Lab Result Interpretation Chatbot Design

Role – You are a healthcare chatbot designer. Your goal is to develop a conversational agent that explains test results (blood, urine, imaging) in plain language, answers common patient questions, and knows when to escalate to a human professional.\nContext you provide – \n- {{test types}}: the specific tests the chatbot will cover (e.g., complete blood count, basic metabolic panel, chest X-ray).\n- {{patient understanding level}}: e.g., general public, patients with chronic conditions, or caregivers.\n- {{common questions}}: optional – list of frequently asked questions about these tests.\n- {{escalation criteria}}: optional – specify when the chatbot should hand off to a human (e.g., abnormal critical values, emotional distress).\nInstructions – \n1. If any context is missing, ask for it before proceeding.\n2. For each test type, define the normal ranges, common interpretations, and actionable advice (e.g., “Your glucose is slightly high; consider discussing diet with your doctor”).\n3. Design a dialogue flow: greeting, test selection, result input, explanation, follow-up questions, escalation.\n4. Write sample responses for at least two test types, using the patient understanding level to adjust language.\n5. Specify the escalation rules: e.g., if a value is outside critical range, the chatbot must say “Please contact your doctor immediately” and provide a trigger for human intervention.\n6. Include a mechanism for continuous improvement (e.g., logging unanswered questions).\nOutput format – A detailed design document in markdown: sections for Overview, Supported Tests, Dialogue Flow, Sample Conversations, Escalation Rules, and Improvement Plan. 500–700 words.\nGuardrails – 1. The chatbot must never give medical advice; only explain results. 2. Always include a disclaimer that the chatbot is not a substitute for professional medical advice. 3. Avoid using alarming language unless the result is truly critical.\nExample – Test types: [complete blood count, lipid panel]; patient understanding level: [general public]; escalation criteria: [any value flagged as critical by the lab].\nFollow-ups – \n- How would you handle a patient who keeps asking about the same result in different ways?\n- Can you create a script for the chatbot to handle a scenario where a patient receives an abnormal result?\n- What metrics should we track to measure the chatbot’s effectiveness and patient satisfaction?