AI is Needed Where Humans Struggle with Nonlinear Processes - i.e. 99% of the Time
I was interviewed by the brilliant Shankar Hemmady. We talked about the future of AI and technology in healthcare and the concrete steps for transforming the industry.
Welcome to AI Health Uncut, a brutally honest newsletter on AI, innovation, and the state of the healthcare market. If you’d like to sign up to receive issues over email, you can do so here.
In a recent interview with Shankar Hemmady, a Silicon Valley strategist and Stanford alum, I discuss the future of AI and technology in healthcare, my journey in digital health, and the concrete steps for transforming the industry. I want to thank Shankar for inviting me to the interview, for highlighting my journey, and, most importantly, for addressing how we can improve access to healthcare in America through technology and AI.
Shankar did a great job editing my pauses. Otherwise, the interview would have been three times longer. 😊 Still, for those who would like a summary, I provide it in this piece. As always, I’m open to any questions or critique you may have.
Notable highlights of the interview:
✅ Insights on the transformative power of AI in healthcare
✅ The importance of democratizing healthcare access
✅ Challenges and opportunities in the healthcare industry
Here is the interview outline:
1️⃣ Introduction
2️⃣ Early Life and Education
3️⃣ Move to the United States
4️⃣ Career in Math and Economics
5️⃣ Transition to Healthcare and AI: Advancing COVID-19 Research
6️⃣ Founding a Company in Digital Health, and Facing Challenges in Healthcare and AI
7️⃣ Balancing Innovation and Regulation in Healthcare AI
8️⃣ Transforming the Healthcare Industry Through Technology, Antitrust, and Deregulation
9️⃣ Challenges and Opportunities in Healthcare
🔟 The Need for AI in Healthcare
1️⃣1️⃣ Economic Realities of Healthcare Innovation
1️⃣2️⃣ Nonlinear Systems Are Why We Need AI in Healthcare
1️⃣3️⃣ Addressing Fragmentation in Medical Research
1️⃣4️⃣ Accessibility and Democratization of Healthcare
1️⃣5️⃣ Challenges of a Healthcare Startup
1️⃣6️⃣ The Importance of Democratizing Healthcare Access
Here is the shortened version of the interview transcript:
Sergei Polevikov shares his journey from his early life in Belarus to his current role in digital health and AI. He discusses how his father’s influence as a mathematician and physicist instilled a love for mathematics in him. Due to political instability in Belarus, Sergei moved to the United States to pursue academic opportunities, eventually earning a degree in economics and working in finance on Wall Street.
Sergei’s transition to healthcare and AI was almost accidental, prompted by a restructuring at his previous company. Together with a friend from finance, he co-founded a company focusing on bioinformatics and digital health. During the COVID-19 pandemic, they developed a tool to aid genomic researchers by mapping nonlinear relationships between medical terms, which proved useful in advancing COVID-19 research.
Sergei emphasizes the importance of freedom and innovation in both his personal journey and professional endeavors. He discusses the challenges in healthcare, such as the industry’s resistance to change, the need for transparency and best practices in AI, and the balance between innovation and regulation. He also highlights the importance of democratizing healthcare access and ensuring that new technologies benefit a broader population, not just the privileged few. Sergei remains optimistic about the future of digital health and AI, believing that innovative ideas will ultimately succeed and transform the healthcare industry for the better.
Interviewer Shankar Hemmady is founder and CEO of Blue Horizons, a San Francisco based company focused on the last-mile problem of Generative AI based solutions in two vertical B2B markets. Their team has significant expertise in Generative AI, AI-ML, Blockchain, 5G-6G, Cloud and Edge computing, 3D-IC, RISC-V, Cybersecurity and Functional Safety. Over the past 25 years, Shankar conceived, invested in, marketed and delivered several new products primarily in three verticals: electronic design automation, biotechnology and education. Most of them involved applying new hardware and software paradigms, especially AI.
