Naji Gehchan: Hello, leaders of the world. Welcome to spread love in organizations, the podcast for purpose-driven healthcare leaders, striving to make life better around the world by leading their teams with genuine care, servant leadership, and love.
I am Naji, your host for this episode joined today by Dimitris Bertsimas Professor of Management, Operations Research, and Associate Dean for the Master of Business Analytics at MIT.
A faculty member since 1988, Dimitris’ research include optimization, stochastic systems, machine learning, and their application in different sectors including healthcare. Dimitris is also a serial entrepreneur, he cofounded several companies like Dynamic Ideas sold to American Express, Benefits Science, a company that designs health care plans for companies, and MyA health, a personalized health care advice company.
Dimitris has coauthored more than 200 scientific papers and several books. He has received numerous research awards, including the William Pierskalla best paper award in health care.
His work in healthcare through analytics contributed to improve many patients’ lives, and still more to come!
Dimitris – I am humbled to have you with me today!
Dimitris Bertsimas: It’s a pleasure to be with you.
Naji Gehchan: First, I would love to hear your personal story from electrical engineering, to math, computer science and analytics, and now being a serial entrepreneur and MIT professor what’s behind this impressive journey of yours.
Dimitris Bertsimas: Well, um, I was born in Greece and came to finish the computer science department there graduated in 95 and then came to the U S at MIT in that year of 1995.
And I have been at MIT ever since I finished my PhD in eight in enjoying the faculty that. Uh, so th there is, um, but back in my childhood, I mean, uh, there is, uh, my, both, my parents installed in me the idea to, to Excel in whatever one does. Um, so with these preconditions, um, we see something I try to do, I try to do as good a job as I can.
Um, then, uh, when I find myself in. In the, um, in the Boston area and at MIT, which is, uh, an environment that, um, sort of excellence is a given, uh, I thought, uh, and inspired by the, by the environment. I, I thought that it might be. Um, a reasonable idea to apply what I, I study and I do research on in the real world because in the end of the day, I, I believe then, and now that, um, the best theories are those that solve the problems that originated with theories and other results.
I have been involved in the beginning of the nineties in. In this early in the financial sector of Boston, Boston has a very strong one of the strongest in the world asset management area. But then later in healthcare, healthcare in Boston is very strong. Some of the best hospitals in the world are here.
So the combination, therefore of my predisposition to effect, to have impact, to matter, to affect the world, the environment at MIT that, um, uh, motivates people. Uh, for all ages to do as good a job as they can, and the opportunities that are available to somebody like myself, um, and, uh, who has both the mighty connection, deep connection, and it has an entrepreneurial spirit, um, to achieve these objectives.
Naji Gehchan: Thank you for sharing that you have a passion for healthcare, and you mentioned it’s here, um, as your work contributed 3d to change people’s lives on many of the research and the work that you’ve done. Why healthcare? Why is this passion specifically for this sector?
Dimitris of yours.: Um, at some point, um, I definitely have a predisposition to have impact, but, um, and in the beginning of the 2000, there was an opportunity with a close friend of mine to start a company called who also in healthcare. Uh, and I started, we started the company. Oh, my, he was doing well at decisive point was also that my parents also got sick.
My, I lost my father to cancer in 2009. She got sick in 2007 in the same year, 2009. I also lost my mother. So this is a period of roughly middle age. When you, you think about your. And, um, and you observe that, um, what is important. Um, and I thought, um, affecting the lives of people in a positive way is more important than, uh, perhaps making money and making rich people richer that I was doing when I was.
In an earlier life when I was working in finance and services. So I, I’m not that I have any negative opinion about these matters, but I definitely believe that, um, researching the and helping people to improve the lives, um, is a value is a worthwhile effort. And I have also observed that with the evolution.
All for analytics, namely the data in electronic form in healthcare, the opportunities are significantly higher than they were in the past. So, um, that’s some of my, um, aspirations.
Naji Gehchan: Yeah, and we definitely share this similar purpose of making life better. And in healthcare you would frequently, I think you met with people who share this purpose, uh, deeply, uh, you, you start talking about, uh, data, uh, AI and what this can bring to healthcare.
