HealthTech
MGMT 275
HEALTHCARE TECHNOLOGY
A mix of newsworthy content and thought-provoking opinion pieces Covering healthtech from all fronts, including the latest trends, key issues facing hospitals and practitioners, industry events and emerging technologies.
Healthcare Technology Management
Latest Blog Posts
Closing the Gap in Women’s Health Care: FemTech on the Rise
Closing the Gap in Women’s Health Care: FemTech on the Rise
The storied history of women’s healthcare dates back centuries, beginning in the late 19th century when the first female doctor, Dr. Elizabeth Blackwell, was labeled an abortionist for pursuing medicine. Despite such labels, she set out to increase the number of female physicians, launching the first women’s medical school New York Infirmary for Indigent Women and Children. Fast forward to 1960s when the women’s rights movement fueled the need for reproductive choices, including pharmaceutical contraceptives. This highly abbreviated history illustrates one key point: progress has been slow, painfully slow. But the landscape is changing.
Women’s healthcare has gained more traction in the last few years, attracting new startups, VCs and pharma under the umbrella of Femtech. But where, exactly, is Femtech making an impact on female-centered care and where are there opportunities for growth?
There have been several notable unicorns in the Femtech space, exiting to IPO in a market with double the number of “MenTech” startups valued at $1B+. In 2018 men’s healthcare startups invested in healthcare. In 2018 men’s healthcare startups raised 97.8% of the total VC money invested in healthcare. The disparity is unequivocally weighted in favor of men’s healthcare, despite women spending $20 trillion globally in annual consumer dollars (and making 80% of all household medical decisions). Women are on track to represent a market twice as big as China and India combined.
This imbalance could be due to a number of factors, such as the startlingly low number of women investment partners at US venture capital firms (11%), or 83% of the VC backed healthcare startups having no women on their leadership team. Regardless, it is clear that women’s healthcare needs more women at the helm to drive progress towards what is forecasted to be a $60 billion dollar "niche" by 2027.
One specific aspect of women’s healthcare that is worth watching is pre and postnatal care. Pregnancy care has long been an area of limited innovation and growth, and postnatal care even more so. Startlingly, pregnancy-related mortality in the US has continued to steadily rise since 1987, with the US having the highest maternal mortality rate among 11 developed countries. Nested in these concerning statistics are marked racial disparities: black mothers in the US suffer 3x the number of deaths during pregnancy, labor and delivery, and postnatally to that of white women. Compounding these outcomes around breastfeeding statistics: 60% of women discontinue breastfeeding sooner than they prefer and black infants are 15% less likely to have ever been breastfed compared with white infants. While there have been improvements in data collection aimed at equalizing care for vulnerable mothers, it remains siloed. To improve outcomes and equalize care, three key areas need to be further addressed, and there are several early entrants in the field worth watching:
Preventing postpartum mortality risk
The majority of postpartum deaths are secondary to complications from preeclampsia, hemorrhage, cardiomyopathy and infection. However two-thirds of these deaths are preventable. Detecting risk factors associated with these morbidities through stratifying data based on EMR and employing AI to detect risk factors would alleviate the burden on understaffed healthcare providers and alert physicians before fatal symptoms present. Additionally, moving the care continuum from physicians to midwives and doulas, common practice in other countries with far better maternal outcomes, could aid in the provision of high-quality care. It would also reduce the potential OBGYN shortage due to burnout. One company that recently launched in this space is Mae, a maternal health digital health platform underwritten by payors with the aim of connecting future Black mothers with doulas and healthcare providers.
Addressing postpartum peri-care
Postpartum care plays a critical role in the successful recovery of women’s abdominal and pelvic floor health. Yet, medical training on the surgical repair of perinatal lacerations and caesarean suturing remains grossly outdated. This is an area of ample opportunity for companies like OssoVR. By demystifying perinatal tears, medical students, physicians and midwives can develop critical skills in the surgical management of postpartum lacerations.
One company to follow is Granville Biomedical based in Canada. They are the first in their field to 3D print 3 anatomically accurate perinatal lacerations for use in medical training. This combined with VR training could drastically improve post labor and delivery outcomes.
Addressing postpartum support for breastfeeding
Breastfeeding and bonding with a new baby after an intensive labor or caesarean can be challenging without support. Yet, decades of research shows the current medical model does not support women in successfully breastfeeding, despite the evidence supporting the benefits.
Enter Nest Collaborative, a virtual care platform supporting women pre/postnatally with breastfeeding. By partnering mothers with expert lactation consultants, Nest Collaborative aims to reduce the racial disparity and support moms 24/7 with breastfeeding.
As the we begin to plan for the secondary impacts of this past year, women should emerge as the most important sector of consumer purchasing, specifically in healthcare. While gains have been made to address inequities, there is much room for growth. Wise healthcare companies will capitalize on this disparity and position themselves at the forefront of the female consumer to promote better outcomes and increased VC funding for Femtech.
