In this month's "Discussing Data Science" episode, I talk with Jeff Kozloff, a serial entrepreneur, advisor, and investor in the biomedicine/healthcare industry.

You can watch the video below or on Youtube. But if you'd prefer to read, keep scrolling. The complete transcript (edited for length and clarity) is below.


Spencer Hey (SH): Hello, my name is Spencer Hey. I'm the co-founder and chief science officer at Welcome to Discussing Data Science.

SH: My guest today is Jeff Kozloff. Jeff initially started in the biomedical space as a neuroscience marketing consultant for Bristol-Myers Squibb. However, during the 2000s, he pivoted to entrepreneurship, founding several companies, and eventually selling them. One such company was "Verilogue", which focused on aiding pharma and other significant institutions in navigating the complexities of data and regulatory compliance. More recently, Jeff was the CEO at "TrialScope", a clinical trial transparency management solution, which was acquired in 2021 by Informa.

Jeff has also been serving as a board member and advisor to a variety of startups and companies working to solve complex data and analytic problems in clinical trials and adjacent domains. Jeff, welcome to the show, and thank you so much for joining me today.

Jeff Kozloff (JK): Hello, Spencer, it's great to be here.

SH: Let's start by talking about what initially drew you into this space. How did you enter the biomedical field?

JK: It's been a long, but fun journey. I graduated as an undergrad in '99, right around the dot-com era. I began in strategy consulting where most people from my alma mater typically went down either the investment banking or the strategy consulting route. I chose the latter because I always loved healthcare. Many people who are interested in healthcare think they want to be doctors. I did a stint shadowing an ER doctor and spent some time in a pathology lab. But I realized medicine wasn't where I could see my career going.

Despite this, I continued to love medicine. In one of my business classes in undergrad, we had to bring a biotech idea to the market. I ended up doing more of the business side and less of the science side for the team. This experience led me to fall in love with the business aspect of biotech. Consequently, I switched from bioengineering into business school for my undergrad. After strategy consulting, I found my way into market research.

During this time, internet market research was really starting to take off, moving away from paper. I worked for an exciting venture-backed company in this space. Even though I wanted to run my own business one day, I first wanted to get the buyer's side experience. That's how I found my way over to Bristol-Myers Squibb. There, I had my first lightbulb moment. We were traveling around doing what they call a three-city tour for market research.

We spent a week talking to psychiatrists in one-on-one research or focus groups. Then we talked to patients the next week in three different cities, and sometimes we did the same thing with payers. On one of those trips, I was flying home, kind of burned out, and read an article on the plane. It was about how Kimberly Clark Corporation had put little video cameras on mothers to observe them during playtime. What they observed was the complete opposite of what the mothers told them in traditional research.

That was the lightbulb moment for me. I thought: We need to get out of recall and into reality with the doctor-patient conversation. We then built a system to record physician-patient conversations in a compliant way, which led to the world's largest database.

SH: That's fascinating. Around what time did you launch those recordings?

JK: We launched Verilogue around 2005, long before audio health became what it is today. At that time, physicians were using Dictaphones, so we wrote some software for the back of an SD card on the Dictaphones. We had what we call a double opt-in, meaning that both the patients and the doctors verbally agreed and gave us written consent to record. We de-identified all the personal information.

As technology grew, we moved over to a mobile app to do data collection. On the back end for data analytics, we used a two-pronged approach: We employed machine learning to identify those moments of meaning in the dialogues, and then we had human linguists bring the data to life, using linguistic metaphor and imagery analysis.

SH: That's impressive. Just thinking about the recent explosion in AI and large language model (LLM) technology, that data must be even more valuable today.

JK: Absolutely. I think it also shows how long it takes for technologies to mature. I'm really proud of what we built, and the company is still thriving under Publicis who acquired us. Verilogue data was sold to Google, Amazon, and others as training datasets because we had such a large library of two-way audio physician conversations and a married-up transcript. Some of the commercialized models were probably on the heels of the great work that the team at Verilogue did.

SH: So Verilogue was your first venture into the biomedical space as a business, right?

JK: Yes, as an entrepreneur and a CEO. My good friend, Jamison Barnett, and I founded it out of our basements. This was a time when everyone was looking for free Wi-Fi. We knew every Panera Bread in the area to meet up and use someone else's juice. But as you know at Prism, the first few customers are everything to a small business. We were lucky to have a couple of people take a chance onus, and they've become lifelong friends because of the gratitude we have for those who believed in us and helped us get off the ground. 

The Draw of Medicine and Healthcare Industries

SH: You started in the healthcare field, and despite exploring various other avenues, you've remained largely focused on healthcare and biomedicine. What about this domain keeps you attracted and motivated to continue solving its diverse problems? 

JK: Well, I find the healthcare sector exhilarating. It's a highly dynamic field that significantly impacts our society and GDP growth. The role it played over the past few years is even more prominent, which is incredibly rewarding. What particularly fascinates me is the multitude of disciplines and problem areas within the healthcare ecosystem that are waiting to be addressed.

Due to its heavily regulated nature, the healthcare industry can be slower in adopting new practices compared to other sectors. However, this aspect provides an entrepreneur with opportunities to implement best practices and advanced technologies used in other industries, especially if one has healthcare domain expertise. For instance, we're still doing a lot of paperwork, using Excel, and engaging in manual processes that can be disrupted with automation.

