The Challenges in Data Harmonisation in the smart lab industry

Pharma IQ spoke with Richard Caron, Associate Director Global MQIT Eli Lilly, about the challenges the Smart Lab industry is facing when it comes to data harmonisation and how future technologies will help Pharma keep its data more secure. Richard guides us through the major challenges when it comes to managing large quantities of data and how AI may be able to fill the gap in the industry. He also leaves us with a few tips to ensure data integrity and security.

What are some of the biggest challenges currently facing the Smart Lab industry when it comes to laboratory informatics data management and systems visibility?

The biggest challenges we’re currently seeing is data harmonisation. Many methods that we’ve implemented for testing our products aren’t as standardised as they could be. For example, a concentration could be done in many different ways across one organisation- if you take a look at our sites we have numerous different concentrations, and when we try to look at the data that is produced to try to do some informatics or metrics or any types of visualisation with that data, it’s very difficult to compare as each site has its different methods. It’s like comparing apples to apples when the meaning of ‘apples’ is different at every site.

Another major challenge is the sheer amount of data that we’re collecting with all of the IoTs that are being put in place into our laboratories. We’re capturing a lot of data and there’s a lot of noise there. Trying to pull the trends and pieces of truth from that noise is challenging and the data that is a true signal is difficult for us to locate.

Another challenge certain labs are experiencing is that they aren’t getting data that’s surrounding the instruments that we’re running. The data is being collected, but we’re just not pulling it up into our data lakes because we don’t currently have access to it.

This data issue leads into the IoT issuewe’re talking about instruments that cost nearly half a million dollars. You want to get your money’s worth out of them, so these instruments usually last between 10, 15, 20 years, if not longer, whilst technology is moving rapidly. So, there’s a gap there that we’re challenged with because the instruments aren’t as up-to-date as the technology. 

Finally, another issue is that much of the laboratory software that we’re putting in place is not very user-friendly. The software doesn’t have usercentricity at its forefront; it’s more utilitarian and it’s not as easy to use as what labs are used to. This is especially the case as we bring new generations into the industry who are used to userfriendly technology such as iPhones and social media, where everything is simple and straightforward. Our laboratory softwares are not.

Finally, another issue is that much of the laboratory software that we’re putting in place is not very user-friendly. The software doesn’t have usercentricity at its forefront; it’s more utilitarian and it’s not as easy to use as what labs are used to. This is especially the case as we bring new generations into the industry who are used to userfriendly technology such as iPhones and social media, where everything is simple and straightforward. Our laboratory softwares are not.

What do you believe that the role of the solution provider and the role “The biggest challenges we’re currently seeing, is data harmonisation. Many methods that we’ve implemented for testing our products aren’t as standardised as they could be. “laboratory-informatics.iqpc.co.uk of the pharmaceutical company, which is the end-user, looks like in overcoming these challenges? How can they be overcome?

A lot of it has to do with Pharma. We need to be more vocal and clearer about the requirements that we need, those that we actually have, and not get distracted by the buzzwords and the shiny objects that we tend to see. I don’t think that the clarity and voice is there yet. We’re starting to voice these requirements, but it’s only beginning.

The vendors also need to consider the user experience. Right now, they’re very focused on the analytics and data side, but they’re missing the piece as to how the user is going to interact with it.

With younger generations being accustomed to user friendliness this issue will only be exacerbated further unless the vendors have an opportunity to get there. 

I also haven’t seen many vendors consider open source software and cloud technologies. Vendors are still very much into the proprietary software, building their own code and using their own developers instead of using tools and technologies that already exist that could make their lives (and ours) a lot easier. Right now, a lot of the instrument controllers that we’re looking at are still based on 1990s technologies. They’re not building towards the future. 

In the next few years where do you see technology needing to develop to fill the current industry need or the gap that you mentioned?

It depends on which function you’re looking at. The pharma industry is a huge organisation. We’re talking straight from development, where we have to have a lot of creativity, and that’s where I could see a lot of the machine learning and AI becoming useful.

But in manufacturing, when you have to pump out product and it has to be very consistent, and it’s highly regulated, I’d argue that simplicity would be more important in that stage. So, it depends on which area you’re looking at whether AI and machine learning would hinder or help the process.

As far as customisation goes, the industry was moving away from it, but now I think that based on the trends we’re seeing with microservices and cloud technologies, we’re getting back into customisation, but customisation of a different kind. It’s more a user-centric and process focused customisation, rather than customisation on a broader base.

What do you think the impact of future technologies, AI, machine learning and IoT would be in shaping informatics development?

From the manufacturing perspective, we’re going to have to look more closely at the harmonisation of our legacy data to be able to provide the same context and make sure it’s consistent. As the regulatory agencies become more familiar and comfortable with the output of these technologies it will allow us to be able to move more quickly in our product lifecycles. Moving from development to commercialisation, I think a lot of these technologies are going to speed that up so our timeline will be reduced drastically. But first, we’ll have to make sure that the regulatory industries become comfortable with that.

New technologies will also give us better control and visibility into our whole supply chain as the Pharmas are consolidating and expanding, they’re distributing worldwide, and these types of technologies are going to help us see what our supply chain is like and gain better control of it.

This is crucial especially as we’re moving towards biologics where the time to the market is critical because you’re dealing with live cells and things that are expiring, rather than a pill, Technology is going to have a huge impact on the process of understanding where our product needs to be and when.

How is Eli Lilly building their informatics capabilities within the manufacturing space? Does this come from adopting R&D technologies or are the needs different?

No, the needs are fairly different as is the case throughout the whole industry, and especially with large pharma. The manufacturing space has different constraints and different drivers than R&D and development. Eli Lilly is putting the building blocks in place to be able to leverage our data as much as we can with the informatics tools that we’re putting in place.

It’s a huge organisation, so there are several groups that are working on it, but we’ve put governance in place to be able to make sure that we’re heading in the right direction or at least, we’re all heading in the same direction.

When it comes to maintaining data integrity, how do you see new systems implementation and smart technology impacting this?

From my perspective I say that it’s going to strengthen data integrity because we’re going to be able to get closer to the source data. Then there will be a better handle on the chain of custody of that data as it moves through its lifecycle. It’ll be easier for us to be able to keep that in place, so even if we’re putting data into the cloud, so long as we understand that we can control the access to that data, data integrity will be strengthened.

However, that’s just my opinion. There are others who feel that new systems are weakening our control over data because they don’t have physical control over the data because it’s not in a box. They feel that it’s not as controlled as it’s not tangible. Those are the two conflicting schools of thought surrounding data control right now.

What would be your top three tips to ensure data integrity and systems security while implementing these advanced technologies?

Validation requirements, as validating our data is still the law here in the US. The FDA regulates our data, so we have to validate these systems, regardless of where or how they sit. I would say have your group that’s in charge of computer systems validation involved as early on as possible in your project as you’re moving these technologies forward. It’s important for them to provide their input and build the security and the data integrity into the system as you’re deploying it. 

I would say have your group that’s in charge of computer systems validation involved as early on as possible in your project as you’re moving these technologies forward. It’s important for them to provide their input and build the security and the data integrity into the system as you’re deploying it. 

Having the quality organisation involved early is important too. The later they are, the harder it is for them to be able to defend the system or put in the correct controls to defend the system.

Ultimately, data integrity is about control, so to be able to demonstrate that you have control over your data and that you understand where it came from and what the data flow is, is important. If you can demonstrate that and support it, then you should be all right. A third tip is try to avoid data duplication. Whenever possible always use the source data. Data duplication can lead to some serious data integrity issues, especially if you’re not maintaining it properly and then your data is going to start drifting.