Merck

Fernando Schwartz

As a young and passionate individual, Fernando Schwartz came to the US with a dream to become a Mathematician. He attained his PhD in the subject and worked as a professor for about 10 years, building his acumen in data science. After completing his tenure in academics, Schwartz realized, as he says, “I had all the science, but none of the data”, which resulted in transitioning away from his living dream of a Mathematician to getting into the online advertisement industry to demystify the impact of data science and machine learning in the real world.

Schwartz became particularly interested in learning the science behind online ads—showing the right ad to the right person at the right time. As he developed a deeper level of understanding in the same, he thought of repurposing his knowledge in the healthcare sector to treat the right patient at the right time and do something altruistic. Schwartz finally switched to healthcare sector, where he is focused on transforming healthcare standards by using the combined power of data, analytics and AI.

In an interview with Pharma Tech Outlook, Schwartz shares his insights aboutthe adoption of cutting-edge technologies such as AI in the healthcare sector. He also throws light on the limitations and scope of AI for the industry.

How do you think the pandemic has brought about a new perspective in healthcare technologies?
The pandemic has accelerated the desire to take advantage of various technologies and use them for healthcare. Currently, we are undergoing a massive transformation discovering the full potential of deploying cutting edge technologies in healthcare.There is more openness to try new things, take telemedicine, for example, which has proven to be a big hit with patients/consumers because of an improved experience. However, it is going to take a while when it comes to unlocking the full potential of more cutting edge technologies using AI.

Why has the healthcare sector not been able to use AI to its full potential?
Currently, AI is being leveraged by various industries. Some verticals such as, say, online advertising, is incredibly advanced in this front. So far, we can say that AI has been successful in guessing the purchasing intent of a person, and if this same technology were to be used in healthcare, one could hope it could solve most of the problems. However, the fact that AI has been great at guessing consumer intents comes from the massive amount of data it uses to make those predictions and the type of data used to make them. Unfortunately, this does not apply directly in healthcare, where there are vast data challenges as sources are varied, and a patient’s journey is at the center of the predictions.

To make things worse, the access to data is not equal across geographies. For example, in Europe, the data privacy regulation is very stringent. In the US we have access to commercial data, but in many other places, healthcare data is controlled governments. Owing to such different access rules, it is difficult to share comprehensive data assets. Such a varied landscape makes it highly difficult to find a one size fits all solution.

That is why there is a lot of work—in the healthcare landscape—in piecing together the different chunks of data, which many times come from multiple sources with different formatting. I think that healthcare is at a clear disadvantage compared to other sectors in the industry because of the fragmented nature of data sources. Although things aremoving a little slower in healthcare, we will soon witness the widespread applications of AI across healthcare.

What do you think is the future of AI in healthcare?
I would equate AI in healthcare with selfdriving cars in some sense. While AI is an efficient tool to generate complex rules, it may not perform well in some edge cases, where things may go sideways. For example, an edge case for a self-driving car can means the car loses its control, which may result in people’s death.

However, in self-driving cars, we have explored quite a bit of the edge case situations where we can almost guarantee that nothing too extreme is going to happen. And I think that is a good parallel to healthcare. When we talk about healthcare, we consider a broad spectrum of tasks and operations. I think that different companies across the healthcare spectrum naturally carry different degrees of liability in their operations.

The companies that have the most liability will likely be slower in the deployment of AI. Take insurance companies, for example, AI is quite advanced in the payer space because it mostly deals with financial decisions. In such cases, edge cases have just financial implications. But as we talk about patients and their lives, mistakes are much more consequential and may be life-threatening. In a nutshell, I would say that the penetration of AI in healthcare is inevitable, but it wouldn’t be the same as the penetration of computers, which is being used in every single industry. I believe the financial areas in health care will be more advanced in adoption of AI than in clinical areas, where we can expect humans to remain in the loop.

What would be the single piece of advice that you could impart to your colleagues to excel in this space?
AI is a set of techniques that have not fully matured in the industry; there is some degree of speculation around them. So, when they are setting expectations, my piece of advice would be to build a team of highly technical people. If you are trying to solve a machine learning problem—more likely than not—this problem will have some uniqueness to it, and if you don’t have a deeper technical ingredient in the mix, it would be setting yourself up for failure. So, my last piece of advice would be to remain sceptical about the technology, but, don’t let that curtail your ambitions and always seek advice from the technology experts.