AI for Pulmonology Practices in South Africa: Spirometry and Inhaler Review Recall
Pulmonology practices coordinate GP referrals, consultations, clinician-directed investigations, review appointments and ongoing follow-up. Reception teams may also handle missed calls, incomplete referrals and rescheduling requests. Xtreme AI can help organise confirmations and recall without turning a bot into a respiratory adviser.
Key takeaways
- Separate appointment and test administration from clinical interpretation or advice.
- Use clinician-approved recall lists for spirometry and inhaler reviews; never let automation decide who needs review.
- Provide a human hand-off for symptoms, worsening breathing, medicine questions or test-result queries.
- Use consented, minimal WhatsApp or SMS messages and keep private health details out of notifications.
Design around the respiratory patient journey
A referral may arrive with a question for the specialist and supporting documents, but the practice still needs a reliable method to acknowledge receipt, check what is missing and arrange a suitable appointment. Following the first visit, the clinician may ask the team to coordinate a test appointment and schedule a separate review. An administrative system should track those bookings as distinct events, with clear ownership for confirming whether information has arrived and whether the patient has been contacted.
Spirometry recall is especially suited to a human-approved workflow. The clinician or practice protocol determines when a patient should be invited back; staff validate the list and handle exceptions. The message simply asks the patient to contact the practice or select a booking option. Likewise, an inhaler-review reminder can invite an appointment, but it must not tell a person how to use, change or stop an inhaler. Do not infer adherence or clinical need from message replies.
Examples for WhatsApp and SMS
Where the patient has agreed to receive communications, an official WhatsApp Business API connection can support two-way confirmations and staff follow-up. SMS can be offered where appropriate. Keep the wording neutral, identify the practice, and make it easy for a patient to request a person. These examples are administrative templates to adapt and approve locally:
- Referral acknowledgement: confirm that the practice received a referral and explain that the team will contact the patient after administrative review. If documents are missing, staff can request them through a channel the practice has approved for health information.
- Spirometry appointment: confirm the booking and share only preparation instructions that the responsible clinician has explicitly approved for that patient and test. If no approved instructions are available, provide the practice number rather than inventing preparation guidance.
- Inhaler-review recall: invite a patient on a clinician-approved list to book a review. Avoid naming the medicine or condition in a notification unless the patient has specifically asked for that level of detail and the practice has assessed the privacy risk.
- After-hours booking: acknowledge an enquiry, collect a preferred callback time and offer routine appointment booking. Any message that mentions breathing difficulty, an adverse effect or a clinical concern should be handed to the practice's human process; automation must not triage severity.
| Workflow | What automation can do | What staff or clinician must do |
|---|---|---|
| New referral | Acknowledge receipt and create a follow-up task | Review referral and determine clinical priority |
| Spirometry booking | Send time, location and approved logistics | Choose test and validate any preparation instructions |
| Recall invitation | Invite a patient to book and record a response | Approve recall timing and review non-responses |
| Clinical question | Pause automated replies and alert the team | Respond through appropriate clinical procedures |
Guardrails for clinical safety and privacy
AI must not diagnose asthma or another respiratory condition, interpret spirometry, assess symptoms, recommend investigations, offer clinical advice or give medication instructions. A bot should not decide that a patient is safe to wait. The treating practitioner remains accountable under HPCSA professional requirements; staff should follow the practice's established response and emergency communication procedures.
Respiratory information can reveal sensitive health details. Under POPIA, the practice should define its lawful purpose and safeguards for processing, explain communications to patients, respect preferences and obtain or record consent where required. Limit message content, restrict staff access and agree how records are retained. Use an official WhatsApp Business API setup, verify the service providers involved, and ensure that a patient's preference for SMS, WhatsApp or telephone is recorded. If a person opts out, update the communication preference and use an appropriate alternative.
Make the workflow testable
Begin with routine appointment confirmations. Document the trigger, approved recall list, staff owner and exception process. Test missing referral information, rescheduling and a clinical concern. Check that staff see escalations and patients can reach a person. Track unanswered enquiries and incomplete follow-up; do not assume automation improves outcomes by itself.
Read the directory of South African GPs and specialists, the complete healthcare AI guide, the HPCSA and POPIA guide for healthcare practices, the 2026 South African healthcare AI statistics, and the AI ROI calculator guide. For adjacent patient-communication examples, see AI for ENT specialist communication and AI for cardiology practices.
Frequently asked questions
Can AI interpret spirometry results?
No. A qualified healthcare professional must interpret tests and explain their significance.
Can a reminder include inhaler instructions?
Only clinician-approved, patient-specific instructions should be sent; the automated system must not create or change medication advice.
How should after-hours breathing concerns be handled?
The bot should not judge severity. It should hand the message to the practice's defined human response route and provide only pre-approved contact information.
Can the practice send recall messages to everyone?
Recall lists should be approved under the practice's clinical process, and communications should respect consent and channel preferences.
Want to see how this would work in your practice? WhatsApp Xtreme AI for a free walkthrough.
