Unpacking AI and automation in life sciences: Four impactful scenarios

In 1628, Sweden was brimming with anticipation over the launch of their latest warship, the Vasa. They’d invested three full years into its creation, hoping it would establish their superiority on the seas. But minutes after the Vasa embarked on its maiden voyage, the ship tragically sank in ‌Stockholm Harbor.

What happened? There were two primary factors that contributed: a misunderstanding of the new technology and a series of design errors that led to the ship being woefully unbalanced with insufficient reinforcement to the hull and keel. The result was an almost guaranteed disaster.

Modern AI technology presents a “cargo hold” full of transformational benefits for life sciences organizations today—but it also presents significant risk. Unless organizations are careful, they can overload the "ship" with fancy capabilities, while neglecting appropriate governance. While AI is the “brain” for your business, the technology needs the “muscles” of automation. A Vasa-level failure is no longer a concern with the right combination of AI plus automation that leverages enterprise level guardrails including governance, controls and workflows.

That’s where an intelligent automation platform that combines automation plus Generative AI and Specialized AI models, comes in. Using these technologies together is often called AI-powered automation.

Automation is the path to realizing true value from AI today (not three years from now).

Four ways to put AI to work in life sciences

Let’s take a look at four life sciences use cases that AI and automation can enable together.

1. Pharmacovigilance (PV) process

PV is a systematic process used to collect, monitor, and analyze data—usually through a complaint—on the safety of medicines. Complaints can come from a variety of sources, and AI-enabled automation can analyze the data to determine if a submission is required. AI-powered automation can then update the regulatory document and submit the form to the appropriate regulatory agency either directly or after the review of an employee. The intertwining of the automation with AI capabilities creates a more intuitive and efficient experience for the user.

As a result, automating PV in this way is more valuable and user friendly.

2. Patient registration

Many newer therapies are personalized based on an individual’s genetic makeup and require a screening and/or diagnostic test to ensure the patient is eligible to take the treatment. Automation can coordinate the appointment for the patient, ensure the machine required for the diagnostic is available at a specific time, and guide the patient and/or employee through the questionnaire that is based on a decision tree approach to the questions.

Specialized AI underlies this process, with models for communications and document understanding to process patient requests and extract the data needed for automation. In combination with automation, Generative AI capabilities can then be used to generate any number of follow-up documents that summarize the results of the tests, confirm the appointments, and provide any other relevant information for the patient.

3. Regulatory submissions (including NDAs and BLAs)

Earlier, we explored the potential for AI to update the information required for regulatory reporting across a variety of operational and clinical use cases. As discussed in a recent white paper, one of the areas currently under analysis is the potential to point an AI large language model (LLM) at ‌clinical trial data. AI can make the first pass at creating the narrative for a new drug application (NDA) or Biologic License Application (BLA) for a product submission to the United States (U.S.) Food and Drug Administration (FDA). It is still early in the evolution of this technology, but the LLM capabilities align with the updating of the new drug submission forms.

To further strengthen guardrails around AI, the UiPath Platform includes the UiPath AI Trust Layer. The AI Trust Layer enables transparency, trust, and control over the interactions between Generative AI features across the UiPath Platform, your organizational data, and third-party LLMs. This includes the ability to leverage usage auditing and cost control, Gen AI governance, harmful content filtering, personally identifiable information (PII) and sensitive data masking, and LLM gateway monitoring.

Typically, the process required to analyze, compile, and document the data required for any new product submission could take many months to complete. The data must be organized and assembled in a clear and comprehensive manner for regulatory review. Gen AI could generate the data in a few hours, and automation can let researchers know when the data is ready to be reviewed and edited for accuracy and completeness, rather than having to draft the report from a blank sheet of paper.

Initial estimates indicate that organizations could save 10 to 20 weeks by leveraging AI and automation to prepare the required submissions. At a revenue estimate of between $1 million and $3 million per day for a new drug, this could provide a significant benefit of between $70 million and $420 million in accelerated revenue for global pharmaceuticals.

4. Query standard operating procedure (SOP) documentation

One of the most remarkable aspects of AI is its ability to review vast amounts of data. Many life sciences organizations have extensive libraries of SOPs that define the steps required to successfully execute a process.

For example, any regulated manufacturing facility will have thousands of pages of documentation that precisely detail the specific processes and procedures required to manufacture a pharmaceutical product. The documentation is extensive, and organizations will be required to manage all updates of that documentation to ensure the current procedures are not only documented but integrated into the manufacturing process.

AI provides a tremendous opportunity to allow users to query the SOP database and locate the specific regulation for review. This provides the pharmaceutical company the ability to demonstrate the full understanding of the regulation while highlighting the documentation in real time if they are ever audited by a regulator. Further, the user could use a voice or text-based inquiry to receive a response in real time.

How one global pharma company is using AI-enabled automation

Picture this: a global pharmaceutical giant aims to fuel its digital transformation by bringing AI-powered automation onboard. Their goal? To not just smooth out back-end processes, but also to take tedious manual tasks off the plates of their global staff.

What they needed was an end-to-end platform that could support a strong citizen developer program and a wide-scale usage of both unattended and attended robots.

Their pick? The UiPath Business Automation Platform.

Along with providing enterprise-grade capabilities across the AI-powered automation platform to streamline core behind-the-scenes processes like invoicing, UiPath offers comprehensive training and engagement programs. The UiPath Platform also provides user-friendly, low-code tools, consistent governance, and a great user/developer experience that amped up their organization-wide citizen developer initiative.

The global pharmaceutical company’s results so far have been nothing short of impressive. They've reaped over $35 million in annual automation savings across more than 350 processes, over 200 developers, and more than 250 robots.

More ways automation empowers AI in life sciences

The combination of automation and AI will benefit all aspects of the enterprise, including research and development (R&D), finance, manufacturing, supply chain, commercialization, and compliance.

With AI-powered automation, global life sciences organizations will reduce cycle times, improve time to market, end human errors, and accelerate time to revenue.

See how other life sciences organizations are realizing more value combining AI and automation. You have complimentary access to our virtual Healthcare and Life Sciences Summit (now available on demand).