Point of view

Data + AI + Expertise: The Trifecta for Health Tech Success in Pharma

Dimitrios Skaltsas /
Feature image

Since 1885, the pharmaceutical industry has developed treatments that have increased life expectancies and provided life-improving therapies, from the 1899 introduction of aspirin to the discovery of penicillin in 1928 to the launch of Lipitor in 1997. Understanding disease biology and drug mechanisms has matured and will continue to evolve and grow. New modalities, mechanisms of action, targets and advances in personalized medicine continue to elevate the realities of what’s possible.

At the same time, artificial intelligence (AI) is poised to reshape drug discovery and development for the better. However, the hype doesn’t always match the reality. Through our work at Intelligencia AI, we know that for AI to unleash its full potential, well-defined use cases and realistic applications, access to appropriate data, expert talent and building industry trust in AI are mission critical.

The Underlying Challenge in Pharma: Risk

Let’s start with the big-picture challenge in pharma: developing a single drug can take a decade and over $2B. This challenge is exacerbated by the fact that the vast majority of drugs that start down this long and expensive path never make it to the other end—only about 10-15% do. These high failure rates greatly diminish focus, waste significant resources, and delay the delivery of the most promising drugs to patients. The “R word”—Risk with a capital R—is a challenging reality in the pharmaceutical industry.

Throughout drug development, there are numerous inflection points where critical and highly consequential decisions are made. The more quality data to be interrogated by powerful analysis tools, like AI, result in more actionable insights that answer critical questions and enable insight-driven, not just data-driven, decisions. Such insights can help answer pharma stakeholders’ questions:

  • What is the right balance between early- and late-stage assets in our drug portfolio to best achieve the desired returns?
  • Which assets should we acquire/in-license?
  • How can we design our clinical trials to minimize risk?
  • How advanced is the competition?

The Core Ingredients: Purpose-built Data and Explainable AI

To mitigate risk in drug development, purpose-built data and powerful, reliable and trustworthy AI are required.

  • Expertly curated and harmonized data: Data alone has become somewhat of a commodity. However, not all data are created equal. High-quality, expertly curated, purpose-built data harmonized from trustworthy sources are required to power AI models and generate reliable, actionable insights. The oh-so-true principle of “garbage in, garbage out” applies to AI like any other method. If AI algorithms are fed incorrect, incomplete, or biased data, the results will produce inaccurate, incomplete, and biased answers.
  • Explainable AI models: While AI has existed for several decades, only recent advances have enabled novel models to ingest and crunch the vast amount of data generated in drug development quickly and generate insights in a transparent, explainable manner that instills trust in the results.

Reliable data is the backbone of trustworthy AI—and in the pharmaceutical industry, where AI-generated insights can directly impact patients’ lives —precision and accuracy matter.

AI Alone Is Not Enough

Not just any AI, but explainable and transparent AI, is needed to help take the pharma industry to the next level for AI-driven insights. AI alone is often perceived as a mysterious black box that hinders adoption. A lot of information goes in on one end and on the other end, an answer magically seems to appear. How did we get from here to there? How do we know we can trust those results?

Explainable AI sheds light on which of the many input factors contributed significantly to the answer AI provides, therefore adding transparency and allowing experts to understand the drivers and perform a “sanity check.” Developing trustworthy AI is vital—it must be explainable, and methodologies must be transparent.

Oversight by Human Experts Connects it All

Perhaps one day, AI will become smart enough to independently detect issues related to incorrect, incomplete and biased data. Until then, humans will play a significant role in cleansing and curating the data used to train the AI models. Companies utilizing AI, therefore, need to focus on building a robust and highly reliable data foundation.

Training AI algorithms on expertly curated, quality, comprehensive, and harmonized clinical and biological data sits at the core of the company I lead, Intelligencia AI. We go to great lengths to make sure what the algorithms digest is correct, complete, and free of bias. AI is not a replacement for people’s expertise. It’s another very advanced tool that needs human oversight to tap into.

Intelligencia AI: Startup Success in the Pharma Sector Powered by Data, AI and People

At Intelligencia AI, it is part of our DNA to help our current and future large pharma customers understand which factors inform the results AI calculates. Our patented accurate and explainable AI-driven probability of success (PoS) assessments are a critical metric in the industry that helps make better decisions in this risk-laden environment. We can go behind the scenes into the data to show the ingredients informing the AI algorithm.

A common thread connecting data and AI is driven by one irreplaceable ingredient—talented people. The foundation of any solid and successful company is its people.

I started Intelligencia AI in a large market – New York City in 2017 – and then returned to – Athens, Greece – to further develop our R&D division here. At one point, many of my current colleagues left Greece, lived abroad, and now have returned home, contributing to the brain gain and impressive and growing tech ecosystem in Greece.

While no Silicon Valley (at least not yet), what is being built here in Athens is truly unique. As I currently split my time between New York City and Athens, I hope this international blend of building companies and finding the best talent on both sides of the world continues and becomes more commonplace. This is a shared vision with Big Pi and Aristos Doxiadis, who sits on our board of directors. I also want to highlight Nick Kalliagkopoulos’s recent post, which captures the untapped potential markets like Greece that hold to be tech meccas. I could not agree more with his evaluation of the landscape here.

Quality data and explainable AI will only take all industries—including pharma—so far. People are essential in building a solid foundation, and no matter what industry, we need individuals’ expertise and passion to continue raising the bar. We’re fortunate to be one of the companies doing just that.

Connect with Dimitrios on LinkedIn and visit intelligencia.ai to learn how the right combination of data, AI and talent can reduce risk in the pharmaceutical industry.