Reproductive Medicine: Fertile Ground for AI Advancement
Reproductive Medicine: Fertile Ground for AI Advancement
Reproductive medicine has made tremendous progress over the last several decades. The industry has averaged a compounded growth rate of approximately 6.5% during the last decade and is expected to grow by as much as 10.5% annually over the next decade. Advances in embryo culture and genomics have significantly improved treatment safety and treatment outcomes. Gone are the days of reproductive endocrinologists transferring numerous embryos in hopes of one implanting. And fortunately, the days of neonatal intensive care units taking care of extremely premature multiple gestations that resulted from such reckless behaviors have passed. Genomics helps to minimize the risk of passing genetic diseases onto one’s offspring. By eliminating karyotypically abnormal embryos, fertility specialists have reduced miscarriage rates and the need to transfer multiple embryos with the goal of achieving a viable pregnancy. The current state of fertility care is marred by the lack of efficient and reliable data collection. The inability to apply artificial intelligence (AI) and machine learning (ML) limits our ability to diagnose the root causes of infertility and limits the ability to tailor treatment protocols that are unique and personalized to each patient more efficiently.
Lack of attention to data is limiting advances in reproductive medicine:
The fertility industry is expected to grow to a market value of $33.9 billion by 2030, but largely exists as a collection of private clinics. Lacking institutional funding and organization, there is little consensus on how data is collected and shared. Due to the lack of public funding of fertility care there are few mandates that require the use of Electronic Medical Health Records Systems (EHRs). Approximately half of all clinics operate in pen and paper models and as such their data is relatively useless for data analytics. To compound the problem, the vast majority of EHRs available in reproductive medicine are legacy systems that are server based and are incapable of collecting accurate or reliable data points.
The ability to collect large reliable data sets is further hampered by the lack of agreement on how to define diagnostic criteria and the lack of consensus regarding laboratory units of measure and grading criteria for embryos. HIPAA laws further limit collaboration between clinics.
Creating an actionable solution based on today’s AI technology:
The fertility industry is fiercely competitive for consumers. In the United States, intended parents will spend approximately $45,000 to achieve a pregnancy through reproductive technologies. Consumers seek care from providers who are “perceived” to be on the cutting edge of technology and have high pregnancy rates. These are drivers that can be harnessed to create systems to improve the capture of reliable data to be applied to analytical algorithms.
There are several EHRs emerging that might be utilized to creating the backbone of such an endeavor. Fertility Pro, N-Able and IVFqc are cloud-based systems that are agile in nature but have been poorly received by the end users. Fertility Pro (disclosure: this company was founded by the author) is cloud based and is designed to accommodate data analytics and machine learning. The system’s wireframes were designed with teams in Australia, North America, and Europe so it was designed to incorporate the workflows across the globe.
Regardless of which EHR emerges as the industry standard, there are some key approaches that are needed to bring the power of AI to the fertility industry. First, the system needs to operate intuitively. Users need to be able to harness the full capabilities of the system from day one. Much like users of smart phones – who reads a user manual? Most simply pick up the phone and start using it without difficulties. To be used, the EHR needs to make the lives of patients who use the patient portals to the healthcare professionals easier. The users need to be engaged. Advisory boards need to establish criteria to define diagnostic criteria and standardize units of measure and descriptive terms for sperm and embryo grading. The reality is that not all clinics will agree to these standards. But, by restricting the access of the AI capabilities to those who rigorously follow established protocols has a high likelihood of creating a significant user pool. This will be a simple business solution for many. Fertility patients are savvy and are drawn to those providers incorporating AI into their diagnostic criteria, treatment protocols and embryo selection criteria.
AI is highly dependent on large data sets to train algorithms. It will require many clinics to pool their data. An international collaborative effort would be ideal in creating large training datasets, but this is limited by local healthcare information protection rules and national regulations on where data is hosted. Fortunately, with the emergence of Kubernetes, software products are transferred to remote markets and managed from the developer’s host country. Furthermore, federalized data utilizing block chain technology will allow programs to access data in an anonymous manner in a foreign country’s cloud and share only the de-identified data to the host shows promise in helping create meaningful datasets. Harnessing data from providers that agree to well defined definitions of terminology and data entry criteria will help clinicians and data scientists to better individualize patient diagnostic criteria and treatment protocols.
Developing data-based solutions:
Just over forty years ago, the first IVF baby Louis Brown was conceived. The fertility industry has transformed itself and is at the forefront of technological innovation by incorporating genomic innovations and is starting to incorporate AI evaluation of sperm and embryos, but the industry has not embraced AI/ML/BD regarding patient evaluations, diagnosis, and treatment protocols. We can do better. If clinicians can embrace the same collaborative models that big technology companies have regarding sharing resources and data and building ML engines and datasets, then the fertility industry stands to create better outcomes for all intended parents. We must also be mindful of minimizing data biases and take strides to improve equality of all intended parents. The goal is to expand the field of reproductive medicine into new untreated communities. By increasing efficiencies and the safety of treatments, patients are the clear beneficiaries.