So there’s obviously a huge hype on, on this, on big data AI and how this will disrupt the healthcare industry. You know, those trendy words these days, you’re obviously an expert there. I would love to hear your thoughts about
Dimitris Bertsimas: it. Um, so let’s take an analogy. Um, in, um, I was involved, um, in financial services in the late eighties, early nineties using quantitative approaches, analytics, beta, and so forth.
And, uh, it can have a significant impact in the field, um, in healthcare. The availability of data and electronic form of, uh, both fractured and unstructured, electronic medical records, uh, computer vision scans, uh, language is starting to become increasingly available. I would say in the last decade. Uh, so if you think about human doctors and how they, they, uh, reach decisions, uh, diagnosis and, uh, and prescriptions regarding their patients, they definitely involve data of this type electronic records test.
They do scans, they read reports that they read. Uh, recently genomic information. So it makes reasonable sense to me anyway, to utilize the same type of data that human doctors do to, uh, to make these predictions with a difference that computers, unlike. Typically don’t get tired. They are, they have less bias.
Maybe they might be biased on the developers, but, uh, of the algorithms, but, um, but nothing else. So it makes sense, you know, rationally to consider. The use of methods that have been unusually successful in other areas. I mentioned finance services, but you know, energy production, if you think about energy production in the world regarding why we have electricity today, and it works quite well, and I leave the keys behind it and we use of data and so forth.
So it’s makes sense that, um, given the availability, um, to, um, to attempt to do that, Um, and of course, many people realize. And when people try that, there is a complexity that unlike other areas, we are dealing with humans, both human doctors and human patients. So therefore other elements besides data, mother, understanding what drives them, understanding the culture, how they react to compliance that’s um, that makes it as what are some more active.
Um, in addition to being potentially. Can we double
Naji Gehchan: click on this because I love how you frame it, human doctors. Uh, I dunno if it’s compared to computer
Dimitris Bertsimas: doctors that something, that’s my that’s what I expect in the future might be at least in assisting doctors. Um, so, okay.
Naji Gehchan: Can we talk about this? Because many times we see innovation, some would see innovation as, you know, disruption, or like the end of.
Certain type of jobs. So you shared the complexity of it. Like, what is your vision about it? Shouldn’t be afraid as healthcare professionals, uh, from the technology or embrace it. What is your
Dimitris: view on this? We have seen disruption innovation in many industries over the centuries then. So it’s definitely the case that the type of jobs change, but humans are not replaced.
Take for example. Doctors, the medical education in the world has not changed primarily since 1920s. When John Hopkins introduced the care and structure of training doctors, the appendix, the append, this type where data doesn’t play a significant role. Um, this is, uh, I think in my view, this is about time to change because it makes good sense, given that the technology has now advanced, it might be not yet at the level that we can trust it a hundred percent, but it’s definitely has improved dramatically to train doctors in this way.
Therefore, it’s not that we’re not going to have human doctors. Of course they were going to have, but they are going to be in my office. Uh, the doctors of the future will be drained, trained differently. They are trained today in addition to anatomy, in addition to, um, do what they learn at the moment. And the experiential aspect is very important.
I think understanding. Uh, data and their exploitation using machine learning AI, I think it will be part of the story and programs in digital medicine. I, I already observed them reluctantly being in some universities, but I believe it was. Main stream in the years to come. And if you train young people in this way, this will take a generation.
Don’t take me wrong. It’s not going to be simple, but I have little doubts that in 10, 15 years, uh, the experienced patients will have in visiting doctors. In assessing, um, what will happen to them from a healthcare perspective would be quite different from what is today. But I do not believe that this would replace humans.
It will just be an adaptation of what the doctors do as opposed to replacing them.
Naji Gehchan: Well, you shared a in the class, I had the opportunity and pleasure to be in, uh, in, in your classes. Uh, great examples on how data and AI really transformed, uh, some of the care for patients, uh, and things you worked on. I would love if you can share one or two of these examples to give tangible.
Dimitris Bertsimas: So, um, I have a long collaboration with Hartford hospital. Um, hospital in connected cars and have for healthcare in that we have implemented, um, and, and, uh, machine learning, AI approach that, um, predicts for every patient. The length of. The probability of mortality, the probability of, um, going to ICU, leaving the ICU.
So in other words, for every patient in the hospital, based on the data they have, based on who is updated regularly, because there’s new information, the hospital, we basically can, um, make predictions about. The future and why this is relevant. Suppose you observe a patient. As we have that. The, the mortality probability we used to be, let’s say 2%, 1%, two days ago is now 3% today, 5% tomorrow.
