Signaling the Tipping Point of AI Value in Healthcare
Top minds across a broad spectrum of healthcare disciplines converged to discuss artificial intelligence in Boston April 23-25 at the 2018 World Medical Innovation Forum (WMIF), which drew 1700 registrants and 169 speakers. Researchers, clinicians and industry leadership paraded on stage for a collection of panels and “1:1 fireside chats” to discuss healthcare applications of AI and related technologies, their progress, roadblocks and the promise of future value. The majority of provider participants were from Harvard teaching hospitals as Partners Healthcare runs the conference.
The healthcare innovation community is always looking for applications to illustrate the potential of AI, but also to grasp a tangible that is closer to the now rather than solely as a futuristic vision. At this point, there are many diverse applications for AI and related technologies, as highlighted in the framing of the conference, which opened with a discussion of 19 “AI breakthroughs” and closed with the “disruptive dozen” (scroll to the end of the post for listing). It was encouraging that these innovative pursuits were not limited to any small facet of care or research, but instead cover a broad spectrum of healthcare challenges.
Across the rest of the conference were overlapping discussions of the impediments and implications of advances in AI and the many applications. These AI-centric discussions can be organized into 5 themes that broadly mirror key concerns of the healthcare industry as a whole, indicative of both the hype and potential for these technologies to fundamentally change our methods for care delivery:
Unsurprisingly, the importance of good sources of high-quality data was a core theme across all panels. In some cases, the data already exists but has not yet been analyzed. Drug manufacturers sitting on mountains of clinical trial data being a prime example. New data sources like that those coming from companies like Flatiron – recently acquired by Roche – provide a fresh resource for research. All of this data, however, exacerbates challenges of large data analysis. In order to run AI and machine learning systems, the data needs to be clean. Surveys have indicated that data scientists spend 60% of their time cleaning data. The challenge before everyone includes sourcing data (as clean as possible), cleaning and making data usable for tools.
In many cases, the limiting reagent to progress is our understanding of biology – as much as or even more so than the technology. Technology, in fact, is the means to understand the biology further. Life sciences companies like Novartis, Pfizer and others are actively using, in their words, “higher order data,” including ‘omic data sources. In doing so, they hope to reach breakthroughs faster by redefining how they consider new drug discovery.
William Lane, MD, PhD delivered one of the 19 breakthroughs that opened the conference, bloodTyper, which offers a new outlook for the long-standing method of categorizing patient blood types based on the presence or absence of 2 antibodies and 2 antigens. bloodTyper uses DNA-based categorization as opposed to serologic testing alone. Whole genome sequencing costs have declined dramatically in recent years and are expected to continue to decline making this solution viable for widespread clinical adoption. Technology enables this effort, but much research needs to be done in order to illustrate specific value and bring these new methods into practice. With continued advancement of technological capability, understanding the biology of health and/or disease will continue to be a primary obstacle.
One of the consistent themes across panel discussions was the notion that medical technology, AI in particular, is futuristic, while our care system is stone-aged. The challenge of adoption and change to the care paradigm is not a limitation of technology. Panelists repeatedly remarked about how the pairing of “Jetsons technology and a Flintstones care system” would take substantial time to evolve, because change requires evidence and trust that are not ascertained lightly.
In some cases, however, evidence-based change is already upon us. A group at MGH, represented at WMIF by Erica Shenoy, MD, PhD, is using Machine Learning to more quickly identify cases of hospital-acquired infection, identifying C Diff cases full days before the currently accepted 5-day standard process. We expect to see substantial growth in research publication volume illustrating the value of AI and Machine Learning technologies in areas like this.
On the other side of the curtain, AI is changing the way that industry operates. Drug development, in particular, is undergoing a revolution of sorts. Companies like Exscientia, with diverse data science capabilities, provide life sciences partners with a way to look across different data types including ‘omic data, research text and other sources. This allows manufacturers to potentially repurpose molecules and sub-organize disease for more precise targeting.
Not all areas of healthcare are ready environments, but the ubiquity of efforts to utilize AI technologies to accelerate processes, improve accuracy, increase access, increase bandwidth and offer precise care points to a tipping point.
Another recurring theme was the impact of AI on the workforce. As in other healthcare sub-verticals, there is a large and increasing demand for data experts. Radiologists and pathologists across the conference echoed the surprisingly optimistic resolutions of a panel dedicated to this topic both with respect to the increased demand for data science professionals in all organizations as well as the potential obviation of some roles.
The future will likely position these data scientists alongside healthcare professionals as part of multidisciplinary/cross-functional teams in care and non-care settings alike. There is also an expectation that data science literacy, at least at a high level, will become a core component of education for many healthcare professionals, not just IT specialists.