Shifting Surgical Assessment from Antiquated to Advanced Technology
Shifting Surgical Assessment from Antiquated to Advanced Technology
An estimated seven million people worldwide suffer post-surgical complications each year. In the United States, each major complication was associated with an average cost of over $11,000 to the healthcare system. The New England Journal of Medicine also found that bariatric surgeons with low technical skills had complication rates three times higher and mortality rates five times higher than those with high technical skills. What can we do to ensure surgeons with high avoidable complication rates are not continuing to burden the system? Research has shown that using virtual reality to train surgeons “improved overall surgical performance by 230%”. Can we leverage VR technology to also evaluate surgeons? How do we incentivize parties to consider this approach? OssoVR is a tool at the forefront of Virtual Reality training for surgeons, but I propose it could be used to make surgical board exams more relevant and effective. Through Mimic simulation, the DaVinci robot also provides a way to assess surgeons; perhaps this tool could be used to assist payors with insurance contracting and/or aide hospitals in the credentialing process to ensure patients are provided the best care. Finally, practitioners could use technology to provide transparency to patients about surgical abilities and outcomes, something that has historically been very opaque.
Board Certifications for Specialty Surgeons
Board certification historically began in ophthalmology with an examination. The original purpose of board certification was “to protect the public from practitioners without the knowledge necessary for up-to-date and safe practice”. Today, trained surgeons spend months studying for qualifying and certifying exams. But why are individuals with years of training and experience being subjected to such rigorous written and oral exams that may not be relevant to actual physical surgical skill, particularly with 42% of American physicians experiencing burn-out and burdened by an average of $150,000 in student loan debt? While certification is important in protecting patients, do board certification exams improve surgical outcomes or do they take away from surgeons’ focus on patients? Could a more streamlined method of evaluation minimize lost productivity and undue stress on physicians preparing for board exams? Anecdotally, we examine the board certification process for Female Pelvic Medicine and Reconstructive Surgeons (urogynecologists) who require four years of OBGYN residency training followed by three years of subspecialty training after medical school and prior to independent practice. Following U.S. Medical Licensing Examinations, urogynocologists are required to take both written and oral board exams for OBGYN, followed by both written and oral exams for the urogynecology subspecialty which evaluate clinical knowledge and an understanding of research methodology. For a surgical subspecialty, none of the four exams measure a physician’s surgical ability and objective patient outcomes. As a patient, I am not concerned with how well my surgeon can defend her thesis to the board; I would much rather know that she is adeptly capable of performing my surgery. Theoretically, a well-trained surgeon would much rather be objectively tested on her surgical ability than on research methods or other topics that, for that individual, are nowhere near as important to patient outcomes. I believe using virtual reality or simulations would be a much more effective way for medical boards to gauge specialty surgeons, alleviate pressure of studying for exams and redirect physicians’ focus on patient care.
Hospital Credentialling and Payor Contracting
According to the National Center for Biotechnology Information, “credentialing ensures that patients receive the highest level of care from professionals who have undergone the most stringent scrutiny regarding their ability to practice medicine. Credentialing also ensures that all healthcare workers are held to the same standard”. Without periodic assessment of surgical skills and outcomes, how stringent or standardized can this certification be, particularly for sub-specialty surgeons? Conducting one “virtual” surgery every few years to certify a surgeon’s skills are up-to-date would not only be a minimal administrative burden to any capable and experienced surgeon, but also an extremely effective way to ensure surgeons are unimpaired and performing enough surgery to keep skills fresh and relevant. Both payors and hospitals stand to gain from reducing surgical complications. For hospitals, profit margins from reimbursements declined by approximately 20% when patients were negatively impacted by complications; for payors, complications resulted in increased costs with an average increase in reimbursement of over 50%.
Independent Certification and Transparency Around Surgical Skills
The U.S. healthcare market has been characterized as “shrouded by obscurity around costs, prices, and quality”. While transparency around quality and surgical outcomes has never been at the forefront of the industry, virtual reality and simulation technology affords hospitals and practitioners the means to provide data around surgical capabilities. In fact, in a Kaiser survey about what is most important in determining the quality of health care received, consumers chose “qualifications of a doctor” as the most important factor. Understanding this, well-performing physicians and hospitals should be incentivized to provide assessment outcomes to patients as a way for patients to choose surgeons based on results or outcomes. To patients, this would be viewed as a strong signal as to whether their surgeons are skilled. Considering journalists can make it into America’s “Top Doctor” rankings, there must be a better way for patients to evaluate options.
While there has been research around how virtual reality can assist with training surgeons, the possibilities for assessment are unchartered territory. Ultimately, the current methods of assessment and certification for surgeons, while well-intentioned, are incredibly antiquated. Using technology for assessment could result in better surgical outcomes, lower system costs, and greater standardization and transparency. Now that we have the technology to improve patient outcomes, why aren’t we using them more extensively?