So even though these are still small numbers, In the scale of things, the fact that our increasing might reveal and condition that human doctors, it’s hard to, it’s hard to observe. And in fact, in this particular case, this particular person developed, uh, an infection that was slowly developing. And the fact that we’re able to observe this, the album picked up the increasing probability gave you opportunity to doctors to actually attend to this.
Uh, even though the algorithm in other words was not designed for this purpose, the fact that you can use it in this way helped the outcomes. That’s one example. Here’s another example. I have been involved for a decade now in a company called benefit science that you mentioned its objective is to design healthcare policies for large organizations in the United States, but also around the world.
Healthcare is primarily the responsibility of the employer. So the employer typically provides the funding provides healthcare and they typically self in soar in that, but they also have to decide what type of plan. So rather than basing, only on demographics benefit science looks at actual data to design.
What is the best quality of policy of a policy? What combination of health savings account, um, PPO plans, HMO plans. To maximize quality subject to a budget. And we have found that, um, the companies, um, save money, but the quality also increases. Another example is another company that you mentioned, Maya health, my analytics health, that takes the perspective of not the company, but the patient.
So, so let’s say you are, um, you have a health savings account. So what is the best way from a to. Um, to monitor and optimize your health. For instance, if you want to do an MRI, what do you do the MRI? If you, because the price is very, very significantly, the quality of care at various places for various specialties also, uh, significantly changes.
Data provides you an objective view of reality. Data can allow you to, um, to basically take the bias out and make more objective and overall better. So this is an example from hospitals to two companies to patients. And it’s really endless. I could go on give you many other examples. Yeah. Thank you.
Naji Gehchan: Thank you for those.
You said data gives you an objective way to look at things I want to, I want to get into more leadership, uh, discussion. With data. And my first question on this would be when you shared the example, for example, uh, on, uh, on the healthcare, uh, hospital or the institution you worked with, uh, we always see reluctance from healthcare providers from experts.
Uh, whenever there is data saying something. And we don’t really believe or buy into, right? Like the model is wrong. No, it’s different. You know, we even see it in our industry is when we talk with different countries, like data shows something, but you always have, oh no, we are an exception. It’s different here.
How do you deal with this? Have you seen this and how do you deal with
Dimitris Bertsimas: it multiple times? There is skepticism that comes from culture, but also sometimes correctly. I mean, you deal with significant decisions about patients, life and death, this situation sometimes. So it’s appropriate to be skeptical. So in this particular experience with Hartford is I was fortunate to have met, um, two people.
Uh, there were four men, executive MBA students. I met them in the classroom. Uh, who have leadership positions in the organization? Um, Barry Stein and RJ Kumar, both of them are in the leadership of the Carrefour healthcare hospital system. And, um, I had the opportunity therefore, to be introduced to them and at least start in a relation of trust with these two gentlemen and over time, because I started working with people.
And then so forth. Um, there, the level of trust increased dramatically, a particular important moment is that I gave a class to about, it was just before COVID. It was 2000, January, 2020, just before March. Uh, and I gave a lecture, a set of lectures, um, to about a hundred professionals about the art of the.
We are the other possible in healthcare and this educated many people at the senior level physicians, nurses, administrative personnel, um, about what can be achieved using data and analytics and this open the door. Even for example, we, we developed a model for COVID, uh, for trying to help, uh, hospitals.
Size their ICU needs given that you don’t know how the, how the disease will develop. So the fact that they were already seen the benefits and the realities of analytics. Mainly leaders in the hospital to at least approach it with more trust. Of course you have to verify, but, but the door was open and therefore the method had an impact in the hospital.
It allowed them. Two sides. The ICU is not only my main hospital, but the eight hospitals of the, of the system, uh, and the rest was relevant. And the trust is not only at the leadership level. Healthcare is local, no matter what the CEO of the hospital says in the end, the decision maker is the nurse and the doctor who, who are, who are in, um, Near the side of the patient.
And if they go and that’s what you’re going to say, it doesn’t matter who supports. So my experience therefore, is that, um, the, the literacy aspect allows. The missile contact, but then we, we, the way we developed all these methods is that even today on a weekly basis with various groups, with the patient, the doctors that attend to the patients, the doctors that attend, um, surgeries, uh, surgeons, nurses, and so forth, we have weekly.