The fear of the obviation of certain roles was interestingly framed. At a very high level, the introduction of AI and related technologies will exacerbate the division between the highly educated and the less educated, as Glenn Cohen of Harvard Law School eloquently pointed out. It was also highlighted, however, that a lot of the lower skilled manual roles have already been dramatically reduced with the introduction of EMR systems. The natural targets often discussed beyond lesser skilled workforce are experts in clinical disciplines of radiology and pathology. In both contexts, AI technologies have been shown to approximate or even beat the accuracy of some clinical practice by human experts. There is also a global shortage of these types of experts. The prevailing opinion expressed was that new technologies will be used to make these professionals more efficient and effective, and it is unlikely that technology will be used to truly replace the human element.
The burden of AI technologies on regulators is substantial. They need to be able to evaluate, audit and assure quality of these new technical capabilities. Linguamatics is one example of the many companies that are making progress in crafting technology solutions for healthcare while also targeting the FDA as a customer. Providing a mechanism for auditability of otherwise “black box” AI systems is a great benefit to the regulators.
Standardization is a key to enable scalability and support industry-wide progress. Consider DICOM, for example, which offers a standard for many imaging modalities. The expectation has been set that at least some standardization must be a focus of innovators so as not to run into an even greater challenge of interoperability with more complex technology. Building gold standard training data sets was a solution highlighted that begins to address this challenge with AI.
Precision medicine was an end target for many of the applications and topics of discussion at the conference. The development of a clearer, richer phenotype and genotype (or other “’omic” and new forms of data) essentially yields the potential for a digital twin for each individual patient. To accompany this, the development of precision drugs, diagnostic capabilities and other therapies is the end promise of many of the applications of AI, machine learning and other technologies.
We will only be able to deliver on these promises if our data, technology and systems for the delivery of care are able to adapt. To get there though, we need to establish trust and confidence across the healthcare ecosystem and for patients with AI and its elevated role.
Currently, there are little islands of the right components to make great progress and work is well underway. In oncology, for example, there is data access, a willingness and need from physicians and patients, a motivated industry (opportunity for profit or strategic position), research capacity and funding, payment and regulatory feasibility. At WMIF, Atul Gawande described these environments as “ready environments,” in that they are capable and motivated. The historical example he gave to juxtapose “ready” and “not ready” was the spread of anesthetized surgery over weeks or months, vs. antisepsis, which he estimated to take 20 years to spread.
Not all areas of healthcare are ready environments, but the ubiquity of efforts to utilize AI technologies to increase accelerate processes, improve accuracy, increase access, increase bandwidth and offer precise care points to a tipping point. Within the next year, I expect to see a thicker volume of applications and more importantly more examples of impact.
Matt Guldin · 2 years ago
Chilmark Team · 1 month ago
Chilmark Team · 3 months ago
Matt Guldin · 2 months ago
Connected Health Conference: Digital Health’s Opportunity to Capture a Positive View of Aging
The Connected Health Conference and its predecessor events, the Partners HealthCare’s Connected Health Symposium and the Personal Connected Health Alliance’s Connected Health Conference (a HIMSS event), aim to frame digital health advancements in the context of improving clinical care delivery as well as personal engagement.
It’s “personal” engagement because the conference recognizes that “patient” engagement only occurs in a healthcare setting. Improving health and well-being must be part of our everyday lives – or, as this year’s conference theme put it, part of “the connected life journey.”
In particular, this year’s Connected Health Conference examined digital health’s impact on getting older – a process that speakers such as AARP CMO Dr. Charlotte Yeh attempted to destigmatize by emphasizing the positive aspects of aging, including learning and building social connections. Get it right, Yeh said, and aging makes us happier – and happiness has been linked to better health and longevity.
With that in mind, here’s a recap of various digital health solutions discussed at the conference, along with Chilmark Research’s assessment of the maturity of these solutions.
Condition management: Almost mature. Multiple speakers highlighted the potential of chronic condition management solutions to provide targeted engagements and interventions, especially those that draw upon principles of behavioral science as opposed to marketing. Plus, Livongo Health announced a partnership with Alaska’s Medical Park Family Care that is notable because providers approached the vendor, with both stakeholders then approaching the payer together; this could be a sign of a shift, as payers or large employers typically initiate the conversation (and may not even get providers involved at all). If there’s a caveat, it’s solution sprawl; Dr. Adrienne Boissy of the Cleveland Clinic noted that the hospital has 27 different apps for engaging with patients; that number has to go down.
Providers don’t want artificial intelligence but, rather, the output of AI – clinical decision support, diagnostics, more personally tailored care plans, 30-day readmission risk scores, and so on – but that hasn’t stopped vendors from promoting the use of AI and machine learning.