Digital Engagement Empowering a Personalized Patient Experience
Digital Engagement Empowering a Personalized Patient Experience
Complex medication regimens, difficulty of care coordination, and low engagement between doctors and patients are among the many challenges of chronic illness management. These challenges contribute to the high prevalence of treatment nonadherence in the chronically ill population, leading to poorer health outcomes and an additional annual spend of $100 to $300 billion of U.S. health care dollars. Patient empowerment, in the form of disease and treatment education and accessibility to one’s health data, is critical to reversing this trend of treatment nonadherence as informed patients are better able to advocate for the care they need. A number of digital health trends have the potential to transform how patients approach chronic illness management.
Immersive Technologies to Enhance Patient Education
Low health literacy in the U.S. poses as a challenge for patient education, but virtual reality can enhance patient education through experiential delivery of information. Virtual reality applications can help patients visualize how the human body works down to the cellular level and connect the science with the symptoms they are experiencing in their daily lives. With a better understanding of what their body is experiencing, patients may be more likely to see the importance of disease management and become more engaged with their care.
Augmented reality can help patients overcome frustrations related to complex medication regimens by providing the appropriate instructions as patients progress through each step of the treatment process. For example, a patient performing a self-injection can use AR to identify an appropriate injection site, confirm that the needle gauge they’re holding is the correct one to use, and ensure the correct amount of medication is drawn up.
Wearable Devices to Improve Monitoring and Provide Relevant Education
With the increasing popularity of wearable devices like Apple Watch and Fitbit, patients are being accustomed to having continual access to their health data. Patients can share the health information captured on the devices with their doctors, giving doctors more insight on the patient without burdening the patient to actively record the information. Furthermore, companies producing wearable devices have an opportunity to provide patient education in real-time as it relates to patients’ specific metrics. Providing patients with relevant information at the right time can better engage patients and facilitate understanding of their health information.
Telemedicine to Facilitate Personalized, Coordinated Care
With 4 in 10 adults in the U.S. having two or more chronic illnesses, it is critical that a patient’s healthcare providers can provide a coordinated, holistic care plan that takes into consideration the complexities of the patient’s comorbidities. Telemedicine provides a channel for a patient and their healthcare providers to come together despite geographic differences to develop a care plan that the patient understands and can commit to. With telemedicine, patients have greater access to doctors beyond their geographic area and can shop around for a doctor who they can trust. For patients whose native language is not English, that may mean a doctor who can speak their language and provide culturally relevant education.
The unique value that digital health brings to the mission of addressing medication nonadherence and poor health outcomes is the ability to empower patients in a way that is relevant to their experiences. Patients can learn about their body and treatment experientially through VR/AR, understand what all of the health data that is being collected actually means to them and how they’re feeling, and play an active role in building and communicating with their care team. Patient education is a key step to engagement and an engaged patient who knows how to advocate for their care may be more successful in managing their chronic illness. However, to be successful, digital health companies can never lose sight of the patient experience when they are building and reiterating on these patient empowerment strategies. Already burdened with the complexities of chronic illness management, patients will have no tolerance for non-user friendly and irrelevant products. Companies need to meet the patient where they are, whether that’s designing patient education to account for low health literacy levels or making product choices that are appropriate for patients impacted by the digital divide.
Designing effective patient solutions will require intentionality and a laser focus on the historically neglected patient perspective, but the healthcare industry cannot shy away from this challenge. Not only does it have an obligation to provide effective care for chronically ill patients but doing so will also lower the costs and stress on the U.S. healthcare system.
Will Amazon Win It All?
Will Amazon Win It All?
Founded in 1994 in Seattle, Amazon started out as an online bookstore. Just 25 years later it is the largest e-commerce leader in the world with a market capitalization of $1.6 trillion and has penetrated the fresh grocery and nutrition market by acquiring Whole Foods. In 2018, together with Berkshire Hathaway and JPMorgan Chase, Amazon co-formed the non-profit joint venture, Haven Healthcare, aiming to “provide employees and their families with simplified high-quality, and transparent health care at a reasonable cost”. Despite the companies’ combined 1.2 million U.S. employees and incredible market power, Haven never fully got off the ground and disbanded in Jan 2021. However, during the same time period, Amazon, the e-commerce giant, has increasingly sought to take on the $3 trillion healthcare market on its own.
Amazon Alexa
Amazon announced new Alexa healthcare skills on April 4, 2019. Amazon Alexa is the only HIPAA-eligible voice assistant on the market to provide administrative, technical, and physical safeguards for personal health information (PHI). The HIPAA Self-Service Program allows healthcare providers and other developers to build the skills that meet specific needs of patients. For children who had cardiac surgery, Boston Children’s Hospital created the Alexa-built instructions tailored to the children’s needs for caregivers to follow. They can share the post-operation progress with doctors without having to physically go to the hospital. For senior population living alone at home, ChristianaCare Home Care built Alexa skills to remind them to take medication and instruct them to do physical therapy without in home visits during the pandemic. Alexa Health connects customers at every stage of their health and wellness journey.