So as a result, when somebody asks, I don’t explain, they explain and it’s much more effective if your colleague who is, of course you have a decades relations whom you trust say something, it has a different gravitas. And, um, I have found that, uh, Understand, you know, understanding the culture of the environment and gaining trust, not gaining trust by, by basically demonstrating to people that we, first of all, you, you trust their opinion.
For example, if the algorithms benefited from their comments and if this, if this happens, it’s not anymore, the album I developed is the algorithm that we have. And that’s how it is presented. So I would say this is a combination of leadership from the top, but also literacy from the base. Uh, and in some cases, one is much more effective than.
Naji Gehchan: Yeah, that’s that’s great. Um, you, uh, well, we had Barry in one of our episode, various time for the listeners who want to know more about his story, you shared Demetrius about, uh, all this power of data and really how they impacted lives and the most recent example with COVID, as you said, Um, there’s all this informed.
Decision-making also that as leaders we can do now, even better with all the data that we can, we can process. How do you see leadership? In fact evolving? Based on those data analytics algorithm that we can use. Do you see it changing or will it change? And what is the role of leadership overall and this a
Dimitris Bertsimas: massive word of that to set an example from yes.
So yesterday, one of the largest, um, Italian companies contacted me with the idea to they, they decided to increase, uh, the digital, uh, aspect in their company. So the use of data throughout. Right. I mean, they have no groups, but they have this desire to do it. So they are asking me very detailed questions about my experience on that.
So here’s a company that is not really not. It is not, I mean, yet they are thinking about, and this is of course the story of many other companies. It’s not that this is an exception. Um, I expect it has already been happening. It’s not even if this is not even a prediction, I’m probably stating a fact. Um, it is already today, but I think it’s going to increase in the future that I’m a chief digital officer.
It’s if analytics officer, there are multiple names and so forth, we’ll, we’ll be at all that many companies already have, and we will be fulfilling in the future. That would be the accurate reporting to the. Which means if, you know, typically a CEO of a large company has 10, 10 direct reports, um, you know, 10, 15, no more than that.
And I believe that one of them would be that that officer, it already is happening in many companies. I know. So it is my. And this is also by the way, through the hospital systems that I collaborate people that have responsibility on, on data and analytics report to the hierarchy of a company, the highest echelon of the company.
So it is my view that this is already happening and it’s going to increase as a result. Another instance of that. Is this access that the master of business analytics program that I started 2016 has had over the years. So currently the program has of, uh, maybe this year 1600 applications for 80 positions.
That’s a 5% selectivity. There aren’t too many places at MIT that have that level of applica application. And I would expect that in years to come, um, the demand for MBA. We’ll be lower than the demand for people with that. This experience has already happened in many places, not at MIT yet. So currently we applications for MBAs, roughly double, maybe 3000, 3,500 versus, um, 1600.
But, but I think this gives will meet as they have met before, because there is a high. I observe and I don’t observe it now. I’ve been observing it for decades. That’s why we started the analytics program. Uh, I have been observing it and in healthcare, I would say definitely I see the need. I see, uh, places, some of the major hospitals having one or more analytics group.
So it’s definitely, this revolution has started.
Naji Gehchan: I know, not to jump into a section where I will give you one word and I want your reaction, a word or a reaction to what happened to the word that would mention. Uh, so the first word is leadership.
Dimitris: Commitment is the word that comes to mind, um, and understanding.
Naji Gehchan: Can you tell us a little bit more, I’d love to hear a definition
Dimitris: from you. Yeah. Um, I believe that, um, the way I see it is that I better like the work on analytics, which is the science of using data to build models. That make that lead to decisions that impact the world positively data models, decisions, value, uh, AI is nearly a synonym in that it also uses, um, data sources, not traditionally utilized.
For example, computer vision. Um, languages, language, you know, answer actual data, but, um, the process, which I believe is our future, our collective future, using data to make decisions that impact the world.
Naji Gehchan: What about personalized medicine?
Dimitris: Personalized medicine is, is, um, also, uh, a very bright, an aspiring future.
