India has 1.2 million doctors for a population of 1.4 billion — roughly 1 doctor per 1,160 people, well below the WHO recommendation of 1 per 1,000. Administrative burden makes this worse. Documentation, patient communication, reporting, and compliance paperwork consume hours of clinical time that could go to patients. These 20 prompts target the administrative and communication tasks where AI can genuinely help.
Important: These prompts are productivity tools for administrative and communication tasks. All clinical content — diagnoses, treatment recommendations, prescriptions — must be reviewed, verified, and approved by a qualified medical professional before use. Do not use AI-generated clinical content without physician review. Nothing here constitutes medical advice.
Clinical documentation prompts
1. Discharge summary drafting
Draft a discharge summary for this patient based on the following clinical notes. Format:
- Reason for admission
- Hospital course (brief chronological narrative)
- Procedures performed
- Medications at discharge (drug name, dose, frequency, duration)
- Follow-up instructions (appointment, investigations, warning signs)
Clinical notes: [PASTE NOTES HERE]
Note: This is a draft for physician review and signature. Do not finalize or share with the patient without attending physician review and sign-off.
When to use: after complex inpatient admissions where the treating team's notes are thorough but unstructured. Claude will produce a readable draft in 30 seconds; the physician review and edit takes 2-3 minutes instead of 15.
What requires physician attention: the "hospital course" narrative and the discharge medication list. Both need verification against the actual chart.
2. OPD consultation note structuring
Convert these consultation notes into structured SOAP format.
- Subjective: Chief complaint, history of presenting illness, relevant past history, medications, allergies
- Objective: Vitals, examination findings, relevant investigations
- Assessment: Clinical impression (this section must be completed by the physician)
- Plan: Investigations ordered, medications, referrals, follow-up
Raw notes: [PASTE DICTATED OR TYPED NOTES HERE]
Leave the Assessment section with [PHYSICIAN TO COMPLETE] — do not infer a diagnosis.
Important: never use Claude to generate the Assessment (diagnosis) section. Structure the other three sections, let the physician complete Assessment.
3. Referral letter generator
Draft a referral letter from [Dr. Name, Designation, Department, Hospital] to [Specialist Name / Hospital Name] for this patient.
Reason for referral: [CLINICAL REASON] Relevant clinical history: [HISTORY] Current medications: [MEDICATION LIST] Recent investigations: [KEY FINDINGS]
Include:
- Brief clinical summary (2-3 sentences)
- Specific question or service requested from the specialist
- Urgency level: [Routine / Urgent / Emergency]
- Contact details for the referring physician
Tone: professional, concise. Length: 1 page maximum.
4. ICD-10 code lookup assistant
Given this diagnosis description, suggest the most appropriate ICD-10 code(s).
For each suggestion provide:
- The ICD-10 code and official description
- Why this code applies to the described condition
- Whether a more specific code may exist (flag if 4th or 5th character codes are available)
- Any commonly confused codes and how to differentiate
Diagnosis description: [CLINICAL DESCRIPTION]
Note: Final code selection must be confirmed by a qualified medical coder or the treating physician. ICD-10 coding has billing and regulatory implications.
Practical use: code lookup is one of the most reliable AI tasks in healthcare — it's factual, verifiable, and the stakes of a slightly wrong code are manageable (claims rejection, not patient harm).
5. Discharge medication list formatter
Convert these medication orders into a patient-friendly discharge medication list in both English and Hindi.
For each medication provide:
- Drug name (both brand and generic if possible)
- Dosage and form (tablet, syrup, injection)
- How to take it (with/without food, timing)
- Duration
- What it's for (in simple terms)
Medication orders: [LIST]
Format the Hindi version in simple, everyday language — not medical Hindi. A patient with 8th standard education should understand it.
Patient communication prompts
6. Appointment reminder (bilingual)
Generate WhatsApp appointment reminders in both English and Hindi for this patient.
Details:
- Patient name: [NAME]
- Doctor: [DR. NAME, DEPARTMENT]
- Hospital: [HOSPITAL NAME]
- Date and time: [DATE/TIME]
- Location: [ROOM/FLOOR/WING]
- Preparation instructions: [IF ANY — fasting, bringing reports, etc.]
English message: professional, under 150 words, include hospital helpline number [NUMBER]. Hindi message: warm and clear, simple language, same information.
7. Lab result explanation
Explain these lab results in plain English that a patient with no medical background can understand.
For each result:
- What the test measures (one sentence)
- What the normal range is
- Whether this patient's value is normal, high, or low
- What being outside the range generally means (general health information only)
Results: [PASTE LAB REPORT]
Do not suggest a diagnosis or treatment based on these results. End with: "Your doctor will explain what these results mean for your specific situation and what steps to take."
