Playbook: Healthcare and Wellness

Scope: this playbook covers administration, education and research support. AI must not be used to diagnose, prescribe or make clinical decisions. Those remain the responsibility of licensed professionals.

Doctors, physiotherapists, dietitians and therapists spend a large share of their day on work that isn't patient care: explaining the same things repeatedly, keeping up with research, writing follow-ups and managing appointments. That's where AI helps — safely, if you respect two lines.

---

1. The two lines

Line 1 — no diagnosis or treatment decisions. AI can summarise, explain and draft. It can't examine a patient, and models can produce confident, wrong medical information.

Line 2 — no identifiable patient data in public AI tools. Health data is sensitive personal data. In India, the DPDP Act and its 2025 Rules govern it; in the US, HIPAA requires a signed Business Associate Agreement (BAA) before a vendor handles patient information. Consumer AI accounts don't give you that.

Practical rule: never paste patient names, phone numbers, record numbers, or a combination of details that could identify someone. Use approved clinical systems for anything identifiable.

2. Patient-friendly explanations

Patients who understand their condition and plan follow it better. AI is excellent at translating jargon.

Rewrite the following clinical explanation for a patient.
Audience: 60 years old, no medical background, reads English
comfortably but prefers short sentences.

- Use plain words; explain any medical term you must keep
- Focus on what they should DO: 3-5 clear actions
- Include "When to call the clinic immediately" warning signs
  exactly as given in the source — do not add or remove any
- Under 250 words
- Do not add medical advice that isn't in the source

[PASTE GENERIC CONDITION/PLAN INFORMATION — no patient details]

Then review it clinically, line by line, before any patient sees it. Ask for versions in Hindi or other regional languages, and have a fluent colleague check them.

Build a library of reviewed explainers for your most common conditions — you write it once and use it for years.

3. Keeping up with research

Summarise this study for a practising [specialty] clinician.

1. Question, design, population, sample size
2. Main findings with effect sizes as reported
3. Limitations — including ones the authors don't emphasise
   (sample, duration, funding, generalisability to Indian patients)
4. What this does and does not change for practice
5. Quote the sentence supporting each finding

[PASTE ABSTRACT OR FULL TEXT]

Use research-grounded tools (e.g. PubMed search, Elicit, Consensus, OpenEvidence where available) to find papers, and read the key sections yourself. Never cite a paper you haven't opened — models invent plausible-looking references.

4. Appointments and follow-up automation

This is where most practices save the most time.

A typical safe workflow (built with the automation tools from earlier in the course):

1. Booking system (Practo, a calendar, or a clinic management system) triggers when an appointment ends.
2. A template — pre-approved by the clinician — is filled with the first name, next appointment date and a link to the relevant generic exercise sheet or explainer.
3. Sent via WhatsApp Business API or SMS with the patient's recorded consent to receive messages.
4. Replies go to a human. The automation never answers clinical questions.

Hi {first_name}, thanks for visiting {clinic} today.
Your next appointment: {date} at {time}.
Your exercise guide: {link}
If your pain gets worse or you notice {warning_signs}, call us on
{phone} — don't wait for your appointment.

Timely, personalised reminders improve attendance and adherence without extra staff time. Use WhatsApp's approved message templates and follow its opt-in rules.

5. Countering "Dr. Google 2.0"

Patients now arrive having asked a chatbot. Rather than dismissing it:

• Ask what they read, and address it directly.
• Publish your own reviewed educational content — FAQs, short videos, explainers — so the good information online includes yours.
• End every piece with a clear message about when to see a professional.

---

⚠️ Common mistakes

• Using AI for diagnosis or treatment decisions.
• Pasting identifiable patient data into consumer AI tools.
• Publishing AI-written health content without clinical review.
• Citing studies you haven't read.
• Automations that reply to clinical questions instead of routing them to a person.

---

What's next: the playbook for students and academics — using AI to understand more, not think less.