There may be no greater goal for artificial intelligence than to contribute to the betterment of the human condition by optimizing and personalizing therapies or care plans ― or by curing deadly diseases altogether.
The power to understand volumes of data can play a critical role in accelerating life sciences and medical devices R&D, and the deployment and support of digital solutions that serve as extensions of existing drugs, therapies or devices, or that enable digital health care communities. In a recent report on artificial intelligence (AI) trends through 2023, Research and Markets
“Clinical trials are data-intensive tasks that need continuous patient monitoring and generate vast sets of data every day. Subjecting these data sets to intelligent AI algorithms can help the researchers to filter meaningful correlation even between loosely coupled data.”
The report said that “the tremendous demand for AI in life sciences applications, such as drug discovery and patient monitoring, is opening up new opportunities,” but it added that organizations may be hesitant about solutions that reduce jobs or come with high initial costs.
Executives in life sciences believe that tomorrow’s industry leaders will be defined not by AI itself but rather by “how effectively companies use AI.”
Organizations are also wary about the escalating buzz around AI. In a recent survey, Emerj
found that executives in life sciences believe that tomorrow’s industry leaders will be defined not by AI itself but rather by “how effectively companies use AI.”
So, as your organization considers its next ― or first ― venture into AI, how can you ensure that it effectively accelerates and streamlines your work?
Start with the process
AI is in a period of rapid growth and development, with many solutions pushing the boundaries of the possible. Some are successful, and some are not. So, it can be hard to clarify exactly where AI has succeeded and where it’s likely to succeed for your organization.
Among Grant Thornton clients that have successfully implemented AI, one common thread is that they began by looking for optimization. Rather than asking “How can we use this new technology?” they asked, “What processes within our organization provide the biggest opportunity for optimization?”
To help identify these opportunities, some organizations created teams within their finance divisions. These teams look holistically across the organization to find processes where optimization could provide business benefits. Fully understanding and streamlining a process is the first step to automating the process, and optimization can achieve business benefits apart from new technology.
Even if an emerging technology solution is sold as not requiring process changes, a process evaluation and an organizational technology evaluation can help create a strong business case for a technology implementation.
As Roy Nicholson
, principal, Grant Thornton Digital Transformation and Management leader noted: “We do see some organizations that don’t have a solid business case but look immediately to trying to fit technology into operations. We strongly encourage the development of a process assessment within a function, looking for opportunities where different intelligent automation technologies can be applied and then building a business case that can be agreed to and measured against.”
We do see some organizations that don’t have a solid business case but look immediately to trying to fit technology into operations. We strongly encourage the development of a process assessment with in a function, looking for opportunities where different intelligent automation technologies can be applied and then building a business case that can be agreed to and measured against.
Roy Nicholson, Principal
Leader, Digital Transformation and Management
Grant Thornton clients in the life sciences field have found a number of potential use cases for automation, machine learning and other AI capabilities:
- Automated lab testing in R&D, which allowed one organization to significantly reduce its staff of lab technicians
- Predictive pilot solutions that identify which chronic disease patients a physician is likely to see soon in order to help field reps present any relevant treatment alternatives
- Call center transcription analysis that helps identify trending key words and questions for training call center and medical staff (This functionality has become comparatively mature through development outside of life sciences.)
- Sales forecasting, especially for chronic disease treatments, analyzing a combination of data from patients, prescriptions and physicians
When an organization has zeroed in on some opportunities for optimization, the next question is “How?” With a host of AI vendors and consultant agencies, life sciences organizations don’t need to go it alone — but they do need to balance costs and develop key in-house AI expertise.
Build expertise where it matters
Given the dynamic and complex nature of AI, most organizations can’t afford to wait while their in-house expertise matures. But organizations miss out on important value and create a dependency if they use only external AI agencies.
With AI models available now, organizations can design use cases and formulate solutions on their own. So, what types of expertise matter for in-house teams? How can organizations establish the right balance of external expertise with internal insight on proprietary information?
- In-house leadership: At a high level, life sciences organizations have found that AI efforts should be led by internal resources who can help capture and integrate the unique lessons learned and proprietary information that needs to be passed on to future projects.
- External expertise: External consultants, teams and tools can help organizations get started — especially in implementing new solutions, enterprise resource planning (ERP) systems, and an enterprise platform or data warehouse across the organization.
Internal resources can provide the best insight into an organization’s data. And it’s important for internal resources to understand how the organization feeds data to its new solutions, how the master data is managed, and other structural aspects for a data platform. A central enterprise data platform is a powerful starting place for any enterprise data effort; it helps provide a standard for the tools and solutions that ultimately deliver business value.
