AI for pharmacovigilance

The pharmaceutical industry is undergoing a historic surge in innovation. With thousands of new drugs and therapies entering global markets each year, ensuring patient safety has never been more critical or challenging. As regulatory bodies tighten requirements and global pharmacovigilance standards evolve, companies must process and respond to growing volumes of safety data in real time.

Adverse drug reactions are one of the leading causes of death in hospitalised patients^[i] and have a high economic impact^[ii]on both societies and pharma companies. And yet a vast portion of adverse event (AE), healthcare compliance (HC) and product complaint (PC) information is buried in unstructured free-text records such as CRM notes, emails, call centre transcripts and patient support programmes, to name but a few. Owing to the massive volumes of data involved, these records are difficult to analyse using traditional rule-based or manual methods.

To stay on top of the compliance and regulatory requirements, it’s crucial to implement intelligent, real-time systems that track data from diverse data sources early on and flag them for review. Detecting issues early before the prevalence rate is large helps prevent complications from escalating and turning into major challenges down the road.

Fortunately, the landscape of GenAI- & AI-enabled solutions simplifying this process has greatly evolved. We believe global pharma companies can benefit from the emerging AI & GenAI solutions to drive pharmacovigilance and regulatory compliance, helping them to identify adverse event risks early on and at scale, drive operational efficiencies and anticipate regulatory issues.

Our industry landscape analysis indicates that:

86%

of pharmaceutical representatives believe AI is revolutionising pharmacovigilance

[iii]

12-15%

CAGR for AI in pharmacovigilance expected up to 2030

[iv]

The key challenges we hear from our clients

Some of the key challenges faced by our clients include:

Navigating the increasing volume and complexity of data and integrating diverse data sources

Huge manual effort involved in identifying AEs and PCs, as traditional processes lack automation

Sample-based monitoring approach instead of 100% monitoring of all records, owing to a lack of bandwidth

Latency owing to manual processing that delays safety reporting and increases compliance risks

Increased operational cost owing to licences associated with the legacy system

Increasing compliance and regulatory pressure coming from regulatory agencies who expect faster and more accurate signal detection

Ultimately, these challenges result in a risk of missing adverse events, product complaints and healthcare compliance identification and reporting.

How does the AI & GenAI-powered solution help?

Our AI & GenAI-powered solution accelerates the end-to-end pharmacovigilance value chain by identifying adverse events, product complaints and healthcare compliance situations rapidly and early on – enabling pharma companies to take corrective actions as soon as possible.

With a human-in-the-loop approach, the AI outcomes can be validated by the compliance team to understand the presence of adverse events and product complaints and thereby drive efficiencies. As soon as adverse events and product complaints are identified, the team can automatically retrieve and extract key information such as the patient data, reporter data, product data and adverse events experienced.

Thanks to this human-in-the loop validation, a safety case can then be automatically entered into the safety database. Following on from the validated AI outcomes, the compliance team can suggest and leverage corrective actions to mitigate the risks.

The end-to-end process of adverse event and product complaint risk flagging includes four main steps:

Expected outcomes

We’ve partnered with several global pharma companies to drive transformational outcomes by leveraging AI and GenAI accelerators. Below are some of the key impacts delivered across clients:

70% faster identification of adverse event risks, with real-time detection, increasing reporting accuracy and improved compliance.
50% effort reduction thanks to enhanced automation, reduced manual checks and streamlined workflows.
40-50% reduction in total cost of operations, with a financially sustainable model and lower maintenance and enhancement costs.

How can PwC help you develop the solution?

We have a dedicated team of cross-skilled experts developing our tailor-made AI and GenAI solutions for the pharmacovigilance space. They share a common vision articulated around four pillars:

Our AI- and GenAI-based solution is trained based on the pharma company’s proprietary, secondary and third-party data to identify risks. It flags AE, PC & HC events from the various available data sources, including CRM free text records, the voice of the customer, call centres, voice-to-text records and patient support programmes. Multimodal data source types can be ingested and used to train the models.

The solution leverages an intuitive user interface, with a human-in-the-loop approach enabling reviewers to validate the AI output, provide feedback, suggest corrective and/or preventive action in case of risks or quality control issues – keeping reviewers constantly in control to validate the outcomes.

The solution needs to be designed and developed in alignment with the company’s code of conduct and regulatory requirements as well as being designed to become a scalable product in, for example, AWS/Azure, hosted in the company’s ecosystem and acting as a single source of truth that centrally processes all data sources.

We propose a “one-time” pricing model. This means we ensure the lean development of the solution, co-designing it with our client’s key stakeholders to ensure that needs are responded to and training the solution with relevant data. Once the solution is tested, approved and ready to be launched, our clients completely assume ownership of the solution. There are no licensing costs for the solution and the maintenance costs are low.

