Leveraging AI and Technology Now for Multilingual Safety Management

Managing global safety reporting is becoming increasingly complex with ever-changing regulatory requirements, increases in data volumes, data coming via new channels (such as social media), and the expansion of countries where drugs are being marketed. As most countries require safety data to be submitted in English, market authorization holders need to turn around translations of content, often within a 24–48 hour window. With medium and large pharmas handling tens of thousands of safety-related cases each year, the requirement for translation can have a tremendous cost and time impact on compliant safety reporting and running an efficient drug safety system.

The rise of AI has brought about new applications and technology to help streamline multilingual safety reporting. Here are some ways to make your process more efficient:

Implement a translation workflow management system: Not all safety content is created equal, and with potentially thousands of documents in various languages, managing content through the appropriate workflows is a daunting task. Also, with the advent of GDPR, managing content in a secure way is even more pressing. Leading pharmas are leveraging the power of technology to implement a secure structure for routing content through the correct translation processes using the appropriate linguists and/or language providers; tracking and auditing key KPIs such as delivery timelines, spend, etc.; routing content through suitable review processes (if required); and centralizing all language assets for the purposes of consistency and efficiency. Recently a top 10 pharmaceutical company reduced their internal safety team’s project management time for translation by over 75% as a result of leveraging technology for translation workflow management.

Look for partners able to integrate with existing safety systems: Nothing is more inefficient than pulling content from one system just to put it in another. Organizations are increasingly looking at ways they can integrate back-end systems (case intake, PSMF, etc.). When considering translation workflow management systems, it’s important to consider if/how it can integrate with key safety systems already in place.

Leverage optical character recognition (OCR) solutions: Safety content has not always been the most “translation friendly”—old-fashioned adverse events used to be faxes of a doctor’s shorthand notes, and local literature was often ripped from a journal and scanned/emailed. In order to translate and reformat into a submission-ready document, the content must be put in an editable format. The cost to reformat in a live, translatable format can be up to 35% of the total cost for translation.

OCR solutions allow for the conversion of uneditable, scanned, handwritten text into editable formats such as Microsoft Word, which can help significantly reduce the time and cost of reformatting for translation, as well as allow for use of other translation efficiencies, such as translation memory and AI.

Start applying AI for translation: With AI for translation becoming stronger and stronger, companies are employing it as a tool to help accelerate safety reporting and reduce costs. AI is being successfully applied for the following instances:

• Case intake level: Frontline staff enter call details in safety systems in local language. APIs between back-end safety systems and AI engines allow for real-time translation of call details, resulting in less time/error with frontline staff completing translation.
• Case triage: To understand the severity of a case, safety departments are using AI to get general information and route through the appropriate channels for reporting.
• Voicemails/audio content: Voice-to-text technology paired with AI allows for real-time insight into the nature of patient and physician voice messages.
• Safety content localization for reporting to national competent authority: AI paired with professional, therapeutic expert post-editors is being applied to translate content faster and cheaper while also providing high-quality, certifiable safety content.

Confidentiality and the ability to build in professional human post-editors to ensure quality are important considerations when evaluating AI partners for translation.

Automate your redaction process: It is estimated that the time to redact a patient case can be anywhere from 30 minutes to one hour. If a company has 10,000 cases a year, that’s the equivalent of over 5,000 hours (or more than 2 years’ worth of work for one full-time employee). As a human process, there is inherent risk of omission of PII redaction. AI and technology are being deployed to help support the automated process of PII redaction paired with human review. It has been estimated that through deploying automated redaction resources, the time required for this process is cut in half while risk of error being <.02%.

Use digital monitoring solutions built to manage multilingual content: Market authorization holders have a responsibility to document and report adverse events coming through digital media outlets, and life sciences organizations are investing heavily in monitoring solutions. When exploring media monitoring solutions, it’s important to ensure you’re capturing global content in non-English forums.

For consultation on how to apply AI and technology to improve the quality and efficiency of your multilingual workflow, please contact lifesci@transperfect.com.