Telehealth: Getting there. As Carla Kriwet of Philips noted, the holdup for telehealth adoption is not the maturity of the technology; it’s the clinical culture that still emphasizes in-person visits and the related struggle to define telehealth’s value-add. (Kriwet didn’t mention the regulatory and reimbursement challenges, which Twitter users in the audience were quick to point out). Plus, telehealth requires broadband access, which millions of Americans still lack. That said, physicians are a bit more bullish on telehealth and other technologies at the point of care, AMA President Dr. David Barbe said, pointing to AMA survey data he shared on stage at the Connected Health Conference.
Decision support: Getting there, slowly. Providers don’t want artificial intelligence but, rather, the output of AI – clinical decision support, diagnostics, more personally tailored care plans, 30-day readmission risk scores, and so on – but that hasn’t stopped vendors from promoting the use of AI and machine learning. The baseline ongoing challenge, of course, is bringing disparate data sets together to let AI algorithms do their thing, and then deploying them at the point of care in a timely manner, which is no small task when those data sets include months-old claims and years-old HHS reports. Oh, and don’t forget data streams from remote patient monitoring devices. Smart vendors and providers are using an API approach to address this data deluge, but progress has been slow.
Chat bots: Getting there, slowly. Conversations with bots allow engagement solutions to scale, especially for low-acuity care as well as general wellness and mood. But for every Conversa Health (where chat bot responses come from a database that also identifies whether a particular response must automatically trigger an escalation to higher-acuity care), there’s an Ada Health (where users must escalate on their own, for a $25 fee). Here, the market has yet to separate the wheat from the chaff, as is evidenced by Ada Health’s latest $47 million funding round.
Blockchain: Emerging. There is certainly potential for blockchain to provide secure data sharing, and use cases are beginning to pop up. Hashed Health is addressing provider licensure in the state of Illinois, which will help the state store up-to-date credentialing information and share it with payers; the hope is to eliminate the bottlenecks in an often-manual process. Meanwhile, MintHealth, promises a personal health record platform powered by blockchain. The presence of a unique global identifier, which blockchain enables, could let patients manage their records better than previous PHRs; plus, MintHealth plans to sell to commercial payers, who have clear financial incentives to better engage members. Even if these efforts fall flat – and, let’s face it, PHRs have a poor track record by placing too much burden on the patient – they have at least demonstrated practical examples of blockchain in healthcare.
Virtual reality: Incubating. Qualcomm and ForwardXP announced a VR simulation at the Connected Health Conference that lets users experience stroke symptoms, in an effort to help them better identify warning signs. While this could be an effective educational tool for medical professionals, we’re not sold on its viability for consumers, who would either need to use it on their own VR devices or travel to an educational session somewhere with a VR device. In one session, Dr. Brennan Spiegel of Cedars-Sinai Health System cautioned against “overpromising and under-delivering” with VR. Pain management is an effective clinical use case; giving patients the 21st century equivalent of a ViewFinder probably isn’t.
Wearables: Incubating (still). Wearables provide data such as steps taken, exercises completed, hours of sleep, but they don’t yet provide insight – and that hasn’t changed since I wrote about it for CIO.com way back in 2014. One reason is the separation between wearable information and clinical workflows, which hampers the ability of wearable data to reliably contribute to decision support. Another is the accuracy of the data; we still earn “steps” for folding laundry or doing dishes, but not for riding a bicycle, so physicians can’t be blamed for viewing wearable data with a dose of skepticism. Finally, for all their talk, consumer device makers such as Apple, Fitbit, Google, and Samsung have yet to crack the chronic condition management market – only Nokia has, and its market share lags substantially.
The smart home: Not even close. Speakers mentioned a number of futuristic use cases, ranging from home-based robot caregivers to fall alert sensors tied to home security systems to sensors that monitor water usage, detect abnormalities, and alert caregivers of potential problems. No one mentioned which healthcare stakeholders would subsidize the high cost of this technology, train end users (including caregivers), or monitor data and escalate a case as needed. Nor did anyone mention the ROI for such technology deployments, or the impact on clinical outcomes.
Ultimately, the goal of these solutions is to meet the needs of people managing their health, whether on their own or through the clinical experience. As such, these solutions need to ease their way into clinical workflows and everyday life and prove their potential to play a part in the connected life/care journey. Until then, they will simply remain stuck in the technology hype cycle.
Analytics – An Integral Part of Care Management Success
Last week, I attended the 3rd Health Analytics Summit (HAS). This was my first time attending an event that now attracts over 1,000 attendees. Providers were well represented at the event with nearly 80 percent of the attendees coming from various provider HCOs.
While Health Catalyst focuses on analytics, my observations were largely focused on the care management related aspects of the conference. Here are some of the main impressions from the event:
Line between analytics and care management has blurred further: In our Care Management Market Trends Report, there were some analytics-focused ratings criteria with ‘Risk Identification and Stratification’ being the most straightforward analytically-oriented criterion.