If you can, only medicine is by and large, not personalized. This is what, one of the major reason in my mind that we haven’t been in cancer yet. Um, so if you look at how, um, one of these major killers in the world cancer is, uh, I lost my father to that is being treated it’s more or less a size 50. I had this, you, you, you diagnose somebody with cancer.
You give them some protocols that has been, uh, have been developed for, uh, for everybody, not the personal, uh, human, a significant human. We observed very different outcomes. In my personal case around the time that my father was diagnosed with, with cancer, gastric cancer, there was a lady friend of my father who was diagnosed with exactly the same disease and, um, sees a, to.
Um, and see what’s the same way, but it worked for her. It didn’t work for my father. So it is if there’s overwhelming evidence that, um, personalization mothers, I mean, we already see, I mean, we observe that some women, for example, have mutations that lead to development of breast cancer very early in their life.
Uh, at least in this case, we have taken action, but it’s hard for me to believe that cancer is not one disease. It’s multiple diseases who have an composite in one, in one name, and then they meet personally. The attention. This is not only about cancer. I mean, you know, think about, um, think about impact of COVID.
There are many people have COVID very few die, but, but, uh, clearly there’s personalization aspects. I mean, humans are different. They have different genomes. They have different. Even cultural components. They have many different things, diabetes. I, who do I start? I mean, you know, I, I think medicine, when I talk about this is all medicine and personalized medicine.
Uh, these are medicine can lead to personal life.
Naji Gehchan: And we’ve been, we’ve been hearing a lot about personalization in medicine, but yet it’s, we’re seeing a little bit more personalized care. Uh, but what do you feel is in the way, is it truly time technology and time that will get us there? Or do you feel like there’s something else?
Dimitris: I mean, um, the key aspect is education. Thinking about how we educate our doctors. We have not changed the education of our doctors since the 1920s. It’s a century ago. If you look what people learn, um, then, and how they learn it, you know, Does not play a significant only vis education. So when you educate young people, very talented, the doctors are fantastically talented people.
I do not believe they are appropriately educated at the moment. So as a result, you have to educate them in the art of the possible of personalized media. We have some successes. If you, if you educate a large collection of intelligent people, um, so that their mind goes into that, they start doing research, they start doing developing new methods.
Personalized medicine will be a reality, but you have to start. And I would say the key is, in my opinion is education and it’s starting, but you know, it takes as everything in life. It’s not. It takes some time
Naji Gehchan: and it’s, but it’s definitely a future we should aspire for. The last word is spread love and organization.
Dimitris: And organizations you would like you ask, you want to ask do as my reaction, you know, I, you know, in my view, uh, love is this most significant aspect of human happiness. Um, so specifically I say to my, uh, to my students that the two most important aspects of life is to find somebody to. And to find something that you love to do.
So love is in there in the definition of happiness. So spread love is in a way, um, aspires the way I understand it to increase the overall happiness of receivers. As well as givers of love and my, uh, I mean, in a way, one of the reasons I, um, I mean, there’s no security. I love working with my doctoral students.
I have a sizeable group of very young, very, very talented young people. Um, and the big aspect of our relation, at least on my end is luck. So, um, And as I mentioned, um, it increases my happiness increases. There’s
Naji Gehchan: any final words of wisdom? Uh, Demetrius for healthcare either is around the word.
Dimitris Bertsimas: The only, uh, the most significant in my opinion is, um, to, to make change. You have to have. Um, sometimes you enter an unknown, you know, utilizing data. You never know what you’re going to find. You might actually find that one or one of your departments using data.
It’s not doing a good job, but on the other hand, if you don’t know it, it still does it not a good job, but you don’t know it. But if you allow open your mind, And, uh, allow data to enter over time. Fantastic things will happen to patients, which is our ultimate objective as, as health professionals, but also to ourselves, to the hospital, to ourselves.
I think our life would, um, would be, I would say in a higher plane.
Naji Gehchan: Thank you so much. Uh, the interest for being with me today and this amazing.
Dimitris Bertsimas: Thank you. Nice to be with you.
Naji Gehchan: Thank you all for listening to Spread Love in Organizations podcast. Drop us a review on your preferred podcast platform
Follow us on LinkedIn and connect with us on spreadloveio.com. We’re eager to hear your thoughts and feedback. Most importantly, spread love in your organizations and spread the word around you to inspire others and amplify this movement, our world so desperately needs.