Table of Contents
Artificial Intelligence is Transforming the Pharma Supply Chain
When news broke of Pfizer’s successful Covid-19 vaccine trials, there was jubilation around the world. But that joy gave way to caution, as people began to realize that the existing pharma supply chain might not support a mass vaccination program.
The Covid vaccine candidate needs to stay at -94 degrees Fahrenheit at all times, or else the heat might render it inactive. This isn’t a unique problem, and pharma companies have been an existing cold supply chain for other drugs. The problem is that a Covid will mean scaling up existing supply chains at an incredible speed.
But we have a secret weapon that can help: Artificial Intelligence.
How AI fits into the pharma supply chain
First of all, what exactly is AI?
Right now, the most prevalent form of AI is Machine Learning. ML isn’t “intelligent” as such. An ML algorithm looks at massive volumes of data and studies them in intense detail, identifying patterns that might escape human observers.
The ML algorithm “learns” from these patterns. It can identify similar patterns and extrapolate future events with surprising accuracy, as long as you provide the algorithm with enough data.
Which is why ML is perfect for supply chain management. The supply chain is, ultimately, a data problem, with lots of moving variables. You need to track things like:
Current stock levels at the production facility, including raw ingredients and ready-to-ship packages.
The number of units emerging from each facility. Many companies produce small batches of bespoke medication, which makes this metric even more important.
Information about the entirefleet of delivery vehicles, including cars, trucks, planes and boats. This data includes location, fleet availability, and information about each vehicle (such as on-board temperatures.)
Details of the consumer of each shipment, including their location, their order, and whether the order was accepted.
Some orders will be rejected or marked as undeliverable. This involves a reverse logistics process, shipping back to the pharma company’s facilities.
Thanks to Internet of Things (IoT) technology and 5G, we now have access to highly granular data about every step of this process.
So, the challenge now is to make sense of all of this information and to respond to it in real-time. Machine Learning algorithms can help to do this with greater accuracy than ever.
AI and ML use cases in the pharma supply chain
When we talk about artificial intelligence, we tend to think of a single platform that does everything.
That’s not actually how AI is rolling out. Instead, we’re seeing AI and ML creep into existing systems, adding new functions while improving efficiency. Many supply chains are already dependent on AI to a surprising degree.
Here are a few ways that AI and ML are improving the pharma supply chain already.
Cold Chain management
This is possibly the most urgent issue in the pharma supply chain today, thanks to the Covid vaccine. Perishable drugs such as vaccines have to stay under a specific temperature, from the second they’re produced until the moment they’re administered to the patient.
One of the biggest problems in cold chain is transportation. You need to verify that the shipment has remained below the recommended temperature at all times – even a few hours of heat could render it inactive.
ML and IoT can really help here. IoT devices can track temperatures on individual items, picking up the minutest changes. ML can track these changes and decide whether there is any risk of overheating. If so, then the system can send an alert to the driver, who can take action to protect the shipment. When the shipment finally arrives, there’s a detailed event log for the entire journey.
Choosing the right route can have a huge impact on supply chain efficacy. The right planning can help to shave hours or even days off the journey time. This is vital when shipping perishable medication, as a shorter journey could mean the difference between success and failure.
ML can help optimize journeys by breaking the journey into chunks, and solving each chunk as a discrete problem. The algorithm can look at historical data and available traffic data to avoid delays, and identify opportunities to combine deliveries and reduce costs.
This is an important step towards the ultimate goal of autonomous delivery vehicles. For now, AI tools can plot the best journey and make spontaneous route adjustments if required.
Fighting counterfeit medicines
The trade in falsified medicines is worth over $77 billion per year. That’s not including the cost of patients receiving dangerous counterfeit treatments that could make their health worse. From a supply chain issue, there are two major issues to consider: how ingredients fall into the hands of counterfeiters; and how falsified medicines get back into the supply chain.
Both problems are ideal use cases for machine learning, which is used extensively in the financial industry to identify fraud. It does this by learning transaction patterns commonly associated with unauthorized activity. When it spots a similar pattern, the algorithm flags up potential crime.
ML in supply chain uses a similar technique to fight counterfeit medicines. The algorithm analyses all supply chain data, and flags up any activity that looks suspicious. Even if the counterfeiters change their technique, the ML tool will simply learn the new pattern.
Pharma supply chains often work on a just-in-time model. This is especially important when dealing with perishable medicines, such as the Covid vaccine. But just-in-time only works if you can anticipate demand with sufficient precision. Otherwise, you’ll end up with a warehouse full of stock, or lots of impatient customers.
Demand planning is a trivial task for ML, as long as the algorithm has access to the right data. To make accurate forecasts, the system needs:
- Historical data: Order records from previous years
- Market and economic data: Global trends that may impact demand, including financial trends that might affect consumer confidence
- Consumer data: Data about consumer behavior patterns, with as much detail as possible
- Competitor data: Information about competitors, such as discounts or new products
- Epidemiological data: For pharma demand planning, there’s an additional requirement for data about any disease outbreaks that might impact demand
The more data available, the more accurate the demand planning.
Mechanical failures are one of the biggest problems in supply chain management. Assembly lines get jammed, freezers stop working, trucks get flat tires. All of these things can affect supply chain efficiency, and lead to disaster with perishable medicines.
IoT allows companies to start gathering information about every moving part in their supply chain, from the belts on the assembly line, to the wheels on the trucks. Artificial intelligence can use machine learning techniques to analyze this and identify potential failures. Companies can then send a crew to make repairs before anything goes wrong.
This technique is called predictive maintenance, and it’s already used in many high-tech production facilities. The next challenge is to use predictive maintenance in the transport fleet. For example, on-board software could let the driver know that there’s a risk of a flat tire in the near future.
Procurement is the first step in any supply chain – you need a supply of raw materials before you produce anything. It’s a full-time job in itself, as procurement professionals have to continually weigh up metrics like:
- Payment terms
- Contractual obligations
This is another area where ML can step in, if there is a sufficient supply of data.
Chatbots for ordering and returns
Natural Language Processing (NLP) is another popular branch of AI. NLP is the power behind conversational AIs such as Siri and Alexa, as well as the text-based chat apps that appear on many websites.
Chatbots can perform routine admin functions, including order processing, which means that customers can place orders directly via the website, even if the customer service team is unavailable. Chatbots can also handle returns, and offer the client guidance on the return process.
A human-like interface can really help in times of crisis, such as during a major vaccine distribution. Pharma companies may find themselves interacting with people all over the world with varying technical abilities. Chatbots are a friendly and universally understood way to raise queries and place orders.
The future of the pharma supply chain
2021 will see the biggest test of the pharma supply chain to date. There’s going to be a global rollout of an in-demand perishable medicine. Every dose needs to arrive at its final destination, safe and on time.
What will happen after 2021? Perhaps the biggest change will be the arrival of fully autonomous delivery systems. Thanks to 5G and IoT technology, we might be on the brink of AI-powered trucks on our road, making deliveries without pausing for a break.
The pharma supply chain has some way to go before it reaches that point. There are some outstanding issues, including the final mile problem – the distance between the core supply chain and the end recipients, such as pharmacies and medical clinics. Although it’s possible that AI might even solve this problem, with the use of autonomous delivery drones.