Harlan Krumholz, MD, SM accepted an invitation to be the inaugural guest speaker at the Department of Statistics and Data Science’s virtual Commencement 2021 celebration.
The esteemed Dr. Krumholz delivered a wonderful and inspiring speech to attendees on Sunday, May 23, 2021. Please read:
Congrats to all graduates and your families!
What a pleasure to be with you, even virtually, on such a fantastic day.
You have chosen well by selecting Data Science and Statistics as a major. We are in just the beginning of an age of data science, and this age will bring on major changes in almost every facet of life on the planet. In many ways, it already has. And yet, we are still firmly at the beginning of the digital and data science age to come.
However exciting the possibilities, it is important to remember that data science is a means, not an end. It can help us see what was formerly invisible. It can produce capability and insight and understanding — and, for many, it can confer power. This power will have profound implications for people and nations.
Your lifetimes — your careers — your accomplishments — will be importantly intertwined with how that power is used — and what comes of it. The end result could elevate society. It could also do otherwise.
In a way, I liken where we are today to where we were when physicists began to envision the splitting of the atom and nuclear chain reactions. George Kennan, the eminent American diplomat, and historian, in his eulogy of J. Robert Oppenheimer, the brilliant physicist who led efforts to build the atomic bomb, talked about ‘‘the recent conquest of a power over nature out of all proportion to [our] moral strength.”
The physicists of that age found themselves not just among equations but also deep into moral quandaries. They were extraordinary; but their actions unleashed the power to create and destroy.
I look around, and I see data science at a similar juncture. For all its promise for good, there are those who are using it in ways that divide us, weaken us, and undermine us.
Data science does not have a natural valence… its applications do. And to date, to paraphrase Keenan, our power over data sometimes seems out of all proportion to our moral strength as a civilization.
Data science can empower individuals; it can limit individual freedom. It can promote participatory democracy; it can be a tool of autocracy. It can make the world much better; it can make the world much worse.
The same power to do good has the potential for evil.
Yet, I am someone who believes deeply in the promise of data science and statistics to improve society — and not just to make society and our transactions more efficient — but to promote health, freedom, equity, community, climate, and peace. Data science and statistics should help strengthen society — you are ideally positioned for that. And some of that optimism is because I believe in you graduates — and the potential of your generation.
Already, in so many fields, we have seen the beneficial fruits of data science. All aspects of our lives have been transformed by data science applications. From the humanities to the social sciences to the life sciences and the physical sciences — breakthroughs are predominantly occurring because of data science. Data are more abundant than ever — and the computing power — and analytic advances — are creating an exceptionally fertile moment for the application of data science. Even this virtual graduation digital platform — the fact that we can get together like this — is a product of data science.
In my field, medicine, we are rapidly transforming into data science. From discovery in the life sciences to clinical care to population health, we embrace the possibilities of a future that better delivers to each person and each community to the possibility of achieving their health goals — and do so in a way that is equitable, affordable, and accessible.
What is the traditional approach? People get better with experience. Doctors accumulate knowledge over time from their lived experience. If you want wisdom, you need to go to someone who has seen a lot of a certain thing. But in the past the learnings from any individual patients have tended to be sequestered — trapped in the memories of those in contact with the patient. Papers are written on studies — and occasionally they focus on case series of patients — but the vast majority of our experience is siloed and inaccessible to others. And even among those in contact with patients, they only have a partial view.
With data science, we can get to a day where we get smarter from every interaction and produce real-time learning and decision support. It already happens in other fields… Google gets smarter with every search… it doesn’t launch a traditional research project to determine whether certain people prefer particular search outputs…it creates knowledge pipelines that not only learn rapidly from experience but put that knowledge into immediate use — and they throw in some experiments — A/B testing — along the way. Imagine in medicine if we could learn from each person in this way — someone across the country or the world has confronted a similar health challenge — and the next person can benefit from what was learned.
With data science, we can release that learning from the small number of people who interacted with that patient — and harness it, along with the experience of hundreds and thousands of others in similar situations, to help the next person. In a digital world it is possible to learn rapidly and see those learnings available immediately for the next patients. We can create the possibility that no matter where you live, you can benefit from the best knowledge available — and up until the prior patient. Just as your search on Google benefits from what was learned from all the prior searches… and Amazon’s services from all the prior clicks… that can happen in medicine too.
This is the ultimate pay it forward. Your experience helps the person a mile behind you… you were helped by the person a step ahead. And it is data science that is the key to it.
Data science can improve surgical procedures through augmented reality and smart applications at the interface of data science and engineering. Surgeons may not even need to be in the room — and the devices may ensure top performance. For air travel we have embraced all kinds of AI and technology to ensure the safety of planes and to elevate the performance of pilots. We do not tolerate errors and work hard to make it difficult or impossible to make a mistake. That approach will also be what we will do in medicine.
