Imagining Health Care in 2030

I would like to leave you with the text of forecasts made by a number of leading Harvard faculty on technologies that will characterize medicine in 2030, which was published in World Medical Innovation Forum program which took place on May 11. I feel it is important information or anyone considering what healthcare will look like going forward. Thanks to World Medical Innovation Forum team.

As a part of World Medical Innovation Forum, I had a chance to be looked for forecasts made by a number of leading Harvard faculty on technologies that will characterize medicine in 2030.

The last ten years have yielded remarkable advances in biology and medicine, from a new class of immune-based treatments for cancer to the
emerging use of AI to guide clinical decision-making.
But where will this wave of innovation take us in the next decade?

Drugging the Microbiome

by Lynn Bry, MD, PhD, Director, Massachusetts Host-Microbiome Center and Crimson Core, Dept. Pathology, BH; Associate Professor of Pathology, HMS

The vast community of microbes that live in and on us are more than just mere passengers. They are active participants in our health. This microbial universe, the microbiome, is increasingly recognized for its roles in exacerbating — and preventing — disease.

As a field, we have been sifting through the microbiome to determine which microbes are the key players. What parts of their biological makeup are important? And what are the molecular targets in the human body? Can we zero in on what is happening biochemically— the small molecules made by microbes, for example — and harness those as therapies rather than the microbes themselves? Once we understand the biological mechanisms, our power to help patients is even stronger.

Over the next ten years, these activities will start to pay off. Therapies will move toward precise combinations of microbes and, increasingly, their metabolites and small molecules. We’ll see microbial-based therapies aimed at various conditions — like improving the function of immune checkpoint inhibitors
for cancer treatment or lowering the risk of strokes and cardiovascular disease.
We now recognize that microbes release neurotransmitters as part of their normal metabolism. How do those impact your enteric nervous system, the network of nerves that controls your gut? Believe it or not, you have more nerves in your gut than in your brain. This gut-brain axis could prove to be an important therapeutic touch point in a variety of diseases.

By 2030, we’ll also see the first wave of approvals for therapies that deliver defined mixtures of different microbes for conditions such as food allergies, eczema and other conditions. A decade ago, making microbes
into medicines was almost unthinkable. But with the explosion of biopharma companies in this space, particularly here in Massachusetts, this effort is not just thinkable — it’s doable.

Putting Technology to Work in Hospital Rooms

Alistair Erskine, MD
Chief Digital Health Officer, Mass General Brigham

Technology is transforming multiple facets of health care. But one often overlooked aspect is the hospital rooms where patients recuperate from surgery or other serious health conditions. How will those change in the next decade?
First, large flat-screen TVs will be ubiquitous, acting as major hubs of communication. With always-on, two-way video conferencing, patients will stay connected with loved ones, and family and friends will be able
to keep a watchful eye on their recovery. These capabilities will also enable physicians, who often work at multiple, distant sites, to connect more readily with patients and their families.

Microphones will be deployed throughout the room — like miniature, mute Alexas —constantly listening and recording clinicians’ notes as well as orders for medication or blood tests. Such ambient voice technology will render in-room computers and keyboards obsolete, freeing up more time for face-to-face connection.

These video and voice features lay the foundation for a broad spectrum of AI-based tools — for example, to help monitor patients at risk for falls, a task that now lies in human hands with dedicated patient sitters.
Location-based services, the equivalent of an in-hospital GPS, will also help maximize clinicians’ time and efficiency, making it possible to track clinical equipment, identify a patient’s whereabouts (is she in x-ray or
resting in her room?), and help doctors, nurses, and other clinical staff meet up in person.

Importantly, this bubble of technology will not burst when patients head home. Rather than being discharged with a stack of paper and a few prescriptions, patients will be given a disposable patch or other wearable device that can measure their vital signs and signal for help if they are not well. That way, we can extend this state-of-the-art support network beyond the hospitals’ walls.

Expanding the Toolbox for Cancer Drug Discovery

Keith Flaherty, MD
Director of Clinical Research, MGH; Professor, Medicine, HMS

When it comes to defeating cancer, our toolbox is woefully small. For example, of the roughly 20,000 protein-coding genes in the human genome, only about 45 are currently targeted with a cancer drug or other molecularly honed therapeutic. When you consider the full molecular machinery that fuels cancer growth, our current defenses amount to a handful of dents. Of course, the ideal anti-cancer armamentarium doesn’t necessarily need to number in the tens of thousands, but it should certainly contain hundreds. How can we resolve this dramatic mismatch by 2030?