1. Introduction
0:08
Shankar Hemmady (SH): Hi friends, welcome again to another episode of “Beyond the Clouds: Edge to Transformation”. This transformation is engulfing every part of our life, and definitely healthcare. How healthcare is changing is amazing to watch, and I learn a lot from Sergei. Sergei, you’ve been doing a lot of writing. I truly appreciate that. Can you tell us how you became the Sergei we know today?
2. Early Life and Education
0:24
Sergei Polevikov (SP): Thank you for having me, Shankar. It’s great to be here. I had great family beginnings back in my home country, Belarus. I learned really early how to survive. We didn’t have much, and also the value of education and science was instilled in me, mainly because of my father. He’s a mathematician and theoretical physicist. He’s the person who instilled the love for mathematics in me. I still really enjoy reading math papers and doing some calculations from time to time.
3. Move to the United States
1:01
SP: I came to the U.S. partly because of the political processes that were going on, and still are ongoing, in my home country, Belarus. There were some poor leadership and dictatorship that unfortunately were instilled in Belarus right after the breakup of the USSR. Unfortunately, those people are still in power. Back then, as a student, I was trying to make sure that my country is free and we have some good life. I was very active as a student, and unfortunately, the government, the authoritarian regime, took notice, and it became just very dangerous for my life. At the time, it was a pretty hard decision. I still had a lot of friends and family there. I decided to pursue something outside of the country, particularly in academic circles as a graduate student and then as a researcher. That’s how I came to the U.S. as a graduate student in economics. Once I got my math degree, I started looking around where I could apply my mathematical knowledge.
4. Career in Math and Economics
1:54
SP: I’ve read quite a few papers in economics and finance, and I found it very interesting how they use pure theoretical models to apply them to real-life economic systems. The main reason this whole journey started is the freedom. I feel like all of us humans need freedom. In the United States, when we have it from day one from our birth, we don’t appreciate it as much. When it’s somebody like myself or yourself who came as immigrants, we have great appreciation. We value all of the freedoms and liberty we have in this country.
2:44
SH: You remind me of the story of many immigrants. Whether it’s political freedom, economic freedom, or the freedom of thought to do different things. I truly appreciate that. How did you move from one field to another, from economics to now healthcare and AI? Can you tell us about your journey through that?
3:00
SP: Yes, it’s been quite a journey both professionally and on a personal level. To be honest with you, there have been some really great successful stories on the way up, but there were also some really hard situations and adversity. There were times when I was really thinking about what I was going to do next with my life, how I was going to represent myself as a person and as a professional. It wasn’t a smooth ride. The values of hard work and family values that were instilled in me eventually paid off. The education that I got, especially in mathematics, statistics, and later in data science, was very fortunate.
4:18
SP: I wrote a couple of academic papers in finance, but I also continued working with my dad, a mathematician and theoretical physicist, in particular in the field of thermodynamics. I was fascinated by how elegant some of those physics laws are. You can just look at the formulas and derive a law that represents the motion of a tiny object in space. That’s completely opposite from data science, where you need a lot of data. Some of the laws have already been proven, and you just reap the fruit of that labor. There are some areas that are really interesting from a theoretical standpoint. I almost did a full loop from complete theoretical work to now being in data science. Again, it started in economics and finance, where interesting mathematical models were applied.
I almost completed a full loop from theoretical work to data science. It all started in economics. At that time, there were many interesting mathematical models applied in economics and finance. We had Nobel Prize winners, such as Merton and Scholes, doing significant work in option pricing. I got a job on Wall Street in risk management, later moving to portfolio management and financial modeling. I dealt with the mathematics behind financial and insurance models.
5. Transition to Healthcare and AI: Advancing COVID-19 Research
SP: My journey to medicine was almost accidental. Four years ago, we had restructuring at a company I was working for, and I was looking for new opportunities. I got together with one of my friends, Daniel, who is now my current business partner. He was in finance but on the opposite side, producing fintech and risk management products. We got together, and he said, “I’ve done some work in bioinformatics. There are a lot of interesting areas where we could use your expertise as an AI expert and maybe build something.” Both of us are math guys, not necessarily experts in healthcare or any of the medical fields. We found some interesting connections and people to work with. During the COVID-19 pandemic, we developed a tool to aid genomic researchers by mapping nonlinear relationships between medical terms. This proved useful in advancing COVID-19 research.