Three general models:
Data science model
Managed service model
As an organization considers how much AI and machine learning expertise it should develop internally, it can consider three general models:
- Proprietary platform model: An organization can build its own team of technologists and data scientists who are entirely responsible to build solutions on the organization’s unique data in order to achieve unique AI goals.
- Data science model: An organization can acquire an AI platform, hiring a supporting team that includes one or more technologists and data scientists who help the organization use the platform to achieve its goals.
- Managed service model: An organization can call upon a cloud machine learning capability or other service-based AI capabilities, providing data for external data scientists to process and report upon. This tier is more common for initial pilot projects.
Once an organization aligns itself with an AI approach, it needs to choose the tools that will best integrate with users and processes.
Find a platform and flexible tools
Behind any successful AI solution is a source (or sources) of data and a set of tools that deliver the key capabilities. To deliver those capabilities, you need the tools to perform three types of work:
- Mining: The data mining, or the plumbing, extracts data from various sources and puts it into an environment ready for analytic processing.
- Analytics: The analysis applies the logic, calculations, hierarchies and algorithms that are specific for analytics (as opposed to transaction processing).
- Presentation: The presentation tells a story or presents a visualization that informs business decision-making, transformative action or even compliance reporting.
So, what tools are right for your team? The majority of organizations already have access to a range of mining, analytics and presentation solutions. For life sciences organizations, some tools have proven particularly valuable:
Ensure quality with governance
- Birst – Some organizations use Birst business intelligence software to help mine volumes of financial, prescription and clinical data, as well as input from dialysis machines and other remote devices. While it can take time to determine the best way to apply this analysis, the analysis effort can quickly start to identify places where data is missing or needs to be standardized. Some life sciences organizations have found that the Birst reporting interface is accessible enough for use by C-level executives, clinicians and other direct end users.
- BlackLine – Some organizations have found that this cloud-based accounting tool can essentially automate reconciliation processes to save valuable time.
- Microsoft Excel – While some of the top platforms steer users away from working in Excel, it remains the top business intelligence tool — and most platforms do offer some option to connect to Excel. Or, Excel can connect to the data model so that users can access the enterprise data warehouse subject areas for dynamic reporting capabilities.
- NetSuite – Life sciences companies have found NetSuite’s cloud-based ERP solutions useful for processing financial data. While your data platform and tools can play a critical role in the success of your AI efforts, it’s critical to remember that your processes for data generation are the foundation of your AI solution’s quality. It’s important to have tight controls and process discipline to help your solution provide accurate real-time data to the field.
- Power BI – This Microsoft platform allows organizations to create interactive dashboards to analyze data for decision-making.”
- Qlik – Some life sciences organizations have found that Qlik can analyze physician data to better inform sales reps in advance of their calls, and help to track in-market performance, physician behaviors and more.
- Salesforce – To help field staff track opportunities and forecast business or inventory, Salesforce has been a proven tool for some teams.
- Tableau – Some organizations have used Tableau for financial and marketing analytics, identifying when to combine different datasets and providing an interface readily accessible for executives.
- WeR.ai – Providing use case-specific AI solutions that are supported via an AI as-a-service subscription agreement. WeR.ai’s solutions can be deployed in a fraction of the time and for a fraction of the cost of others.
Infrastructure may not play an essential role in an initial AI test or pilot solution. But the background processes that ensure data governance set the foundation for long-term value and success in your organization’s AI efforts.
It’s easy to focus on the goal at the end of an AI solution, but an enterprise strategy can help teams ensure that they work together to build value in a central data standard that will sustain business value for present and future solutions. The enterprise strategy needs to clearly articulate the processes and procedures that govern data — and it falls squarely within the realm of an organization’s internal expertise.
Process optimization, data management and governance may not grab many headlines, but they are the foundation of AI solutions that accelerate the search for tomorrow’s critical treatments and cures.
A medical device use case
A large, global medical devices client has to submit what are essentially "audits" of their devices' performance to European Union authorities three times per year for each of their 30 devices ― a total of 90 reports. Each report must include clinical evidence related to that specific device from a corpus of 6,000–7,000 medical publications. This requires skilled medical reviewers to an enormous amount of time reviewing numerous documents to find out which documents are relevant.
Solutions company WeR.ai created a device-specific tool to 1) read all of the documents, 2) identify which documents are relevant, 3) rank those documents by relevance, 4) identify which content triggered the relevance score and 5) generate an Excel for the medical reviewer team to prioritize their review efforts.
Principal; Leader, Digital Transformation and Management Group
+1 408 346 4397
Senior Manager, Digital Transformation & Management
+1 858 704 8082