While there's considerable interest in disrupting the status quo, many of these changes aren't revolutionary; they have been applied elsewhere. It's more about strategic adaptation, which I find quite exciting. As an entrepreneur, I'm continually energized by innovation and by people who consistently challenge the status quo. The idea of maintaining old ways simply because "that's how we've always done it" is like nails on a chalkboard for me. This innovative spirit is what gets me out of bed every day.

One Solution or Many?

SH: You've explored multiple facets within the healthcare value chain. It seems there are still plenty of opportunities to streamline processes, expedite work, and provide better insights. However, the industry's size and complexity creates challenges, such as fragmented data and data silos, which can inhibit effective communication and coordination. I'm curious to know your perspective: Do you envision an opportunity for a unified platform to bring all these different pieces of data together? Or do you think that specific solutions are needed due to the varying requirements of each segment?

JK: Like most things, the answer is nuanced. The healthcare field, from drug discovery to commercialization and patent experience, has so many different facets, many of which I'm not an expert in. However, in areas like clinical trial data analytics and commercial launch, I believe that standards play a crucial role. While people joke about not wanting another standard, I think bodies like CDISC are doing an excellent job in harmonizing some of these standards.

That said, as much as we talk about open APIs and data sharing, these often remain theoretical concepts. The competitive nature of business and the difficulty of articulating win-win scenarios for sharing can make it hard to transition from theory to practice. I believe that specific use-cases and strategies that provide mutual benefits for all parties involved can help to overcome these challenges.

One such challenge is data duplication, which is not only wasteful but also slows down discovery. Streamlining this process could make trials faster and more affordable.

SH: Can you provide an example of data duplication that you find particularly unnecessary?

JK: Sure. One glaring example is in clinical trials, where data from EHR (Electronic Health Records) has to be manually entered into EDC (Electronic Data Capture) systems. This duplication not only consumes valuable time but also requires additional checks for accuracy. If we can automate these processes, we would free up healthcare professionals to spend more time with patients rather than technology. This could accelerate scientific discoveries and deliver breakthroughs to the market faster.

What are the biggest data gaps in the field?

SH: Transitioning to our final segment, what do you see as the biggest data gaps in the field?

JK: One critical gap that comes to mind is in the area of diversity and inclusion in clinical research. There is a pressing need for trials to be more representative of the population. Early in my career, I worked on understanding doctor-patient communication, categorizing different types of patients and physicians, and studying the role of language in healthcare communication.

I believe we can leverage advanced language models to identify effective communication strategies for different communities and implement them in medical discourses. This approach would help to overcome cultural barriers and foster more effective communication. Other organizations, like MRCT at Harvard, have done excellent work in this area, but we need to take this further. Technology can help automate some of these processes and disseminate this information to a wider audience, leading to significant improvements in the field.

What excites you the most about the future of research?

SH: For our second question: What excites you most about the future of research? 

JK: What truly excites me is the remarkable innovation happening right now. The COVID-19 pandemic, while devastating, served to expedite a multitude of ideas in our industry. It's unfortunate, but it also pushed us decades forward in terms of research pace. If we can maintain this momentum, we will continue to innovate at an increased rate, which is very exciting.

Large language models are particularly interesting, as their capacity to provide context-rich answers fills a gap for many individuals. Whether comfortable with data or not, everyone wants to make data-informed decisions. Accelerating research in this way—providing comprehensible and insightful data—allows for quick decision-making and adaptation, which is absolutely amazing. This can be applied everywhere, and to me, that's very thrilling.

Also, as we were discussing democratization, this is going to attract more entrepreneurs, leading to even more ideas and disruption of the status quo. This could reduce the time it takes to bring new ideas to life indifferent domains from five or ten years to one or two years, which I find quite exciting.

Wave the magic wand...

SH: Third question: If you could wave a magic wand, what's one thing that you think you would want to change about the field?

JK: My wish might be more practical. Starting and scaling a business, especially in the slow-moving pharmaceutical industry, is difficult. Even innovation teams can take a long time. My wish would be to give more young companies the chance to try things out, to find budget dollars and introduce more ideas, to see what works. If two mediocre groups were combined, they could achieve incredible things. My hope is to speed up the process for entrepreneurs, helping them secure early business so that they can either fail quickly or scale faster. This would be beneficial for everyone in the industry.

Although I don't have this magical wand, public-private partnerships and hospital systems have done some good work building ecosystems and incubators to support young entrepreneurs. However, we need more of that, and we need it faster. We need to cut decision cycles from 12 months to 45 days and simplify the contracting process. That's what I'd love to see.

SH: That's great. As you were speaking, I thought it would be wonderful to make it easier for people from academia to venture into entrepreneurship. There are so many intelligent people who, despite having the necessary training and great ideas, feel constrained by the academic system. Whenever I talk with young academics now, I encourage them to expand their horizons to consider business as viable career path, as there are just so many opportunities out there. If we could create an easier way to test out their ideas in the industry, that would be very exciting.

JK: I absolutely agree. This is similar to what we were discussing earlier about community physicians and patients. There should be more opportunities available to academics, just like there are for patients. If we make this a regular part of the conversation, we might see more people choosing this path.