Industry
Pharma companies prioritizing technology for growth: ZS CEO
Summary
- ZS, a US-based tech services and consulting firm, has a significant India presence, with around three-fourths of its workforce and four-fifths of its research based across five offices here
New Delhi: ZS, a US-based tech services and consulting firm, has a significant India presence, with around three-fourths of its workforce and four-fifths of its research based across five offices here. In conversation with Mint, Pratap Khedkar, chief executive of ZS, said the company reported sustained growth in contrast to the top IT service providers due to resilient tech spending across the pharmaceutical sector. He also spoke about how global pharma giants are prioritising tech investments to grow revenues, over margin expansion, and the distinction bletween artificial intelligence advanced statistics, and process automation. Edited excerpts:
How is the downturn in IT affecting ZS?
Most of India’s tech majors operate across nearly 50 industries, while ZS focuses only on healthcare. When macroeconomic fluctuations happen, the healthcare sector remains recession-proof. But, healthcare has its own challenges independent of economic factors, such as replacing the revenue generated by a blockbuster product that moves out of the market. These are up and down cycles, which do affect us.
How are pharma majors maintaining discretionary spending on technology?
Digital transformation and tech spending in pharmaceuticals and healthcare are resilient for many factors—one of them is a patent and product expiration slope. In the next five years, products that account for 41% of global sales in pharmaceuticals will expire. As a result, firms are trying to squeeze more returns out of these products by getting these products to more users. Also, digital insights can increase proliferation of a medical product. Nearly 60% of pharma companies globally are adopting omnichannel orchestration to expand the reach of drugs to physicians and patients. Over time, we are seeing that, doing so, increases the top line by 8-10%, through AI-driven precision-marketing. The other area of improvement is R&D, where the impact takes longer to be seen. Organizations spend $2.6 billion per approved molecule (discovery and implementation of a new drug). This was quite wasteful so far. Now, firms are looking to reduce this cost by up to 20%, with AI designing clinical trial protocols. Adaptation of new drugs is now faster, thanks to a digital framework governing the drug discovery process.
Is AI and digital transformation aiding the pharma industry to improve margins?
For pharma industry, top line is more important than margins, because the variable margin of drug business is fairly high since after the expensive R&D process, one needs to only keep producing enough. Its not true for other sectors, for instance, med-tech had supply chain issues during the covid-19 pandemic, which did not hit pharma much.
The importance of margin comes primarily to healthcare providers—they are recession-proof, but their margins are super thin. For instance, average operating margin of a hospital in the US is actually zero, or even slightly in the negative. Nearly 60% of all hospitals in the US lose money. This leads them to focus a lot on operational efficiencies, which is where digital transformation comes in.
How big is ZS’ India presence?
Around 80% of ZS’s research experts are based in India, and 75% of its overall employee base is also here.
Is our understanding of AI primarily driven by hype, given that most products seem to use process automation instead?
Through the first two decades leading up to 2,000, and even beyond, data businesses were using complex multivariate statistical models. We used this to offer insights on advertising, representatives and other avenues where they were spending money. The models would already offer results in terms of returns on investment, all of which was already possible through advanced statistics.
Ten years ago, we reached a point where AI kicked in, and advanced statistics ceased. A decent statistical model can already offer prediction—AI can do it more precisely. In some areas, we started realizing that if you put in enough data, you do get better predictions. For instance, in case of a new drug discovery, use cases where AI can become useful is how often a physician can prescribe it—a factor on which we do not have any current data, since the drug launch is a year away.
AI would infer this by taking into account prescription trends, size of the physicians’ practices, trend of adopting new drugs, as well as non-related lifestyle data such as preference of food, magazines and so on. All of this could help us generate, say, 3,000 data parameters. This is where a neural network can help process predictive analytics, while statistical models simply fail.
How did AI improve predictability? Did we lose anything in the process?
A very complex model can offer insights with nearly 80% accuracy, by narrowing down all data to 30 relevant variables, given that the model was trained on enough data. This is the jump that AI offered—by making our predictability much more powerful. However, what we sacrificed for it is interpretability, which is where we don’t always understand why a certain result is being showed by an algorithm.
