether its ambient voice technology or natural language processing to autochart, voice recognition for info summaries, robotic process automation in care coordination or machine learning in clinical decisionmaking.

"The jobs are the same. The tools are evolving. So you're going to see a lot of digital health companies evolving hopefully to bring you those tools," Gupta said.

Cerner, for example, for eight years has only deployed about three to four clinical models to clients. In 2020, after reshuffling investments and ramping up needed infrastructure, Cerner deployed two models 35 times.

And this year, the vendor is on track to deploy 10 models 350 times, Gupta said.

Hota also said that AI and ML tools were becoming more mature, saying Rush is building predictive models to try to bolster its social determinants of health screening platform in a bid to connect in-need patients with resources in their community.

Meanwhile, payers are also taking note of AI's potential.

Anthem Chief Data and Analytics Officer Ashok Chennuru said Tuesday it's "embraced AI" to transmit structured and unstructured data into "proactive, predictive and personalized insights."

But "it's not as simple as developing a model, throwing it in there and then letting it go," Gupta said. Models need to be frequently calibrated, which takes time and money, and developers need to make sure they're not building on dirty datasets. Additionally, it's preferable to end users like clinicians if any output is explainable, which helps with adoption down the line.

And machine learning models are often inconsistent when applied in the real world, according to a Monday presentation by Mujeeb Basit, associate chief medical informatics officer at the University of Texas Southwestern, on an algorithm to predict the likelihood a patient becomes critically ill during a hospital stay.

"There is just not enough data, and models generally aren't designed to deal effectively with that missingness," Basit said.

And the future of AI is iffy, as many healthcare leaders remain uncertain on the ROI of algorithms. One statistic from a survey conducted by management consultancy the Chartis Group, released Wednesday found 7% of healthcare executives said they believe AI and ML are the top priority for the future, while the same percentage called them a "distraction."

"There's a general belief that AI/ML is going to create new operational capabilities, but it's not really clear how it fits in the agenda," Tom Kiesau, senior partner with the Chartis Group, said.

One new frontier for AI/ML, however is emergency forecasting in event of further public health crises, experts say.

"Health systems need to predict catastrophes that are going to overwhelm both labor and non-labor supply, in advance," Gupta said. "We should be able to do it with reasonable certainty. We do it with earthquakes, we do it with forest fires, we do it with weather."

Researchers at UnitedHealth's health services arm Optum started studying the COVID-19 pandemic while tracking a flu outbreak in the U.S. in 2019, and began using AI and ML to create infectious disease forecasting, Optum Director of Research Danita Kiser said Wednesday/

The computational epidemiology system — which Kiser said was as accurate as current weather forecasting — can make a global impact by text messaging caregivers early warnings of outbreaks so they can isolate patients.

Currently, Optum is sharing this information internally, and is developing plans to share it with third-party providers.

"If we can forecast a week or two in the future, we can put prevention measures in place to save people's lives," Kiser said.