Transformative success stories from our clients

Achieving 100% monitoring with AI-driven pharmacovigilance

We helped a global pharma company perform 100% monitoring of all the free text records from its CRM and rolled out the solution globally across all affiliates, leveraging a tailored AI and GenAI solution on the AWS platform.

The client wanted to analyse the free text records in the CRM system capturing interactions with the HCPs (e.g. from meetings, surveys, etc.). Over 500,000 records were captured and legacy solutions were used to perform sample-based monitoring of records to flag AE, PC and HC events.

This led to the incomplete and inaccurate representation of the actual adverse events and product complaints for their marketed drugs.

The existing solution was not scalable and did not meet the business requirements holistically, as it required a huge number of manual checks.

Recognising a situation that potentially results in the risk of missed adverse events, the client aimed to implement a scalable solution to enhance efficiency and the cycle time and quality of its AE, PC & HC processing.

We supported our client by gathering a cross-skilled team to partner with its key stakeholders in the pharma data safety, marketing technologies and healthcare compliance teams to define business and technical requirements for developing the solution. We created a product backlog for phase 1 pilot implementation and a roadmap for phase 2 solution scaling.

We iteratively co-designed and finalised the wireframes for the user interface to enable the reviewers to analyse the risks flagged by AI – ensuring that the human-in-the-loop approach met the needs of the end users.

PwC leveraged AI and GenAI to implement the AE, PC, HC flagging solution in the AWS platform, training the models based on the client’s available data – including the free text records from its CRM system.

We supported the solution roll-out to affiliates to ensure uniform compliance standards and the reporting of AE, PC & HC risks.

After we had developed an AI and GenAI solution tailored to our client’s needs in terms of AE, PC & HC flagging, the client successfully rolled out the solution globally.

Analysis has demonstrated that these markets have achieved a monitoring rate of around 100% for all reported records, finally capturing the free text from the client’s CRM system.

Additionally, the client expects a reduction of some 40-50% in its end-to-end pharmacovigilance cycle time and is ready to scale the solution to remaining markets.

Enabling a 40% operational cost reduction with tailored AI and GenAI solution

We supported the client in streamlining AE intake, triage and case processing and reducing pharmacovigilance operational costs. This helped when it came to anticipating any risks related to upcoming product launches.

Our client recorded over 250,000 adverse events per year from a variety of data sources such as call centres, emails, literature and free text input.

The company anticipated a 10% annual rise in reported adverse events owing to new product launches and increased variety in data sources, raising client concerns that its current system couldn’t handle the growth.

Additionally, the client suffered from many manual touchpoints, resulting in laborious human work and a growing inability to detect risks early on. Ultimately, the workforce complained that it was overwhelmed and unable to perform quality checks.

Keen to ensure the success of upcoming product launches, the client wanted to increase its ability to detect risks detection early on, reduce operational costs and increase user productivity.

We partnered with the client’s pharmacovigilance team to enhance business processes by automating AE intake, triaging and case processing.

Our team focused on user experience to increase adoption and the productivity of individual users. This ensured that the solution was developed using agile methodology and continuous feedback from end users.

The solution’s technical components were co-designed to respond to the case processors’ needs. We developed AI models to interpret the case and determine the seriousness, validity, priority and coding based on the case processor’s input and guidance. We also made sure that the solution was integrated with the safety database.

The intuitive user interface helps users to quickly verify, validate and submit cases to the appropriate spot in the downstream workflow.

After we had developed an AI and GenAI solution tailored to our client’s needs in terms of AE, PC & HC flagging, the client successfully rolled out the solution globally.

Analysis has demonstrated that these markets have achieved a monitoring rate of around 100% for all reported records, finally capturing the free text from the client’s CRM system.

Additionally, the client expects a reduction of some 40-50% in its end-to-end pharmacovigilance cycle time and is ready to scale the solution to remaining markets.

^[i]Bouvy JC, De Bruin ML, Koopmanschap MA. Epidemiology of adverse drug reactions in Europe: a review of recent observational studies. Drug Saf. 2015;38(5):437–53. https://doi.org/10.1007/s40264-015-0281-0

^[ii]Starfield B. Is US health really the best in the world? JAMA. 2000;284(4):483–5. https://doi.org/10.1001/jama.284.4.483

^[iii]https://pharmacovigilance.institute/res/pages/files/hs-artificial-intelligence-in-pharmacovigilance.pdf

^[iv]https://meditechinsights.com/ai-in-pharmacovigilance-market