BC Cancer: Connected Chatbot to Improve Patient Support Custom Case Solution & Analysis
1. Evidence Brief
Financial Metrics
System Scope: BC Cancer manages over 80,000 patients annually across six regional centers.
Funding Model: Publicly funded via Provincial Health Services Authority (PHSA); budget focus is on operational efficiency and cost-avoidance rather than revenue generation.
Resource Allocation: High administrative costs associated with nursing staff and navigators answering repetitive, non-clinical queries.
Operational Facts
Volume: Thousands of monthly inquiries regarding appointment scheduling, side effects, and facility navigation.
Technological Infrastructure: Requirement for integration with existing Electronic Health Records (EHR) and Provincial Health Information Databases.
Service Delivery: Currently relies on telephone-based navigation and in-person consultations, leading to bottlenecks during peak hours.
Regulatory Environment: Governed by the Freedom of Information and Protection of Privacy Act (FIPPA) in British Columbia, requiring strict data residency and encryption.
Stakeholder Positions
Dr. Kim Chi (Chief Medical Officer): Prioritizes clinical accuracy and the reduction of physician burnout.
Sarah Roth (President & CEO): Focused on patient experience and organizational modernization.
Nursing Staff: Express concern regarding workload shifts and the potential for AI to provide incorrect medical advice.
Patients: Report high anxiety levels due to information gaps between appointments.
Information Gaps
Implementation Cost: Total capital expenditure for software licensing and integration is not specified.
Technical Latency: Current uptime and response speed of the existing EHR API for external connections.
Staffing Impact: Specific FTE (Full-Time Equivalent) reduction or reallocation targets are not quantified.
2. Strategic Analysis
Core Strategic Question
How can BC Cancer deploy an automated interface to alleviate administrative pressure without compromising clinical safety or violating provincial privacy mandates?
Structural Analysis
Applying the Jobs-to-be-Done framework reveals two distinct needs. Patients need to reduce anxiety through immediate information access. Staff need to offload low-complexity tasks to focus on acute clinical care. The current manual delivery model fails both by creating a high-friction queue.
Value Chain Analysis indicates the bottleneck exists at the Information Dissemination stage. By automating this, BC Cancer shifts from a reactive to a proactive support model.
Strategic Options
Option
Rationale
Trade-offs
Informational FAQ Bot
Low-risk deployment of static information.
Limited utility; cannot provide personalized patient data.
Connected EHR Chatbot
Provides real-time, personalized appointment and lab data.
High technical complexity and significant privacy risks.
Hybrid Triage Model
AI handles FAQs and routes complex issues to human nurses.
Requires constant monitoring and staff training.
Preliminary Recommendation
Pursue the Connected EHR Chatbot in a phased rollout. Static information bots do not solve the core problem of patient anxiety regarding individual care plans. Integration is the only path to meaningful operational relief.
Phase 2 (Months 4-6): Develop secure API bridge between the chatbot and the EHR database.
Phase 3 (Months 7-9): Pilot at one regional center (e.g., Vancouver) limited to non-clinical scheduling and lab status.
Phase 4 (Month 10+): Province-wide expansion with clinical decision support modules.
Key Constraints
Data Residency: All patient data must remain on Canadian servers, limiting the use of certain global cloud-based AI providers.
Clinical Liability: The risk of the AI misinterpreting a symptom as non-urgent when it requires immediate intervention.
Digital Literacy: The elderly patient demographic may resist the interface, requiring a fallback to traditional phone lines.
Risk-Adjusted Implementation Strategy
Maintain 100% of current nursing navigation staff during the first 6 months of the pilot. Use a shadow-testing period where the AI responses are audited against human responses before being shown to patients. This ensures clinical safety while building the training dataset.
4. Executive Review and BLUF
BLUF
BC Cancer must implement the Connected Chatbot immediately to manage the 80,000-patient load. The current manual navigation model is unsustainable and contributes to staff burnout. By integrating directly with EHR systems, the organization can automate 60% of routine inquiries. Success depends on a strict data-residency strategy and a phased clinical rollout. The primary objective is to convert administrative friction into immediate patient support.
Dangerous Assumption
The analysis assumes that EHR data quality is consistent across all six regional centers. If data entry is non-standardized, the chatbot will provide conflicting or incorrect information to patients, increasing rather than decreasing call volumes.
Unaddressed Risks
Cybersecurity Breach: A single unauthorized access point through the chatbot interface could compromise the entire provincial EHR. Probability: Low; Consequence: Catastrophic.
Algorithm Bias: The AI may perform poorly for patients with English as a second language, a significant demographic in British Columbia. Probability: High; Consequence: Moderate.
Unconsidered Alternative
The team did not evaluate a Decentralized Navigator Model. Instead of a chatbot, BC Cancer could invest in a mobile-optimized patient portal with no AI interface. This would provide the same data access with lower clinical liability and zero risk of AI-generated misinformation.