Amazon Pharmacy
Amazon launched Amazon Pharmacy on November 17, 2020, two years after its acquisition of PillPack, with two offerings: 1) online pharmacy store, allowing customers to complete the entire pharmacy transaction on their desktop or mobile device via Amazon App; and 2) Amazon Prime prescription savings benefit, providing free two-day delivery and up to 80% savings when paying without insurance. Amazon Pharmacy allows customers to order commonly used medications, including both generic and prescription medications through convenient and reliable access with price transparency, without leaving home. Customers also have online self-service options combined with phone access to consult pharmacists (24/7). Amazon has the opportunity to simplify the pharmacy supply chain, improve the customer experience, and lower the cost, which matters for patients, payers, and manufacturers.
Amazon Care
On March 17, Amazon announced that its health care service, Amazon Care, is rolling out in all 50 states starting this summer. Amazon Care Pilot Program, launched 18 months ago, has provided Amazon employees and their families immediate access to high-quality medical care including both virtual care (connecting patients to medical professionals via the Amazon Care app) and in-person care (dispatching a medical professional to a patient’s home). Amazon Care joined with Intermountain Healthcare and Ascension along with other health systems and home care companies to form the Moving Health Home coalition, which aims to lobby Congress to make permanent changes to home health care reimbursement policies.
Amazon’s strengths and its approach
Amazon’s strategy of building an independent healthcare company “free from profit-making incentives and constraints” was empowered by its ecosystem and cash to be deployed. It has reinvested its revenue into massive infrastructure build-outs in the logistics and data center spaces (e.g. AWS for enormous healthcare data loads and analysis, fulfillment centers/supply chain/Whole Foods for distribution of healthcare goods and services).
Amazon has a large testing ground and aggregated consumer demand. It has a direct distribution advantage to over 300M active customers, 100M Prime Members, and approximately 5M sellers on the site. Amazon has impressed its customers with its convenience and fast speed. Even without the Prime membership, the customer satisfaction could be leveraged to persuade customers to switch to Amazon’s healthcare offerings.
By providing customer-friendly products and patient-centered services, Amazon can build economies of scale, network effects, and leverage for negotiating with other parties (e.g. suppliers, providers, insurers) and influencing policy making. From a partnership perspective, Amazon is providing platforms and outsourced version of services on a rent-to-own basis (e.g. Fulfilled By Amazon, AWS) which are traditionally expensive to build/develop independently. The cross-side network effects with platforms allows Amazon to create transparent and competitive markets for both buyers and suppliers. Amazon can also standardize suppliers’ offerings on its platform and use this standardization as a means of scaling up healthcare.
Lack of standardization, transparency, and focus on consumer experience has resulted in an incredibly fragmented, opaque market in health care. With Amazon’s entrance, companies focusing on information formatting/coordinating, relying on opaque pricing, and treating customer experience as an afterthought will be particularly vulnerable. Middlemen that are value extractors and have large profit margins (e.g. Pharmacy Benefit Managers) are under risk.
What’s next?
Amazon just started its disruption in healthcare. There are many areas it can continuously improve. Amazon can build on infrastructures by having all licenses (e.g. wholesale pharmacy license) and logistics pieces (e.g. cold chain or temperature controlled logistics to transport drugs that have specific environmental needs) in place; develop platforms and health management systems such as claims management systems for payers (e.g. detecting inaccuracies in submission), benefit management systems for self-insured employers (e.g. stop-loss insurance), and lifestyle management system for Medicare/Medicaid populations (e.g. Alexa daily checking buildout for patients having chronic diseases or mental health issues without physically hospital/clinic visiting); and create synergies such as providing in-person health care at Whole Foods and adding meal tailoring to Amazon Fresh to capture values in the health and wellness market.
Amazon’s approach into healthcare has already caused existing players to scramble and reevaluate their core competencies. Amazon’s offerings allow them to outsource non-core parts of their businesses and focus on their core services. Pharma/Biotech can outsource the packaging and transportation to Amazon and focus on research and development for new drugs/treatments. Providers can outsource documentation management (e.g. EHR) and supplies to Amazon and focus on providing care to patients. Payers can outsource claims and focus on offering their services.
Amazon is particularly well-positioned to change the healthcare space. Its entrance into healthcare will either change how the system is designed or force incumbents to become more competitive.
New Angle in Health Care: Can Digital Technology Deliver on the Triple Aim and Iron Triangle?