Our journey shaped the work we do.
6. Founding a Company in Digital Health and Facing Challenges in Healthcare and AI
6:07
SH: You’ve been writing extensively about what’s likely to happen in digital health. Could you tell us briefly about what’s happening and what we can anticipate?
SP: That’s a question I’ve been asking myself for the past couple of years. To add to my previous story, as I started working in digital health and artificial intelligence by co-founding WellAI, I got to know a lot of people on both the clinical and academic sides of healthcare. There’s a lot of great research going on. I met some people in genomics research, and they invited me to join a working group during the middle of COVID-19.
Some of the methodologies and algorithms we developed at my current company, WellAI, were very attractive at a time when fast answers were needed. I got to know many great researchers, and we founded this working group in AI and genomic diagnostics as part of the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC).
A lot of big philosophical questions came out of being part of that group. How do you take all these rapidly developing methods and apply them responsibly? We have to be cautious and understand how these models operate. As part of the working group, I started focusing on areas of transparency, explainability, and best practices in AI. I even published a paper on it. Some of the questions you’re asking have direct answers in that paper.
Side note (SP): Our work at IFCC’s Artificial Intelligence and Genomic Diagnostics (WG-AIGD) culminated in a unique research paper “Artificial Intelligence-Powered Search Tools and Resources in the Fight Against COVID-19” that was peer-reviewed in record time, given the urgency and life-or-death reality of the pandemic. The paper was published on June 2, 2020, and the AI research tool we developed has hopefully helped medical researchers around the world fight the COVID-19 pandemic more efficiently and more intelligently.
7. Balancing Innovation and Regulation in Healthcare AI
8:30
SP: As part of the IFCC working group, I started working on areas of transparency, explainability, and best practices in AI. I published a paper on that. Some of the questions you’re asking have direct answers in that paper.
There has to be a balance. We need to be very careful with regulations. For example, I welcome what’s happening in the European Union, but at the same time, I feel that the freedom of innovation must be addressed. We need guardrails, especially in medicine, but innovation—particularly in AI and digital health—is what’s going to shape the future of healthcare.
People who are cautious and asking questions are asking the right questions. In the end, it’s technology that’s going to move healthcare forward. I see that every day. On a daily basis, I talk to many primary care physicians and specialists. It’s amazing that people still use faxes, and we have electronic health record systems from the early age of DOS.
We’re developing great technologies, but we’re facing outdated platforms that people are currently working with. We’re trying to connect the two. We need to continue developing best practices, particularly in digital health, due to the rapid advancement in AI.
At the same time, I want to be clear that especially for small company startups, we don’t want to kill innovation. Innovation is the key in digital health.
8. Transforming the Healthcare Industry Through Technology, Antitrust, and Deregulation
8:58
SH: Very true. New technologies are quite easy to use. Is it possible that [in healthcare] they can leapfrog and upgrade themselves to what we have today?
SP: Yes, but it could also take time. Let’s be honest, especially in healthcare and digital health, no matter who you are, it’s hard to make predictions, especially in an industry that’s been so full of inertia and monopolistic power.
The fastest way to change the industry is to do something drastic in terms of its structure. Historically, look at what happened in other industries. In the 1980s, the U.S. had software monopolies. In the 1990s, the Department of Justice broke up the Microsoft browser into two separate businesses, opening up competition. I believe companies like Google were born because of that proactive stance by regulators.
10:54
In healthcare, you’re looking at the FTC and the Department of Justice. They’ve been doing great things, but at some point, you need to sit down with stakeholders, regulators, CEOs, and definitely physicians. They should be part of the conversation to think about what should be done to benefit the patient. In the end, all of us have to take a hard look at how to change the industry and perhaps even deregulate it, as has happened in other industries. Once that monopolistic power is shaken, you see so many great developments and adoptions.
I see resistance from big players and incumbents in the industry. Many don’t want innovation, even when it could become part of their business. It almost seems irrational to me. Certain companies and people are used to doing things a certain way and like the status quo in healthcare, even to the detriment of the patient. There has to be a conversation, even with organizations like the American Medical Association.
