Configuration & Monitoring for Chatbots aimed for External Communication
While the majority of requirements for chatbots is similar in terms of technology and content management, chatbots aimed for external communication have to respond to a different set of regulatory requirements with regards to its respective environment.
In order to ensure that:
- Chatbot responses are fully controlled by Bayer, chatbot responses shall use approved standard response and frequently asked questions (FAQ) response scripts.
- Chatbots comply with requirements for digital external communication material, relevant Pharmaceuticals and Consumer Health QSDs shall be followed, as applicable.
- Users are fully informed about how the chatbot operations are managed, all Legal Notices (Conditions of use, Privacy Statement, Imprint) need to be implemented and approved according to local regulations. All three legal notice templates shall be modified based on chatbot characteristics and local law/regulations.
- Users are aware of how to report Adverse Events or Product Technical Complaints, chatbots need to include a safety and quality reporting disclaimer. A template can be found in Bayer Corporate Policy No 2095 / Product Safety and Quality: Reporting Obligations of Employees. This can be locally adapted to comply with local rules and applicable legislation.
- Adverse Event and Product Technical Compliant reporting requirements are fulfilled, chatbot conversations should be recorded, if technically possible. Users shall be informed about recording in the legal notice.
Rules for Chatbot Monitoring
As chatbot conversations are considered as user-generated content, a regular monitoring process needs to be in place for all chatbots in the Healthcare space.
- Recordings shall be screened daily for any potential Adverse Event or Complaint
- Findings shall be reported as quickly as possible to Bayer Pharmacovigilance within 24 hours (for Bayer employees) or within 1 business day (for external vendors).
- Receipt verification of any reported Adverse Event or Product Technical Complaint is in place, therefore a local quality check process (e.g. reconciliation based on chatbot conversation recordings) is implemented.