First, we need to understand the full spectrum of cancers’ vulnerabilities. To do that, we must be able to scrutinize tumor cells isolated directly from patients, not those that are grown in the laboratory. Just as antibiotics are screened for their sensitivities — by collecting a patient sample, isolating the bacteria, and exposing them to antibiotics to see which ones are effective — we can apply similar principles to patients’ tumors. This approach would allow us to catalog all the ways that patients’ tumors are vulnerable — and importantly, to zero in on the vulnerabilities that lack corresponding drugs. Teams in academia and
industry are now working to accomplish this.

Yet once these cancer vulnerabilities are identified and validated, probably somewhere on the order of 200 to 300, we’ll need to massively accelerate the process of drug discovery. That sounds like a pipedream,
but with recently developed methods in chemical proteomics, it is becoming feasible. Researchers can rapidly and efficiently screen proteins — thousands at a time — for potential drug-binding sites and develop small molecules that bind to them. These methods are beginning to flip the script on how chemistry has been traditionally applied to drug discovery. Hopefully, they can also help us gain the
upper hand against cancer.

A Three-Legged Stool and the Future of Oncology

Daphne Haas-Kogan, MD
Chair, Department of Radiation Oncology, BH; Professor of Radiation Oncology, HMS

Cancer care and treatment have completely transformed in the last 25 years. During my clinical training, I watched too many patients succumb, often quickly, to their disease. Fast forward to today, and even with advanced or aggressive forms of cancer, many patients go on to live long, happy, productive lives.
Of course, much progress still lies ahead. Over the next ten years, I see our efforts focused in three key areas, which all join together to drive progress in oncology. The first leg is AI and machine learning. With these technological advances, we will be even better equipped to plan patients’ treatments and determine which
approach is best for which patient.

While AI-based methods will unburden oncologists from some of the more laborintensive, time-consuming tasks, deep scientific and clinical expertise will be especially critical. That’s because we’ll have to judge the validity of what the algorithms tell us. After all, those algorithms are only as good as the data used to design them, and as AI end-users, we won’t always know if that data were good or bad. So expert knowledge of oncology, both its underlying science and clinical care, represents the second leg of the stool.
The final leg is the rapid progress we’ll see in personalized cancer care. This personalized approach has many inputs, including the molecular biology and genetics of patients and their tumors as well as the anatomy of their organs and tissues. It is about choosing which treatments are likely to work best while
avoiding those with the greatest potential for harm, whether short-term or long-term.

It is also about treating our patients and their families as the individuals they are, with unique psychological and emotional needs, and partnering together to heal the whole person.

The Ethical Dilemma of Imperfect AI

Jayashree Kalpathy-Cramer, PhD, Director, QTIM Lab, MGH; Associate Professor of Radiology, HMS

Is imperfect health care better than no health care? As we look ahead to 2030, when AI and machine learning will be more embedded in medicine, we’ll need to deeply consider this question.

Today, the majority of data used to train machine learning algorithms are “WEIRD” — derived from Western, educated, industrialized, rich, and democratic countries. If those training data contain any bias — for example, sampling more white patients than non-white patients — we know that the resulting AI can propagate and even amplify those biases. For example, we’ve seen this manifest as racial disparities in the performance of speech and face-recognition algorithms as well as algorithms used to allocate health services.
Those have been chilling wake-up calls.

At the same time, we already live in a world where billions of people lack access to basic health care. For example, India is home to over 1 billion people, yet has around 200 or so pediatric ophthalmologists to care for its youngest residents — an order of magnitude fewer per capita than in the US where the
disease is much less prevalent. Across the globe, babies go blind simply because they can’t get the diagnosis and eye care they need.

Over the last few years, my colleagues and I have developed an AI-based algorithm that can detect such highly treatable forms of eye disease in premature babies — in many cases, outperforming expert ophthalmologists.
Particularly in areas where health care access is limited, it could augment the skills of local health care workers, who often lack ophthalmology training, and enable them to provide better care. In India, that could mean preserving the sight of thousands of babies — and all of the opportunities that come with healthy vision.

