What problem does it solve?
Getting a patient into the right slot is harder than booking a table. Appointments depend on a referral, a triage level, the right clinician and room, preparation instructions and sometimes a test that has to happen first. Patient access teams do this by phone, and at peak times callers wait on hold or give up, so access to care depends on how long someone can stay on the line.
At the other end, a large share of booked appointments is simply lost. Patients forget, cannot get transport, cannot take time off or no longer need the visit but never cancel. Every missed appointment is clinical time that nobody uses while other patients wait. Standard one way text reminders help, but they are sent to everyone at the same moment, cannot answer a question and leave the freed slot empty when a patient cancels late.
- NHS England reports that of 124.5 million outpatient appointments in England in the previous year, eight million (6.4%) were not attended, at an estimated annual cost of £1.2 billion.NHS AI expansion to help tackle missed appointments and improve waiting times (2024)
- The UK government reports 8 million missed appointments in elective care in 2023 to 2024.Power to patients as government sets out plan to cut waiting lists (2025)
How does it work?
- Take the request on any channel. The agent answers calls and messages to book, move or cancel an appointment, and identifies the patient against the record before it discloses or changes anything.
- Apply the scheduling rules. It checks the referral, visit type, clinician, location and any preparation or prior test the rules require, and only offers slots the scheduling system returns for that combination.
- Book and confirm. It writes the booking into the electronic health record or patient administration system and sends a confirmation with the preparation instructions.
- Remind in a conversation. Reminders go out at the times that work best for that clinic, and the patient can confirm, cancel or move the appointment in the same thread.
- Target extra help. A model scores which appointments are likely to be missed; high risk patients get an extra reminder, a better suited time, or an offer of help with transport.
- Backfill freed slots. When a patient cancels, the agent offers the slot to suitable patients on the waiting list so the clinical time is used.
- Hand clinical questions to people. Symptoms, urgent concerns and anything outside scheduling go to staff or to urgent care guidance, with the conversation attached.
- Audience
- Customer facing
- Autonomy
- Supervised agent
- Adoption
- Early adopters
- Channels
- Phone and voice, SMS, Web chat, Mobile app, WhatsApp
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Containment rate | Too few to pool | 30% | 1 | 1 vendor |
| Handling time reduction | Too few to pool | 72% | 1 | 1 vendor |
| Interactions handled | Not pooled | at least 160,000 | 1 | 1 organization |
| Response time reduction | Too few to pool | 87% | 1 | 1 vendor |
| Satisfaction uplift | Too few to pool | 4% | 1 | 1 vendor |
Value drivers: Customer experience, Lower cost to serve, Inclusion and access, Employee productivity.
Indicative value
A hospital with 600,000 outpatient appointments a year
EUR 432,000 to EUR 2.4 million
Value of clinical time recovered from avoided or backfilled missed appointments per year
How this is calculated
Formula: appointments * missedRate * reduction * valuePerSlot. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Outpatient appointments per year appointments, appointments per year | 600,000 | 600,000 | The reference hospital. |
| Share of appointments not attended today missedRate, fraction of appointments | 0.06 | 0.08 | NHS England reports 6.4% of outpatient appointments in England were not attended, and higher rates in some specialties (11% in physiotherapy, 8.9% in cardiology). Replace with your own rate. Source |
| Share of missed appointments avoided or backfilled reduction, fraction of missed appointments | 0.1 | 0.3 | Conservative against the evidence on this page (NHS England reports a 30% fall in non attendance in the Mid and South Essex pilot). Editorial assumption, replace with your own. |
| Value of an outpatient slot that is used instead of wasted valuePerSlot, EUR per appointment | 120 | 170 | NHS England's estimate of £1.2 billion a year for eight million missed appointments implies roughly £150 (about EUR 175) per missed appointment; the euro range is set below that figure to stay conservative. Replace with your own cost per slot. Source |
What it leaves out: Counts recovered clinical capacity only. It leaves out booking calls handled by the agent, shorter waits for patients, the cost of the AI, messaging and integration, and the extra appointments that may be needed when more patients attend.
Market estimates (analyst estimates, not deployments)
- The UK government says focused action, including AI predictions of which appointments are most likely to be missed, will help save up to 1 million missed appointments. PM sets out plan to end waiting list backlogs through millions more appointments (2025)
Who already uses it?
6 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
Mid and South Essex NHS Foundation Trust
United Kingdom · Healthcare · 2024
Mid and South Essex NHS Foundation Trust piloted software from Deep Medical that predicts which outpatient appointments are likely to be missed, using anonymised data and external factors such as weather, traffic and jobs. It offers patients more convenient times (for example evening and weekend slots for people who cannot take time off) and places intelligent backup bookings so clinical time is not lost. NHS England reports that the six month pilot cut non attendance by 30%, prevented 377 missed appointments and let an additional 1,910 patients be seen, and it announced a rollout to ten more trusts. The outcome is a reduction in missed appointments, which has no matching KPI in the taxonomy, so it is recorded in this summary rather than as a metric.