More HCOs though are moving beyond simply importing a risk score from a claims-based risk grouper solution as explained in our Insight Report on the topic. They use this as a starting point, utilizing other data types (mainly clinical and utilization data) to define their own proprietary risk groups including sub-groups within high-risk patients.
Additionally, HCOs are leveraging analytics to measure the effectiveness of their care management programs ‘early and often’ – instead of waiting 9-12 months to examine clinical and cost-based outcome measures. They are also looking at more implementation and process-based measures to assess program effectiveness of particular care plan elements before expanding them to other patient types.
New data are ‘sexy’ but existing data issues consume lots of bandwidth: One of the first poll questions asked was which new data sets were the attendees most interested in for analytics-related projects with the top 3 being: social determinants, patient-reported outcomes, and external demographic data (e.g., credit scores). There was not much difference between the three although most of the conversations I had or heard really focused on the gathering additional social determinants of health especially related to the patient’s home after discharge and what resources were available to a patient.
But clinical data quality and to a lesser degree claims data still remain the biggest issue. HCOs are spending 30-40 percent of time on this single issue during the first several months of a project. Combine this with a multitude of varying and ever-growing value-based performance (VBP) measures, it is no wonder that incorporating additional new data sets into various analytic initiatives is challenging.
Provider-led care management adoption reaching a tipping point: One of the more difficult things to determine is just how many HCOs have actually put a care management program in place in their outpatient settings. There are several different ways we have heard HCOs define this but the most standard definition is having care teams headed by an outpatient nurse care manager who actively manages some percentage of an HCO’s patients through a care plan.
At the Partners HealthCare session on their care management strategy, which Health Catalyst has licensed the IP of, nearly two-thirds of the respondents indicated they had a high-risk primary care management program already in place at their HCO. The conference attendees likely represent some outliers but it would not surprise me if the actual adoption rate for care management among hospital-based HCOs is already at or exceeds 50 percent by early 2017. This number is likely considerably lower in rural settings, community hospitals, or among physician-based HCOs.
There is not some clearly-defined threshold in which an HCO decides to put in place a care management program but the two most useful anecdotal metrics seems to be: number of primary care lives under value-based reimbursement programs (VBR) (est. >15 percent of primary care lives) and percentage of total revenue in VBR arrangements (est. >20-25 of total HCO revenue).
HCOs actually engaging in longitudinal care management for a cohort of patients: While more HCOs have put in place care management programs, the overwhelming majority of these programs are not truly long-term, continuous care management programs. Instead, the vast majority of them are based around enrolling a patient for a finite duration of <30 days, <90 days or <120 days.
This should not be surprising given that HCOs are rationally responding to the measurement periods of various VBP programs most notably the Hospital Readmissions Reduction Program (HRRP) program from CMS. Partners Healthcare detailed how they are engaging in care management for a cohort of high-risk patients from several different payer types including Medicare Advantage, Commercial, Managed Medicaid, and their own employees and dependents.
If provider-led care management programs are going to bend the cost curve, continuous care management including palliative care for end-of-life patients is going to be necessary. Simply focusing on inpatient admission rates is going to be insufficient given early ACO results. The question is how many HCOs have the financial and clinical resources, geographic coverage, and economies of scale to accomplish this lofty goal especially in regions where there is considerable patient churn (e.g., +20% annually).
Broader integration is key for care management: Time and time again access and timely integration with behavioral health services is becoming a critical issue for the success of care management programs. Partners Healthcare reported that nearly 40 percent of their patients in care management program had at least one behavioral health issue. This is not uncommon and several other HCOs have reported to us that 30-50 percent of their patients have at least one behavioral health issue beyond something that can be treated by a primary care provider (PCP).
It is critical that when these patients come in for an office visit to their PCP or seek access to behavioral health services that these resources are available in-person or via a telehealth consult. Other additional areas that are coming up as being critical to provider care management programs for high-risk patients, are tighter integration with substance abuse, pharmacy, and palliative care resources.
Most attendees at the Population Health Summit indicated their HCOs were in the early to middle stages of integrating analytics across their organization with varying degrees of success. Attendees overwhelmingly felt these efforts were having overall positive effects on quality even if ROI remains challenging to determine. Most surprisingly, attendees self-reported that adaptive leadership and culture was the highest-scoring attribute in Health Catalyst’s recently-released Organizational Improvement Readiness Assessment with analytics and best practices being the lowest scoring attributes.
What stuck with me though was just how pervasive analytics are to not only defining the foundational requirements for a care management program but the crucial role they play in helping to set up, refine, and support a care management program over time. Unless a care management program is using analytics to actively measure ‘early and often’ and using this feedback to effectively optimize care management processes, results will be limited.