So, as we consider when it is ethical to use our imperfect algorithms, we must also decide when it is unethical not to use them.

Widespread Whole-Genome Sequencing for Medicine

Elizabeth Karlson, MD Director of the Rheumatic Disease Epidemiology Research Program, BH; Professor of Medicine, HMS
genetics isn’t a part of everyday medicine. But in 10 years it will be.
Today, as clinicians, we assess patients’ family histories to understand what diseases run in their families — a proxy for what genes our patients likely carry and what genetic diseases they might be at risk for. We take
detailed personal histories, with questions about lifestyle, smoking, and diet, that also tell us something about disease risk. And we order lab tests to get a deeper sense of biological factors, like high cholesterol, that indicate an elevated disease risk. With all of this input, we devise our output: evidence-based recommendations to help treat or, even better, prevent disease.
Despite these efforts, our understanding of patients’ risk of disease is often incomplete. Family history can be an unreliable source of genetic information, particularly when it comes to common diseases. But a whole genome sequence — a readout of all of the letters that make up a person’s DNA — is much more precise. And, thanks to the wonders of biomedical technology and human ingenuity, it is also much more affordable than it was just five years ago.
By mining the information within patients’ genomes, researchers have discovered that it is possible to identify patients at highest risk for a range of common diseases, including heart disease, inflammatory bowel disease, breast cancer, and others. By expanding the use of this “polygenic risk scoring” approach, it is likely that even more conditions will be added to this list.
There remain significant barriers to overcome before whole genome sequencing can become a part of mainstream medicine, including issues related to data analysis, genetic privacy, clinical implementation, and, importantly, health equity. But we are already glimpsing its remarkable promise for one of the holy grails of medicine: disease prevention.

Transforming the Role of Radiologists in Detecting Intimate Partner Violence

Bharti Khurana, MD, Director, Trauma Imaging Research and Innovation Center, BH; Assistant Professor of Radiology, HMS

Consider a scenario that plays out millions of times a year across the world: a middle-aged woman walks into her local emergency room with a suspected forearm fracture. As part of her initial evaluation, she is asked the standard screening questions for intimate partner violence (IPV), to which she replies “no.” The ER physician orders an X-ray, the radiologist reads it, identifying a fracture in the right distal ulna, and writes up a report. The patient’s arm is put in a temporary cast. She is given a referral to an orthopedic surgeon and sent home.

The woman’s immediate injuries were treated, but what was missed? Her care team failed to recognize that she is a victim of IPV. But by 2030, with the help of an AI-powered decision support tool that my colleagues and I are now developing, it is my hope that such failures will decrease dramatically.

With AI, we can now harness standardized, evidence-based guidelines to help reduce the variability — and often, subjectivity — that underlies current radiology practice. Moreover, we can begin to integrate automated, imaging based tools with patients’ electronic health records, which in the case of IPV, will help
alert health providers to a history of recurrent injuries and other patterns consistent with non-accidental trauma.

By putting these capabilities to work, we can transform the practice of radiology. Not only will we be able to more readily detect the hidden signs of IPV, but we will also be liberated from a host of mundane, time consuming tasks — making it possible for us to become more deeply involved in patient care.
After all, isn’t that why we pursued careers in medicine — not to be squirreled away for hours in the reading room, but instead, to make a real difference in patients’ lives?

AI Serving Patients: Chatbots, Virtual Visits, and Medical Records at Your Fingertips

Adam Landman, MD, VP, Chief Information and Digital Innovation Officer, BH; Associate Professor of Emergency Medicine, HMS

Much attention is paid to how clinicians’ lives will be improved by AI. But let’s remember that AI can — and will — make patients’ lives better, too.
If you asked me a few months ago to predict how digital technology would improve patient care in the next decade, I would have said that the burden of appointment scheduling would move online, handled more quickly and seamlessly by chatbots and other AI-driven tools. And that triaging patients based on their symptoms and health histories would become automated, with chatbots advising: “Head to urgent care,” or “If you are not better in a few days, book an appointment.” That appointment could be a virtual visit — similar to a secure FaceTime call — and not in person.

