“We need integrated supply-chain data to outsource development and manufacturing more efficiently, be more productive and profitable, and get drugs to patients more reliably and affordably.”
-- quote similar to what you’ve said or heard from biopharma executives
This is a brief essay on our progression to data as divine deliverance, and the role technology – particularly artificial intelligence (AI) – plays within that salvation.
Whether this comes out as perfectly pedestrian or pleasingly profound is up to the reader. It is, though, a deliberation derived from various sources and conversations with biopharma executives, and from books on the (human) interpretation of data. Keep in mind it also comes from someone with an enthusiastic predilection for new technologies (e.g., blockchain and nano).
Just Another Link In the Chain
Data is the language of business in our times. Others consider it – again in today’s vernacular – as currency we trade back and forth.
Yet data itself is nothing more than a type of memory without historical or other context. No matter the means by which it is generated, only until it is infused with some form of human understanding, does data become contextualized information. We turn that it into human knowledge.
What do we use that human knowledge for? The most important utilization is for making decisions that predict – and hopefully create — future outcomes. For the biopharma industry, those decisions are often related to supply chain management of drug development and manufacturing, and for quality and compliance operations.
However, there’s so much data to work with today that we’re having trouble making those necessary knowledge transformations from data to applicable (human) understanding. It’s become a profound challenge within our milieu of supply chains and the entire gamut of the outsourcing of drug development and manufacturing.
To help overcome these challenges, drug sponsors are reaching for technology-based analyses, such as from AI. (Generally in health care, the FDA is jumping on board as well.) In other words, the thinking goes, as AI increasingly provides us more of those data points, it might as well add the actual, contextual understanding of them. Thus, data analytics and the like taking our supply chains by storms.
But in our realm of outsourcing development and manufacturing services, will drug sponsors in fact expect that in place of humans turning data into knowledge, a “non-human” intelligence will perform the final step of cogitation? And if so, does that lead us to “artificial” decision-making along the steps from drug development to commercial sales?
If you answered yes, you’ve embraced a brave new world. If no, you still believe conversion of data into human knowledge – and the decisions taken upon that knowledge – require human intelligence over AI.
But whether a “yes” or a “no,” as we exist today, one thing remains clear: Despite all the new data opportunities and advancing analytics, humans are still held fully responsible for the consequences of decisions made in drug development and manufacturing.
If something goes awry, you can be certain “AI” won’t get the blame. You humans will be held fully responsible by your shareholders, the drug agencies, your patients, and society at large.
An assortment of data, using a simple supply-chain example, may suggest the processing of a batch of API differed from the processing of other batches, but whether that one batch needs to be destroyed or not is still a human decision: For sure it is a more data- and analytically-enriched decision than ever before, but still a human decision. What if there’s a material shortage? Patients waiting? (And what did you learn from that processing anomaly?)
Moreover, Outsourced Pharma readers should be contemplating this more fundamental question: What role would the CDMO even play in a (fully) data-driven decision-making ecosystem? Over years, the CDMOs have moved from being viewed as “a pair of hands” for hire, to a strategic, value-adding partner. Would we now move the CDMOs to a type of “data-delivery service”?
The salient point raised here is we need to put AI in a certain perspective, if not in its place: I’d submit that the additional output of AI – both data and the increasing analysis of which – is not a substitute for human intelligence in ultimate supply-chain decision-making.
Instead, it’s another point in your supply chain.
AI To HK
Therefore, let’s not overshoot the goal of AI.
We are making better-informed decisions via elevated forms of tech-driven analysis. This particular realm of AI we are discussing offers great assistance for executive and managerial decisions, for example, filing INDs or NDAs; increasing or decreasing material volumes; cost of goods and pricing; and regarding the further partnering with or replacing of your CDMOs.
Most of us have been focused on “big data” as a basic (but growing) tool, and initially concerned with “data integrity”: Are we measuring the right things? Is the data itself being generated and recorded accurately? Is it adequately protected? Is it actually reproducible, and/or applicable across the supply chain?
We must now turn – and be adequately concerned about – that next step: the integrity of our thoughts about that data. Those are the thoughts that change artificial intelligence into human knowledge.
Here is where those books on research into human thinking I mentioned in the beginning of this brief discourse come in. They are readily accessible, and have titles such as: “The Art of Thinking Clearly,” by Rolf Dobelli; “Thinking, Fast And Slow,” by Daniel Kahneman; “Predictably Irrational,” by Dan Ariely; and most recent, “The Undoing Project,” by Michael Lewis.
These publications inform us that we humans in fact don’t think as clearly as we think we do – particularly when it comes to drawing rational conclusions from data presented to us.
That conversion of data to “human knowledge,” as I’ve phrased it above, is beset with its own challenges, set within the foibles and flaws of the human process of decision-making. Remarkably, one of those flaws is often we also make worse decisions the more information we have.
Of course the recognition of many of these flaws has in fact been a catalyst on this road to AI and advancing data analytics.
So let’s stay balanced. Let’s indeed welcome the assistance we receive from AI, machine learning, data processing and advanced analytics, but also from AR (augmented reality), VR (virtual reality), IoT, supercomputing and all the rest. We can use all the help we can get.
But the ultimate decisions based on every and any input of data or analysis in our drug supply chains, and the full responsibility for the outcomes of those decisions, will always reside with you.
And the last time I checked, you were all still human.
Unleash Blockchain Technologies On The Entire Supply Chain
The World’s First Biopharma Built On Blockchain
Takeda CEO Mandate Sets Off A Nano Reaction
Nano Aids Pharma In The Business Of Delivering Chemo
Move Over ADCs: Nanoparticle-Drug Conjugates Are Joining The Cancer Figh