The life sciences supply chain starts with a miracle
Is there a more challenging industry than life sciences? In a recent Forbes interview on why it’s so hard to bring tech into pharma, Vas Narasimhan, Chief Executive Officer of Novartis AG pointed out that only one in twenty drugs make it out of clinical trials.
One in twenty. A rate that hasn’t increased in over a decade. The only thing that has increased are the costs. Everything we talk about in supply chain starts with discovering that product you can market and sell. For those in consumer-packaged goods, it’s safe to say we’re fairly knowledgeable about our customers and target markets.
Life sciences is much different than bringing a new phone, app or appliance to market. Narasimhan highlights that bringing a product to market that will have a meaningful impact on curing a disease is an audacious goal at best. “What’s often lost on people is how incredible it is that we find any human medicines at all.”
Imagine knowing so little about your customer, in this case, the human body.
The article reminds us of what Merck R&D head Roger Perlmutter said about the subject back in 2013. “We really understand very little about human physiology. We don’t know how the machine works, so it’s not a surprise that when it’s broken, we don’t know how to fix it. The fact that we ever make a drug that gives favorable effects is a bloody miracle because it’s very difficult to understand what went wrong.”
Although getting a new drug to market has its unique challenges compared to the source, make and deliver part of the life sciences supply chain equation, there is one thing those of us in supply chain have in common with our R&D counterparts, we all need clean data.
The goal of the clinical trial challenge is to use machine learning to dive into the oceans of data and find new insights. The prerequisite for that though is clean data. How many times have you heard we need to take care of our data before we can start a project? Narasimham explains, “We’ve had to spend most of the time just cleaning the data sets before you can even run the algorithm. That’s taken us years.”
Humans still need to define “clean” when it comes to data integrity, but perhaps we’ll start to see technology that actually helps with the data cleansing. Imagine taking years off the clinical trial process and getting the most effective drugs to market faster. Regardless of industry, we want artificial intelligence and machine learning (AI/ML) capabilities to take the robot out of the human being and reduce the wasted time spent cleaning data. That way we can get on with effective and efficient analysis to determine what story the data is telling us. That’s where the human being is the most productive, not swimming through oceans of data.
It will be difficult to replace the rigorous testing that goes with clinical trials. What happens to a patient between study visits is a variable that’s tough to capture. From a supply chain perspective, it reminds me of variable lead times. The reason lead times fluctuate is difficult to capture in a complex life sciences supply chain network. The win comes if you can track the variability against acceptable tolerances. Then action can be initiated to take the pain out of planning with lead times that do not represent reality. Sounds like hype but we can still imagine.
Sure there’s still hype. Who would think an algorithm or black box would resolve the complexities of creating and getting a drug to market in time to make a difference. However, we are starting to see practical use cases for emerging technologies. Narasimham calls out some success stories where AI is delivering results in the areas of clinical trial operations and finance, but it still starts with the data. In supply chain, we’re seeing AI/ML capabilities successfully applied to demand sensing. Think of Google and Amazon making product suggestions and predicting what you will buy, From a logistics point of view you can look to the advancements in autonomous vehicles as an example as well as self-healing capabilities to manage variable planning inputs.
Where’s the planner in all the data, complexity and volatility?
Regardless of the application, AI/ML applied to supply chain problems is data hungry. To add to the challenge, you can throw in the complexity and volatility that is a way of life for supply chain practitioners. While you are working on consolidating, cleaning and feeding massive amounts of data onto a platform for AI/ML, you still need to deliver medicines at scale and in time to make a difference in human lives.
Planners can only do so much, but what they do well is to make the best decisions possible when presented with all the data in a way that makes it easy to see what story the data is telling them. Certainly, some routine planning decisions can be automated. Some call this lights out planning. I prefer to think of it as planning with a dimmer switch. Given all the volatility in the life sciences supply chain and the relationships between all the nodes in a complex supply chain, there are times when the business needs to turn up the lights and allow the planners to course correct in time to have a positive impact on key metrics. At other times, you can lower the dimmer and automate routine parts of the planning processes.
Will we see more miracles?
This takes us back to the first challenge of leveraging the technologies – data. To Narasimham’s point, getting clean data is at the heart of leveraging AI/ML capabilities in the life sciences supply chain, starting with clinical trials through to the delivery of product. Perhaps the reason for some hype is that AI/ML use cases tend to focus on specific processes before taking care of the data. If that is the case, we need to look for technology help in a few areas:
- Synchronizing the network while assisting with the data integrity
- Augmenting, automating supply chain intelligence
- Elevating the planner experience to make confident decisions in a complex, volatile supply chain
The good news is that some of the successful use cases referenced were in the latter two of the three areas identified above. The more strides we make in the first area, synchronizing the network and assisting with data integrity, the more we’ll be able to leverage AI/ML in the rest of our supply chain processes.
Do you have any success AI/ML success stories? We’d be more than interested to hear about them.