If the patient has requested Hindi explanation as well, provide the same in simple Hindi.
The boundary: result explanation (what does "HbA1c 8.2" mean in plain English) is appropriate. Interpretation for this patient's specific condition (what should YOU do about your HbA1c) belongs to the physician.
8. Post-discharge follow-up instructions
Generate patient-friendly follow-up instructions for a patient discharged after [CONDITION/PROCEDURE].
Include:
- Activity restrictions (what can/can't they do, for how long)
- Diet guidelines (specific to Indian food — use familiar food names like "dal", "roti", "rice")
- Warning signs requiring immediate hospital visit (specific, not vague)
- Warning signs requiring a call to the clinic
- Follow-up appointment requirement
- How to reach the hospital (number, emergency line)
Clinical guidelines: [PHYSICIAN TO PROVIDE OR CONFIRM]
Format: numbered list, plain language, no medical jargon.
9. Diet chart generator
Generate a diet chart for a patient with [CONDITION: diabetes / hypertension / post-cardiac surgery / post-bariatric surgery / renal disease].
Constraints:
- Vegetarian / Non-vegetarian / Eggetarian: [SPECIFY]
- Regional preference: [North Indian / South Indian / Gujarati / Bengali / etc.]
- Budget level: [economy / moderate / no restriction]
Provide:
- Morning schedule (wake up, breakfast, mid-morning)
- Afternoon (lunch, post-lunch)
- Evening (snack, dinner)
- Foods to avoid (with Indian food names)
- Special instructions
Dietary guidelines basis: [DIETITIAN'S GUIDELINES OR STANDARD CLINICAL GUIDELINES FOR CONDITION]
This chart is a starting point for the dietitian's review — final chart to be approved by a qualified dietitian.
10. Medication adherence message
Generate a WhatsApp reminder message for medication adherence for a patient on [MEDICATION NAME] for [CONDITION].
The message should:
- Acknowledge that taking medication daily feels tedious
- Explain in simple terms why consistency matters for this specific medication (not just generic "take your medicine")
- Provide one practical tip for building the habit (e.g., linking to a daily routine)
Tone: warm, not preachy. Like advice from a knowledgeable friend, not a lecture. Language: [English / Hindi / Hinglish] Length: under 100 words — this is a WhatsApp message, not an essay.
Hospital admin prompts
11. MIS report narrative
Convert this monthly hospital MIS data into a 3-paragraph narrative for the Medical Director / Hospital CEO.
Structure:
- Para 1: Volume metrics (OPD, IPD, bed occupancy, procedures)
- Para 2: Financial snapshot (revenue vs target, collections, outstanding)
- Para 3: Two key concerns or opportunities with a suggested next step for each
Data: [PASTE MIS TABLE OR SUMMARY]
Tone: direct, data-driven. Flag what's above target in positive terms, flag what's below target without blame. Use Indian number format (lakhs/crores).
12. Duty roster drafting
Generate a duty roster for [DEPARTMENT] for [WEEK/MONTH: DATE RANGE].
Staff available: [LIST NAMES] Pre-approved leaves: [NAME: DATES] Constraints:
- Minimum [N] doctors on duty at all times
- Maximum [X] hours continuous duty for any individual
- On-call duties should be distributed equitably
- [OTHER CONSTRAINTS]
Output as a table: rows = dates, columns = shifts (Morning/Evening/Night/On-call).
After the table, note any constraints that couldn't be fully satisfied and why.
13. Insurance pre-authorization letter
Draft a pre-authorization request letter to [INSURANCE COMPANY NAME] for [PROCEDURE/TREATMENT] for this patient.
Include:
- Patient details: [NAME, DOB, POLICY NUMBER, TPA ID]
- Diagnosis: [ICD-10 CODE AND DESCRIPTION]
- Proposed procedure: [PROCEDURE NAME AND CODE IF KNOWN]
- Clinical justification: [WHY THIS PROCEDURE IS MEDICALLY NECESSARY — physician to confirm]
- Estimated cost: ₹[AMOUNT]
- Treating physician: [DR. NAME, QUALIFICATION, REGISTRATION NUMBER]
- Proposed admission date: [DATE]
Tone: clinical and factual. Insurance companies review thousands of these — clarity and completeness beats elaborate prose.
14. NABH accreditation preparation checklist
Generate a preparation checklist for [NABH hospital accreditation / NABH small hospital standard / NABL lab accreditation] audit.
Hospital profile: [TYPE: multi-specialty / single-specialty / nursing home / diagnostic lab] Bed strength: [N] beds Previous accreditation status: [First time / Renewal]
Checklist should cover:
- Documentation requirements (policies, procedures, records)
- Process standards (patient care, infection control, medication management)
- Infrastructure requirements
- Staff training and competency requirements
- Common gaps found in similar hospitals during NABH audits
Format as a checklist with checkboxes. Group by chapter/standard if possible.