I believe that the question of whether the “iron triangle” and “triple aim” are consistent is a question whose answer has changed over time as new advancements in technology have radically changed the potential of healthcare delivery. It is due to recent advancements such as artificial intelligence (AI), connected wearable devices, remote patient monitoring, telehealth, blockchain, affordable genetic sequencing, electronical medical records, and interoperability that I think we can now make a strong argument for the potential future creation of a healthcare system in which there is consistency between the “triple aim” and “iron triangle”. It will be possible to have a focus on population health, experience of care, and per capita cost, while also pursuing cost containment, access, and quality. This consistency can be achieved through digital-only and hybrid (digital/in-person) integrated practice units (IPUs) that are enhanced by AI, are focused on value/outcomes, and have innovative payment models (bundled payment or data monetization).
A strictly digital, AI-driven IPU that is interoperable with wearable devices and is focused on basic, whole population level preventative care, such as lifestyle habits and basic health monitoring, could be developed for the entire population. This digital IPU could be a mobile application that is always running in the background, monitoring vital signs, collecting data on lifestyle habits, and cross-referencing against genomic/phenotypic data, using AI to offer a suite of personalized health suggestions and warnings (with referrals) if anomalies are detected in the individual’s health data. As wearable devices become increasingly sophisticated as health diagnostic instruments, the digital IPU’s focus on preventative care would help to keep the entire population healthy so that there is lesser need for more expensive acute care. Innovation and competition in wearable devices will eventually lead to a reduction in price, allowing increased adoption at the population level. Since this digital IPU is powered by a single AI system, the quality of care would be consistent across the population and could be upgraded continuously as new data is gathered. Further, the marginal cost of adding a new user to the digital IPU would be minimal, making it economically feasible to operate and enabling universal access. While this digital IPU has the potential to drive down long-term healthcare costs by focusing on preventative care, there seemingly must be some sort of tradeoff to make it affordable (or free) for everyone to use. An interesting, yet very controversial, approach to making it affordable to everyone would be to take a page from “big tech” by monetizing user data. Consumer health data is very sensitive information and should never be taken lightly, but with robust safeguards in place (including anonymity of data and personal ownership through blockchain), personal health data could be a valuable currency that would make a population level basic health digital IPU affordable to all.
This basic health, population level digital IPU, could be designed to be interoperable with hybrid (digital/in-person) IPUs that are focused on specific conditions (e.g., cancer treatment, knee replacement, diabetes management) or specific population subgroups (e.g., healthy over 65-years-old). Unlike the whole population digital IPU described above, the hybrid, condition specific, IPUs would not be “free”, they would be paid for via traditional private insurance, public insurance, or out-of-pocket. However, to drive competition and to dissuade physicians from pursuing volume-based care, these IPU providers would participate in a bundled payment system and would publish outcomes/cost data so that patients and payers could make informed decisions. IPU or traditional providers who do not provide the best patient value would be driven out of business, helping to reduce wasteful spending on low-value care and enabling the top providers to grow and reap financial reward. The provider competition would also serve to reduce prices to market clearing levels, assisting in the reduction of overall healthcare spending. In this hybrid IPU design, the “hub-and-spoke” model, as described by Michael Porter and Thomas Lee, may remain, but the centralized hub component could be replaced by a digital platform powered by both AI and condition specific human specialists. This platform would help standardize quality care across patients, reduce cost by eliminating unnecessary overhead such as real estate, and improve patient outcomes via specialization of care. There would be a lesser need for clinician offices (“spokes”) because people could be monitored in their homes and have AI continuously monitoring patients (via wearable and other in-home connected devices). The AI could be set-up to automatically notify IPU clinicians of needed intervention or inform the patient if additional care is required, either in-home or in a clinician office. Furthermore, clinicians could choose to meet with patients in person or via telehealth visits, reducing travel costs, improving access to care for people not located near a clinic, and potentially improving the satisfaction of care.
As this analysis shows, there is in fact a potential path forward in which the “triple aim” and “iron triangle” are consistent. While the path forward requires continued technological advancement and adoption in healthcare delivery, this past year has shown us that such a situation is not only feasible, but desired by much of the population.
Is AI finally delivering on its promise of accelerated drug discovery?
Is AI finally delivering on its promise of accelerated drug discovery?
In February 2020, the first Artificial Intelligence (AI)-created drug for Obsessive Compulsive Disorder, DSP 1181 entered phase-1 human clinical trials. It was co-developed by a UK-based start-up Exscientia and Japanese pharmaceutical firm Sumitomo Dainippon Pharma in just over one year. This was followed by another announcement by Exscientia in April 2021 about the first AI-designed molecule for immuno-oncology, the A2a receptor antagonist, developed in less than 8 months.
Typically, the entire drug development process can cost a pharma company $2-3 billion, and it can take 12-18 years for a drug to go-to-market. AI and Machine Learning (ML) technologies promise to induce cost and time efficiencies in the process, particularly in preclinical and clinical stages where success rates are currently under 10%. AI can help identify more promising chemical entities, enable analysis and testing of drug structures to fasten drug development processes.