Data scientists and researchers in general should be part of that conversation as well because technology is the future. I wish this happens in my lifetime, that we have innovation quickly implemented in our daily lives. We have to continue working. This is what we’re doing on a daily basis at WellAI, and with all the organizations I’ve been involved with. Everything we do, all the innovations, has to be for the good of the patient and the whole ecosystem around the patient—providers, nurses, receptionists, everyone.
9. Challenges and Opportunities in Healthcare
11:28
SH: Do you think the slowdown in the industry is primarily because of inertia and vested interests, or is it risk averseness because they don’t want to harm patients? What is really going on, in your opinion?
11:59
SP: It’s a very unique industry. First of all, you don’t want to hurt the patient. All physicians take an oath, and all of us who are not physicians should also think this way. We should prevent harm, which is very different from areas like finance. In finance, you make a mistake, you lose some money, and life goes on. In healthcare, caution is a factor that could be preventing new advancements from proliferating. It’s also the culture of the industry. People are used to doing things in a certain way. I talk to physicians and nurses almost daily. They are proud and respected by their patients. However, many physicians and nurses understand that technology is the future and are open to working with us.
10. The Need for AI in Healthcare
12:52
SP: Even the most brilliant physician cannot absorb all the information we have in the world. It’s impossible to read all the new medical studies. We have around a thousand new articles added to our database every day! There’s no way any human can read all of that. There are so many areas, like drug development and drug discovery, where AI can help. Anything with inefficiencies or nonlinear processes that are hard for humans to comprehend is exactly where AI is needed. People are concerned about safety in healthcare AI. Then there’s the fact that people are used to working a certain way, and the business aspect. Many innovations need financing. You can have an innovation that just sits on the shelf until you have partnerships with physicians and investors.
Side note (SP): However, we also need to ensure that the end users—patients and physicians—benefit from the innovation. Unfortunately, we are currently at a low point in the digital health cycle. In many cases, investors reap the benefits of innovation at the expense of doctors and patients, who are left behind. This must change. We can do better as an industry.
11. Economic Realities of Healthcare Innovation
13:56
SP: Healthcare is an area where it’s not easy to have a high-margin business like in finance and technology. This is why it’s hard to advance AI and other technologies in healthcare. Areas like drug discovery and genetics have a lot of promise and financial support, but other areas, like primary care, are challenging. You need to decide if you’re in it to make money or to help people. If you’re fortunate, you can do both. But in many areas of healthcare, you cannot do both. Maybe down the road, you can develop something scalable that helps both investors and patients. But right now, it’s a challenge.
12. Nonlinear Systems Are Why We Need AI in Healthcare
14:50
SH: Sergei, you made a very good point about nonlinear processes in healthcare where AI can help. Can you give examples of that?
14:57
SP: When I co-founded WellAI with Daniel, our basis was scientific. We’re both researchers and mathematicians. We started with a scientific model that used advanced AI methodologies. A few months later, the COVID-19 pandemic started. We had a great model and dataset we’ve been working on for over a year, creating a unique dataset of medical terms. We have one of the most extensive ones. We started thinking about how we could help with COVID-19. Our engineers created a brand new product in two weeks. You open the app, choose your area of expertise, and we map every single medical term, creating a matrix of nonlinear relationships between them.
This allows medical researchers to see the most efficient ways to apply their specialized expertise, even if it seems distanced from COVID-19, to fight the pandemic. The reason we succeeded in this unique project, besides the hard work and sleepless nights, is the advancement of machine learning and AI.
We now have enough compute and GPUs, even at a small startup like ours, to map the entire complex National Library of Medicine into a massive, highly nonlinear, multi-dimensional matrix.
Side note (SP): In my forthcoming article on Explainable AI, I address the challenging task that any AI model in medicine must be explainable and transparent. If you agree with my argument that most things in medicine and in life are nonlinear, you will quickly appreciate the (almost) impossible task of making AI models easy to interpret, i.e., to somehow convert the nonlinear AI models to linear. 🤯 It’s an enormous challenge, and in that article, I will talk about the scientists and researchers who are diligently working on this problem.