But COVID-19 has accelerated the digital transformation of health care.
We used AI chat bots to help triage patients with possible exposure or symptoms to the appropriate care setting. And virtual care was the preferred first step to reduce the likelihood of viral spread. We achieved years of digital progress in a handful of months. Now, we must learn from our experiences and continue to use virtual care technologies when appropriate instead of reverting to our old ways.
Massive changes are also underway regarding patients’ medical data.

Fundamental policy shifts are moving toward a mandate that patients’ data be made available to them — and in electronic form. For example,
a new requirement gives patients the ability to connect their electronic health records to third-party apps, such as those which offer reminders to take prescription medications. Making medical data more readily
available is really stirring the pot of innovation, fueling the creation of novel health-related apps for patients. Apple has been a leader in this space with its Health app. Now, a handful of organizations are collaborating to bring the same data sharing and interoperability standards to Android devices.

As these efforts take root, even more data will be put in patients’ hands, such as the notes clinicians take during a clinical visit. There’s a price to be paid for such liberation — exposing patients’ medical information to potential theft and misuse — so data privacy and security will remain vital. But when it comes to their health data, patients will now be in the driver’s seat, choosing what and with whom they want to share.

What Will We Do with Our AI “Debt”?

Thomas McCoy, MD, Director of Research, Center for Quantitative Health, MGH; Assistant Professor of Psychiatry, HMS

“AI in health care” often means the use of supervised machine learning for clinical prediction. In that framework, investigators and innovators train models for an array of clinical tasks — from recognizing strokes to stratifying dementia risk — using machine learning algorithms applied to the health data generated through routine care.

The resulting trained models help predict the future based on relationships among facts observed in the past. Such reasoning presumes that historical relationships — codified in the data used to train the models — will repeat themselves: what has been will continue to be.
If we rely on these models in clinical decision making, that presumption becomes a self-fulfilling prophecy.

Trained models at the point of care may well deliver great value by homogenizing care — bringing lower-performing clinicians up to the level of the historical high performers. But this “history in a bottle” is ill-equipped to improve the state of the art in medicine and obscures the extent to which it perpetuates the status quo. For historically underserved patient groups, more deeply ingraining the status quo is a commitment to structural inequity.

For patients benefiting from breakthrough cures, the status quo is a commitment to disregarding the new hope and new health arising from new cures.
The models we train on our historical data are a form of debt. That technical debt, like any other, requires service. And that service comes as careful stewardship of algorithms and models, not just during creation, but over their full life cycle.
AI is a plausible means of achieving the critical goal of higher quality care at lower cost. So, we’ll need to take on AI debt if we are going to build the best possible version of health care in the decades ahead. 

As we do so, we must continue to earn patient and community trust — thoughtfully balancing risks and benefits —by dutifully making our monthly payments.

The Promise and Peril of Gene Therapy

Patricia Musolino, MD, PhD, Co-Director, Pediatric Stroke and Cerebrovascular Service, MGH; Assistant Professor of Neurology, HMS

Gene therapy is revolutionizing patient care, particularly for rare, single-gene diseases, which can be devastating for patients and their families. The development of one-time, potentially curative treatments for diseases like spinal muscular atrophy (SMA) and adrenoleukodystrophy (ALD) is a major achievement — decades in the making — that should be appropriately celebrated.

At the same time, as remarkable as gene therapy is, it is incredibly expensive. In fact, these are among the priciest arrows the medical community now has in its quiver. As a society, we will need to grapple with that reality.
But we also cannot rest on our laurels. As physicians and scientists, we must continually push the frontiers of biomedical innovation. For rare diseases like ALD, that means discovering new ways of halting, or even reversing the disease at its earliest stages — ideally, with a pill or other inexpensive and readily-delivered therapy. Because gene therapy will not be a viable therapeutic option for every patient.

Yet these innovations won’t wash away the ethical dilemmas that surround gene therapy. Actually, quite the opposite. With the rise of new gene-editing tools, like CRISPR, it will become possible to engineer ever more subtle and precise genetic changes within patients’ cells. In the next decade, we’ll see clinical trials emerge that apply gene-editing to the treatment of more common diseases, like heart disease and diabetes.