AI coupled with advanced computation modelling has also played a pivotal role in vaccine development for Covid-19. Researchers at the University of Texas at Austin and National Institute of Health used neural network- based software to analyze protein structures of SARS-CoV-2 using 3D modelling. The Covid-19 pandemic has underscored the importance of investing in healthcare for faster and cost-efficient solutions. Therefore, in 2020, investments to bring the power of ML to drug discovery soared to $13.8 billion, more than 4.5 times that invested in 2019, according to the Artificial Intelligence Index, an annual report produced by Stanford University’s Institute for Human-Centered Artificial Intelligence (HAI).
Big tech investments in pharma
Patient records were handwritten when Google initially tried to break into the healthcare industry; now they are all computerized. Cloud computing and machine learning algorithms that are applied to patient data have formed the backbone of the AI revolution in the healthcare industry. This has played to the advantage of the largest IT firms, who have access to massive data stockpiles attributed to the advent of digital devices and spread of digitization.
Top-tier tech companies are increasingly turning their eyes towards pharma and are seeking to leverage their technical prowess in AI and cloud technologies to become partners for drug manufacturers. Their scale, availability of technical talent and huge cash reserves enable them to make an impact in the space. The pharma drug discovery market is touted to be a $64 billion opportunity, which is motivating tech giants to disrupt the industry. In September 2019, Novartis CEO, Vas Narasimhan, who seeks to reimagine medicine with data and digital, announced a partnership with Microsoft to launch the AI Innovation lab to accelerate drug discovery using deep learning. Google has entered in a joint venture with Sanofi to launch a virtual innovation lab for drug discovery as well. With Project Baseline, ran by Alphabet’s life sciences arm Verily and through a partnership with Ascension Health, Google is gathering vast swathes of patient data, which will help in clinical research. Google’s DeepMind AI technology developed a revolutionary algorithm, Alpha Fold, to predict 3D structures of proteins. In the UK, Vodafone partnered with Imperial college to facilitate AI-enabled research through crowdsourcing. By using AI for smart reallocation of computing resources, the project led to identification of 100 anti-cancer molecules in just a few months.
Rise of startups focused on AI-enabled drug discovery; partnerships with biopharma
Technological progress in form of low-cost and accessible computing resources has led to the rise of Healthtech startups focused on drug discovery. Biotech investors have also been bullish on AI-driven biotech firms, and they invested over $2 billion in such firms in 2020. Large pharma companies are also collaborating with AI startups to alleviate roadblocks to drug discovery. In 2020 alone, almost 27 new partnerships were forged between AI vendors and pharma, a 575% increase over the last 6 years. Atomwise, an AI platform for small molecule R&D, has partnered with Eli Lilly, Merck, Bayer, Pfizer, and Abbvie. Its platform is helping identify new lead compounds in days, bypassing the need for costly and long high-throughput screening experiments. Healx, a UK based startup, is using AI to help researchers repurpose existing drugs for rare diseases. Toronto-based Cyclica uses AI to understand biophysics and speed drug design through analysis of the polypharmacology, pharmacokinetics, and structural pharmacogenomics of molecules. Bayer and Merck have partnered with Cyclica to improve drug target identification. The startup, Exscientia has set up partnerships with giants such as GSK and Sanofi, and it has over 20 drug compounds in development.
There are over 200+ startups working on drug discovery from helping understand disease mechanisms, repurposing existing drugs to designing new drugs. US and UK lead in terms of investments in AI tech companies; however, the AI revolution has seen the rise of Chinese tech firms in healthtech. Quicker regulatory approval and widening market access coupled with a focus on becoming a global leader in AI, have accelerated Chinese investments in this area. It currently has the largest number of AI R&D centers in drug discovery; therefore, China is poised to lead the world in drug discovery in the forthcoming future.
Challenges and the road ahead
The use of AI in drug discovery though promising is posed with several challenges and ethical considerations. As newer data sources are used and human health data is utilized for drug discovery, there are concerns about patient data privacy. AI-modelling based results rely heavily on data; thus, it is critical to have quality datasets such as high-resolution images for 3D proteins and heavy investments are required for computationally intensive tasks. It is also important to address any inequalities in data due to genetic differences such as investigating side-effects for rare diseases while developing drugs. The lack of high-quality data remains a barrier to deep learning systems reaching their full potential. Tech companies would also need to invest in creating systems for improving data collection and quality in healthcare organizations, while being cognizant to not overwhelm the healthcare system with a deluge of changes. As AI startups, pharma companies and tech giants share information and collaborate, new systems, policies and regulations would have to be instituted to ensure data safety and privacy. The outcry over Google’s partnership with Ascension Health highlights that tech will have to invest resources in winning the trust of patients and approval of regulators. International authorities would also need to collaborate to achieve parity in rules and regulations for ensuring adoption of AI-designed drugs in the future. As AI becomes the inventor and develops new products or repurposes old drugs, companies will have to reconsider how they attribute ownership rights or protect intellectual property. The risk of error in healthcare is high and the “black-box” difficulty of AI poses a challenge here, especially in unsupervised learning experiments. Therefore, in the near- term AI can only augment human intelligence and not substitute it as much of the data in healthcare is still biased, incomplete or not interoperable requiring extensive validation.