Data science is transforming diagnostic pattern recognition. A lot of medicine is about recognizing patterns, including anything from reading a chest-x-ray, interpreting an electrocardiogram, and classifying a heart murmur. But our brains have limitations — as do our senses — so what we can see and hear — how we can reason — is limited by our individual capabilities. As such there is immense variation in how well doctors recognize patterns — and even the performance of an individual doctor may vary by the time of day, day of week and more. With data science we can develop tools that can out-perform doctors on pattern recognition and free them to focus on other aspects of care. We can decrease error rates and even automate and empower patients with information that that can be generated from home.
These techniques will also show us patterns to help us see what we could not before. Our current labels for disease will be superseded by new classification systems that are much more precise and personalized. For example, today many people talk about two types of diabetes, but there are likely hundreds of types — and soon we will be able to classify people better — dispense with labels that often are 100 years old — and have long lost the ability to convey meaning.
Another trend made possible by data science is the changing venue of care. Hospitals will become the place for only the sickest patients as we will use technology and data science to care for people at home. We will use remote monitoring to anticipate health problems and pre-empt catastrophic events. The health care will become more convenient, more responsive, and more data rich… with concomitant analytics to enhance people’s experience and promote their health.
Health care data is sensitive and fragmented. There is a need to find ways for people to gain greater agency over their data, to have it primed for them to use for research and their clinical care. Ultimately your health data assets should be yours to use as you choose. And your privacy needs to be protected — in ways that are uncommon now in the rest of the digital economy. This will all take data science too.
I deeply believe that health and health care in the future will be transformed for the better through data science… data science that preserves the humanity and promotes the performance of people in health care … and those seeking better health.
But the same data science capabilities that can produce the good — can create mischief. We are entering an age where data science may dictate who is hired, who gets a loan, or even how long someone spends in prison. If these algorithms are biased, then people will be treated unfairly. We can inadvertently strengthen and embed further the institutional and systemic racism and bias that infects our society, even amplifying it beyond where it is now.
Joy Buolamwini, founder of the Algorithmic Justice League and a researcher at the MIT Media Lab, is an example of a force for greater fairness, accountability, and transparency in coded systems. She talks about algorithmic bias, which is just one of the potential dangers of data science, as a way to spread exclusionary experiences and discriminatory practices on a massive scale. Actually, she likens it, in a way, to a pandemic as these algorithms are capable of spreading rapidly and causing widespread harm to individuals.
Our own Elisa Celis is also providing leadership in this area. It is amazing — and important — that this major has a course on Data Science Ethics. She says ‘Fairness and inclusion can be written into algorithms, right alongside speed and efficiency.’
So as you go out into the world, you need to know that you can indeed make a difference.
In 1914, HG Wells published a science fiction novel, The World Set Free, that envisioned an atomic bomb for maybe the first time. In the book he juxtaposes the potential for good and evil more than 30 years before there was the reality of the bomb. One of Wells’ fictional atomic bomb inventors describes a sense of inevitability about what will transpire, perhaps avoiding any sense of culpability for the consequences. Wells writes: ‘It is not for me to reach out to consequences I cannot foresee. I am a part, not a whole; I am a little instrument in the armory of change. If I were to burn all these papers, within a score of years have passed, some other person would be doing this…”
There is no particular future that is inevitable — you should never see yourself as just a part of something that is you cannot influence — you have a chance to define the future.
You, the next generation, need to guide us through your equations and your moral strength. You need to position us to manage this power — and to see its application for good.
What makes me proud about Yale… it is, in part, reflected by what is said at graduation. The President of Yale confers the degrees and admits new graduates to each degree’s “rights and responsibilities.”
You are admitted to the rights and responsibilities of data science — and I am encouraging you to seize that responsibility — and carry it with you throughout your lives. The responsibilities of this discipline — data science — may be more profound than many others, with society’s fate in the balance. That is not just a burden, it is an opportunity to play a key role and care deeply about the effects of your field. It is to know that your voice will matter, your actions will matter. Your education has placed you at the center of one of the greatest of transformational moments. What you do will influence where this will take us.
In closing, I would like to leave you with some concise advice (could this be a graduation speech without it).
Be purposeful with your time. It goes fast… and it is your most valuable asset.
Be confident. With persistence, determination, and grit — you can change the world for the better.
Be engaged. Recognize that you have power… there was a song when I was young that was part of a dialogue between someone indifferent and someone engaged… and the question was posed… ‘will you try to change things with the power that you have, the power of a million new ideas’ … the indifferent one says that he didn’t think it was necessary because everything seemed ‘so fine.’ Well, here is my request… if you remember anything from this graduation speech… realize that things are not so fine for everyone and that the future will be better than the past is not predetermined. Try to change things for the better with the power that you have, the power of a million new ideas.
So the world awaits you … dazzle us with your brilliance, make a difference that helps others, and engage in the efforts to bend the power of this field toward the good it can do.
Best to all on a wonderful day!
~Harlan Krumholz, MD, SM, Harold H. Hines, Jr. Professor of Medicine (Cardiology) and Professor in the Institute for Social and Policy Studies, of Investigative Medicine and of Public Health (Health Policy); Director, Center for Outcomes Research and Evaluation (CORE)