So, we must take seriously our social responsibility to ensure that everyone understands how these technologies work: what they can and cannot do, and what they cost. And we’ll need to create thoughtful policies and legislation to guide how they can be used in research and medicine. If we don’t, we will unintentionally widen the health disparities that already exist here in the U.S. and around the world.

Preparing the World for the Next Infectious Disease Threat

Rochelle Walensky, MD, Chief, Infectious Disease, Steve and Deborah Gorlin MGH Research, Scholar, MGH; Professor of Medicine, HMS

As the novel coronavirus SARS-CoV-2 has spread across the globe, we are now in the throes of a major pandemic. While it is impossible to ignore this viral threat and its impact on patients and health systems across the world, this crisis will subside. And in ten years, we’ll be staring down other infectious disease challenges.

What will those be? Perhaps one of the biggest threats is likely to be a known enemy: antimicrobial resistance. Public health experts predict that deaths due to antimicrobial resistance will outstrip cancer mortality by 2050. In order to address this problem in a serious, systematic way, we must change the
paradigm of antimicrobial drug development and use.

That means preserving the antibiotics we have — preventing widespread, indiscriminate use lest they become obsolete — and building a steady pipeline of novel antimicrobial therapies to hold in tight reserve for only the most severe, drug-resistant infections. And we’ll need to create the right incentives to get the biopharma industry engaged in this vital effort.

We’ll also need innovative diagnostics to help predict the susceptibilities of major infectious agents that result in things like sepsis, in a rapid timeframe, so that treatment can be quickly optimized. Importantly, we’ll also have to think beyond traditional antimicrobial drugs. From an evolutionary perspective, the microbes will always be a step ahead of us.

Of course, we cannot forget the health dangers posed by climate change. As just one example, rising global temperatures have expanded the geographic reach of tick-borne infections in the U.S. in the last 20 years.
How will other, more deadly infections, like malaria, yellow fever, and dengue respond? And what about the emergence of novel microbes, like COVID-19?
As this pandemic has proven, short-sightedness and lack of preparation put lives at risk. We cannot let that happen again.

A New Paradigm for Ear Health

Bradley Welling, MD, PhD, Chief of Otolaryngology, Mass. Eye and Ear, MGH; Walter Augustus Lecompte Professor and Chair of Otolaryngology–Head and Neck Surgery, HMS

The human ear is a marvel of nature, giving us the ability to hear and maintain balance. At the same time, it is housed in the temporal bone — the hardest bone in the body — which makes studying the ear and its components a significant challenge. 

While major strides have been made over the last several years to unlock the ear’s complex biology, we have much yet to learn. Over the next decade, efforts like the Temporal Bone Registry at Mass. Eye and Ear, which collects temporal bones from deceased donors for sophisticated cellular and molecular studies, will help us push the frontiers of knowledge even further.

We’re also going to improve how we apply that scientific understanding to our patients. We’ll no longer divide hearing loss into just two gross categories, conductive or sensorineural.

Instead, we’ll be able to dig even deeper into the root biological causes and pinpoint the origins of patients’ hearing loss. And we’ll have powerful treatments targeted at those underlying causes. For the millions of people worldwide who suffer from age-related hearing loss, that means therapies that can shield vulnerable cells and tissues as well as regenerative therapies that can regrow the tiny hair-cells and synaptic connections required for healthy hearing.

Importantly, we’ll also have the capabilities to address a troubling and widespread condition, known as tinnitus or ringing in the ears. Over 50 million people in the US alone suffer from it, yet there are no medical treatments. While the majority of patients learn to cope with their tinnitus, some 5 percent are so deeply affected that it rules their life. For those people and the many others who suffer from hearing loss, we can — and will — do better.

Related Articles

post 01_מאמר

פיתוח תוכנה למערכות רפואיות ובריאות דיגיטלית

post 01_מאמר

תהליך פיתוח תוכנה למכשור רפואי – דגשים בתהליך הפיתוח והתייחסות לרגולציה ובקרת איכות

post 02_מאמר

התקנות החדשות ברגולציה לתחום המכשור הרפואי מבוסס בינה מלאכותית לשנת 2024




מה צפוי לקרות בעולם המכשור הרפואי ב-2024?


ניהול סיכונים כתנאי מקדים לשימוש בכלי AI ו-GAI בתהליכי פיתוח מכשור רפואי ותוכנה רפואית