Nevertheless, as these challenges are tackled, the future looks promising for AI in drug discovery and we will likely witness collaboration between academia, AI companies and pharma, lower drug development timeline and costs. This would eventually help make healthcare more accessible and improve outcomes, especially in areas for which there are no treatment options currently.
The Role of the FDA in 3D Bioprinting: Innovation and Regulation
Introduction
Preclinical testing of new drugs or biologics in animal models has been the norm since the passage of the 1938 US Federal Food, Drug, and Cosmetic Act. The assumption is that the effects of these compounds on animals is a good predictor of their effects on humans. The evidence, however, shows a poor correlation between animals and humans. A significant portion of new drugs eventually fail human clinical trials (89%) and about half of those failures can be attributed to unanticipated toxicity in humans. If preclinical animal testing for toxicity was closely correlated to human toxicity, we would expect to see significantly less trial failures due to human toxicity. On the flip side, compounds determined to be toxic in animal studies are usually not developed further--potentially wasting drug candidates that might have been useful. 3D bioprinted human tissue models could potentially provide a better toxicity and efficacy filter for drug candidates in preclinical tests. In the slow-moving and heavily regulated environment of biopharmaceutical research and development, the FDA needs to do more to incentivize and support research into animal model alternatives.
Incentivizing Alternatives
Similar to the launching of the Digital Health Center of Excellence in support of digital health technologies, the FDA should formally support investigations into alternative preclinical testing models like 3D bioprinted models through the establishment of a task force to coordinate the conversation. Biopharmaceutical companies are set in their ways due to historical norms and regulatory requirements. Regulatory uncertainty over usage of bioprinted models for preclinical safety data further disincentivizes companies from exploring those alternatives. The FDA’s formal statement of support and signaling of willingness to work with companies exploring these alternatives would go a long way to clearing up the cloud of uncertainty surrounding the future potential regulatory pathway for testing done using bioprinted tissue, and thus the future commercial environment for bioprinting. Companies would more readily invest their money and efforts into an area where regulatory authorities are interested in developing new pathways for preclinical testing. This environmental shift would greatly benefit 3D bioprinted models by potentially increasing industry interest and allowing for more research into its effectiveness as a replacement or improvement upon animal models.
Beyond a broad signal of support, the FDA could also consider more tangible incentives to mobilize more rapid industry change as evidence of the viability of alternative models is found. The FDA could initially allow for alternative models to replace some animal testing and set a timeline for a transition to allow for voluntary full replacement of animal testing at certain milestones. They could also consider reductions in fees or a pathway for expedited review of investigative new drug applications.
Potential Benefits
The FDA incentivizing testing on 3D bioprinted models would only make sense if there were real benefits to such a switch. However, we think there are numerous benefits to this emergent technology, including decreased cost, increased safety, more drug candidates, and better 3D bioprinting technology. A major reason that the transition to bioprinted models would be beneficial is through cost reduction; estimates of potential savings range from 10-26% per drug. When drug development costs are frequently greater than $700M, this can lead to over $100M in savings per drug. One of the main drivers of this cost reduction is through success rate. This is in large part because human tissue is a better predictor of drug toxicity effects in humans than animal testing. In addition, eliminating drugs that prove toxic to human organs before they get to human testing protects trial participants from potentially adverse reactions. On the flipside of this logic, testing on bioprinted human tissue could allow for the discovery of drugs that are safe for humans, but are currently not considered because of their toxicity in animal models.
Finally, the increased use of 3D bioprinting brought about by the creation of an expedited FDA pathway would lead to significant advancements for the bioprinting technology itself. As the bioprinter manufacturers receive more revenue from increased demand for their products, they would be able to reinvest some of that money into R&D for more advanced printing technology. This in turn would allow the companies making the tissue models to use the more sophisticated hardware to expand their offerings of more complex and stable tissue structures.
As the range of tissues that could be simulated through 3D bioprinting grew, demand for bioprinted tissues would further increase. This positive feedback loop would create the kinds of conditions for the printing of even more complex structures, and the eventual 3D bioprinting of whole organs. Being able to print organs would save over 6,000 lives per year in the U.S. alone, and this revolutionary technology would be built on the back of research for bioprinting in drug development.
Conclusion
Ultimately, 3D bioprinting represents a technology that could have significant benefits to the efficacy of pharmaceutical trials. However, to catalyze investment and use of bioprinting in drug discovery, we suggest the FDA create regulatory incentives for companies to pursue this route. We believe changed regulations would allow the 3D bioprinting industry to flourish, benefiting both pharmaceutical companies and society as a whole.