The Challenges of Safety Risk Detection: The Automation Solution

As the world emerges from the COVID-19 pandemic, pharmacovigilance (PV) teams face the same attrition and staffing crisis as other healthcare environments. However, timely detection of adverse events (AEs) remains an ongoing priority. If there’s anything we learned from the pandemic, it’s the critical need to analyze patient outcome data in real time to stay informed on the safety and efficacy of treatments in the real world.

Using artificial intelligence (AI) to automate data analysis can relieve some of the burden on PV teams—without the risk of compromising work quality.

Challenges of AE identification and capture methods

Marie Flanagan

Current methods used to detect adverse events create challenges for the PV community. Manual processes as well as voluntary reporting by patients, HCPs, pharmaceutical companies, EHR reviews, and patient interviews are all subject to under-reporting. The PV community began using AI and machine learning some years ago to combat these issues, but the trend has accelerated in the past two years.

This surge in uptake is partly due to the rise in telehealth and omnichannel reporting. For example, social media conversations involving COVID-19 vaccines burgeoned during the pandemic. In fact, more than 80% of customer interactions could move to digital in the future. With 50 million interactions per week about the vaccines recorded over a four-month period, the sheer volume of data shows just how much information is out there that requires analyzing.

Healthcare already generates some 30 percent of all data, and the volume grows exponentially every day. Global data is projected to total 181 zettabytes by 2025, an increase of 181.9 percent over five years. There are a couple of important points buried in that statistic:

• Collating and analyzing existing data manually for the purpose of AE detection is expensive and imperfect.
• With the increase in telemedicine during the pandemic, data volumes will continue to grow and outpace hiring capabilities. For example, more organizations are combining human, AI-enabled, and digital customer support across all of their channels, a change that 67% of surveyed executives expect to stick post pandemic.
• Capturing information about potential AEs across all the possible forms of communication is inherently complex for humans and will become increasingly challenging.

Non-automated processes and a lack of sufficient integration between clinical trial management systems, safety systems, data management systems, product performance databases, social and digital data also lead to duplication and oversight, resulting in a wholly inadequate outcome for patient safety.

Narrowing the gap through automation

Automating AE detection through a combination of human and artificial intelligence offers the solution. These days, patients discuss prescription drug reactions on websites, forums, and social media channels, or they report them directly online or by telephone to patient support programs or medical providers. Wearable technologies monitor users’ physical responses and return data via their apps.

Despite these methods, regulatory agencies place the onus on pharmaceutical companies to monitor unstructured data sources for risk associated with their products. A digital solution using natural language processing (NLP) and voice detection technology that supports multiple languages provides a means of overcoming these challenges, enabling the PV community to detect AEs across platforms. Patients can report AEs directly in their own language using a website or mobile device and submit it straight to the safety system using end-to-end encryption. Data from CRM reports and chatbots also get added securely to the system.

Such a solution narrows the distance between the person reporting the AE and the PV reviewers and clears the runway of obstacles such as a high number of manual reviews.

The benefits of automated AE detection

The automation of the AE detection process offers substantive benefits for pharmaceutical companies, HCPs, and patients. These include:

• Better patient engagement: The ability to collect information directly from patients increases their level of engagement.
• Reduced regulatory risk: Artificial Intelligence has shown 99% precision in finding potential safety events. This factor unlocks real value by increasing the success rate of AE detection, avoiding missing events, and improving compliance with regulators.
• Fewer manual reviews: Advanced transcription and speech analysis technologies detect AEs with a high degree of accuracy, even from sources such as call center data. Approximately 60% of source data (including calls) require no manual review, because speech analysis technology can zero in on the critical parts of calls to improve understanding and reduce the number of reviews necessary.
• A streamlined process: By integrating existing AE detection processes under a single method, pharmaceutical providers can streamline the entire process enterprise-wide. The optimized workflow increases the speed and quality of AE detection and supports the rapid clearance of call center backlogs and digital data.
• More cost efficiencies: The use of AI, NLP, and bulk processing capabilities reduces the number of human staff required, optimizing the use of resources, and creating better cost efficiencies.
• Enhanced job satisfaction: Reducing workloads allows PV staff to focus their attention on more valuable activities, lowering stress and enhancing their work-life balance and job satisfaction. The knock-on effect of this outcome is likely to be fewer resignations and a slow-down in the staffing crisis facing healthcare.

‘Leaning in’ to advanced technology

Over the next few years, the pharmaceutical industry foresees growing quantities of data, increased regulatory compliance, and more complex reporting demands. Advancements in safety technology through the combination of human and machine intelligence will become essential for accurately identifying AEs.

Companies that lean in and embrace the opportunities afforded by technology to improve and maintain detection processes will ultimately provide better pharmacovigilance and ensure safer treatments for patients.

Marie Flanagan is Director, Offering Management, Vigilance Detect at IQVIA.