13. Addressing Fragmentation in Medical Research
15:45
SP: We realized that most medical research scientists don’t talk to those in other areas. Genetic researchers focus on their own conferences and papers, and infectious disease researchers do the same. Our tool helps make those nonlinear connections, finding relationships between different medical terms. This is how the IFCC working group started. Medical scientists noticed our work and saw its potential to help develop tools for various clinical applications.
14. Accessibility and Democratization of Healthcare
16:30
SH: I find it interesting that on one hand, people say technology will make us immortal within ten years, but on the other hand, basic preventive medicine is not available to many people. What’s your take on that?
16:39
SP: It’s an excellent question. The U.S. is the foundation of many great innovations. In Silicon Valley, there are many developments, but access to these tools is easier for people with certain social status and income levels. Unfortunately, they’re not necessarily accessible to the general population. Over time, the cost of producing certain things comes down, making them more available. But in healthcare, that journey takes much longer.
How do you get the newest technology to average Americans, especially those in rural areas or with low income? There is no easy answer. We need to involve broader groups of people, including policymakers, physicians, venture capitalists, and scientists, to solve this problem.
15. Challenges of a Healthcare Startup
18:06
SP: When Daniel and I started WellAI, it was a personal and emotional journey. We were shocked by the many problems in healthcare. Our initial idea was to build a free direct-to-consumer app that provides medical information to the underprivileged. However, the daily life of a startup and the challenge of going direct to consumer made it difficult. Venture capitalists see it as risky, and there are many liabilities. Healthcare giants want to control their patients and the environment. There are many obstacles, including regulation, monopolies, and people dynamics.
16. The Importance of Democratizing Healthcare Access
20:52
SH: I’ve seen it myself, not just in my venture, but in talking to many other companies in health tech. There are numerous obstacles—regulation, monopoly, and a lot of it is people dynamics. To those listening out there, it’s time we pay attention. It is time to democratize healthcare access and hopefully healthcare research as well. There are real opportunities for curing and preventing diseases and living healthier, longer lives, but today, these are accessible only to the privileged few. I want to see these advancements available to a larger number of people so we can live healthier, happier lives in a way that's more harmonious with nature.
Thank you for bringing up this issue, Sergei. I’ve been reading a lot of things that you publish. To everyone out there, please participate in understanding the complexities of the healthcare system—how medicines are made and delivered—so we can participate in a democratic way.
I’m staying an optimist. That’s why I’m still in digital health.
SP: I’m staying an optimist. That’s why I’m still in digital health. In the end of the day, innovators and people with ideas, especially those helping patients, are going to succeed. We have all these obstacles for a reason, but history is on our side. The most innovative ideas are here to stay. They will help patients, physicians, and all of us.
Please follow what I do.
Digital health, AI, and machine learning are the future of healthcare.
SH: Thanks again for the optimistic note. To people out there, I’m always looking for different ways of looking at problems. There are many complex problems—the simple ones have been solved—but complexity is very much in every part of nature. Therein lies the opportunity to get to the next stage. Thanks again, Sergei.
SP: Thank you, Shankar. I appreciate it.
I hope you enjoy the interview. I realize I don’t come across as outraged and infuriated as I do in my writings, and I apologize to those whose image of me I may have disappointed. 😊 Still, I always look forward to any questions and appreciate any discussion.
Thank you! 🙏
Sergei
👉👉👉👉👉 Hi! My name is Sergei Polevikov. In my newsletter ‘AI Health Uncut’, I combine my knowledge of AI models with my unique skills in analyzing the financial health of digital health companies. Why “Uncut”? Because I never sugarcoat or filter the hard truth. I don’t play games, I don’t work for anyone, and therefore, with your support, I produce the most original, the most unbiased, the most unapologetic research in AI, innovation, and healthcare. Thank you for your support of my work. You’re part of a vibrant community of healthcare AI enthusiasts! Your engagement matters. 🙏🙏🙏🙏🙏