15. Staff training module
Generate a 10-point training module on [TOPIC: infection control / biomedical waste management / fire safety / patient fall prevention / medication safety] for [AUDIENCE: nursing staff / housekeeping / front desk / junior doctors].
For each point:
- The principle or rule
- Why it matters (consequence if not followed)
- How to do it correctly (specific action)
Include:
- A 5-question multiple choice quiz at the end (with answer key)
- Applicable Indian regulations or guidelines (CPCB for biomedical waste, NABH standards, etc.)
Language: simple English. If you can provide a Hindi version as well, please do.
Health-tech builder prompts
16. Symptom checker system prompt design
Design a system prompt for a symptom checker chatbot with these requirements:
- Asks clarifying questions to understand the user's symptoms (max 4-5 questions before a recommendation)
- Recommends appropriate care level: self-care at home / OPD visit / urgent care / emergency
- Never states or implies a diagnosis
- Handles: common acute conditions (fever, cold, stomach pain, headache), injury triage, and medication questions (general info only, no prescriptions)
Target users: Indian patients with basic English or Hindi literacy, accessing via a mobile app
Required elements in the system prompt:
- Clear scope: what the bot can and cannot do
- Escalation criteria: when to always recommend emergency care regardless of other factors
- Disclaimer language (suitable for Indian regulatory context)
- Tone guidance: warm and clear, not clinical or robotic
After the system prompt, provide a brief explanation of the design choices you made.
17. FHIR data extraction
Extract structured FHIR R4-compatible data from this clinical note. Map to the following FHIR resources where data is present:
- Patient (demographics)
- Condition (diagnoses, coded with ICD-10)
- MedicationRequest (current medications)
- Observation (vitals, lab results)
- Encounter (visit information)
Clinical note: [PASTE NOTE]
Output as JSON. For each resource:
- Use standard FHIR R4 structure
- Use official coding systems (ICD-10, LOINC for observations, RxNorm or Indian drug names for medications)
- Flag fields where data was ambiguous or missing as "status": "unknown"
18. Personal health record summary generator
Generate a patient health record summary from these individual visit notes. The summary should be:
- Usable by any new doctor seeing this patient for the first time
- Accurate to the notes (no inference beyond what's documented)
Structure:
- Active conditions (with onset dates where available)
- Current medications
- Known allergies and adverse reactions
- Significant past medical and surgical history
- Recent procedures and investigations (last 12 months)
- Primary care provider
Visit notes: [PASTE MULTIPLE VISIT NOTES IN CHRONOLOGICAL ORDER]
Mark any information that appears inconsistent across notes as [VERIFY: conflicting information].
19. Telemedicine pre-consultation intake form
Design a pre-consultation intake form for telemedicine consultations in [SPECIALTY: general medicine / dermatology / psychiatry / orthopaedics / gynaecology].
Constraints:
- Maximum 10 questions
- Must be answerable on a WhatsApp chat or a simple web form (no dropdowns, just text or yes/no)
- Language level: 8th standard
- Available in English and Hindi
Questions should gather:
- Chief complaint (what brings you in today)
- Duration and severity
- Relevant past history for this specialty
- Current medications
- Anything else the doctor needs to know before the call
After the form, explain which questions are mandatory vs optional, and why.
20. Ayurveda integration context design
Design a system prompt for an AI wellness assistant that can discuss Ayurvedic approaches while operating within appropriate boundaries.
The assistant should:
- Discuss Ayurvedic concepts (doshas, seasonal routines, dietary guidelines from classical texts) as traditional wellness knowledge
- Answer questions about commonly used Ayurvedic herbs (Ashwagandha, Triphala, Tulsi, etc.) including traditional uses and known interactions
- Not make medical claims or suggest Ayurvedic treatments for diagnosed conditions
- Appropriately defer clinical questions to qualified Ayurvedic vaidyas or BAMS/MD (Ayu) practitioners
- Not position Ayurveda against modern medicine — it should acknowledge both as valid systems
Target audience: educated Indian users interested in integrating Ayurvedic wellness practices into daily life, not looking for treatment of diagnosed conditions.
Include appropriate disclaimers that fit Indian regulatory context (AYUSH Ministry guidelines).
Try it now with AICredits.in
Access Claude, GPT-4o, Gemini, and 300+ models with UPI payment in ₹ — ideal for health-tech development without an international card. ₹100 minimum to start. Create free account →
Next steps
- AI prompts for small business owners in India — the same prompt-heavy format for business contexts
- Prompt engineering for fintech India — financial sector applications
- Professions hub — AI prompts organised by profession
- System prompts lesson — how to build better system prompts for production use