No outcome disclosed.
University Hospitals Coventry and Warwickshire NHS Trust
United Kingdom · Healthcare · 2024
University Hospitals Coventry and Warwickshire used AI based process mining with IBM and Celonis to study missed appointments, which were more common among patients with high deprivation scores. It found a spike in last minute cancellations after two SMS reminders and moved to a reminder 14 days before the appointment with a second one four days before, so patients could cancel early and the slot could be rebooked. NHS England reports that missed appointments in this subset of patients fell from 10% to 4%. The AI analysed the process; the reminders themselves are standard text messages.
No outcome disclosed.
WellSpan Health
United States · Healthcare · 2024
WellSpan Health started with a Hippocratic AI voice agent that phones patients to close gaps in colorectal cancer screening, in English and Spanish, and supports low risk patients before and after a scheduled colonoscopy, with transcripts sent to clinicians and live transfer to a human where needed. By 2026 the agent, Ana, answered inbound calls and scheduled primary care appointments, and WellSpan announced an expanded partnership whose first new workflows call patients who missed imaging appointments. WellSpan reports that Ana manages more than 160,000 patient calls a month.
- Interactions handled: at least 160,000, per month, patient calls
"Ana, WellSpan’s generative AI agent developed in partnership with Hippocratic AI, currently manages more than 160,000 patient calls per month, engaging in more than 7,000 hours of conversation."
Claimed by: organization
Sheffield Children's NHS Foundation Trust
United Kingdom · Healthcare · 2023
Sheffield Children's piloted an AI Predictor developed by Alder Hey Innovation that estimates which children are likely to miss ("was not brought") an appointment, using markers that include health inequalities. Families with a predicted risk of 50% or more received an extra text reminder with an offer of support the day before; families at 85% or more were contacted and offered funded transport or a rebooking. NHS England reports that 53,800 texts were sent in the first 12 months, that recorded non attendance came in well below the expected benchmark (almost 200 more attended appointments a month), and that in a 13 week period 152 families had transport arranged and 129 appointments were rebooked.
No outcome disclosed.
Audibel
United States · Healthcare · 2025
Audibel, a network of more than 400 hearing care clinics in the United States, put a PolyAI voice agent in front of its call centre, which received about 2,000 calls a day with 10 to 15 minute holds. The agent collects the caller's name, phone number and zip code, identifies the intent (such as scheduling an appointment), filters spam and hands over to a human with context. PolyAI reports 87% shorter wait times, abandonment down from 46% to 2% and appointment volume up 2% year on year. Automated booking and confirmations are named as the next phase, not yet live.
- Response time reduction: 87%, call wait time
"With PolyAI, Audibel cut wait times by 87% and abandonment by 44%, while spam dropped 88%."
Claimed by: vendor
Howard Brown Health
United States · Healthcare · 2024
Howard Brown Health, a federally qualified health center in Chicago, deployed a PolyAI voice agent named Alex that answers patient calls around the clock in several languages. Integrated with MyChart, it guides patients through scheduling appointments, test results and prescription refills, and escalates distressed callers to a person immediately. PolyAI reports 30% call containment against a 20% target, a 72% shorter average handle time for routine requests and a 4% increase in patient satisfaction. An Epic integration that lets patients create, reschedule and cancel appointments through the agent is described as the next phase.
- Containment rate: 30%
"Initially aiming for a 20% call containment rate, PolyAI exceeded expectations by achieving 30%."
Claimed by: vendor - Handling time reduction: 72%, routine requests
"72% decrease in Average Handle Time for routine requests"
Claimed by: vendor - Satisfaction uplift: 4%
"Patient satisfaction scores also saw an increase of 4%, primarily driven by the enhanced ease and efficiency with which patients could schedule appointments and access services."
Claimed by: vendor
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- Scheduling rules per specialty, visit type and clinician, owned by the service
- Live slot availability and waiting list data from the scheduling system
- Contact details, preferred language and channel consent per patient
- Historic attendance data to train or calibrate a missed appointment model
- Approved preparation instructions per procedure
Systems to integrate
- Electronic health record or patient administration system (scheduling and waiting list)
- Patient portal or app (identity and messaging)
- Telephony and messaging (voice, SMS, WhatsApp)
- Referral management
- Contact centre for handover to patient access staff
Complexity: Medium
The conversation is the easy part. The work is in the scheduling rules (which visit types a patient may book, which need a referral or triage first), a reliable read and write integration with the electronic health record, and identity checks strong enough for health data. Clinical triage by the agent would push this into medical device territory, so keep it out of scope.
- 1
Start with rebooking and cancellations
Cancelling and moving existing appointments is lower risk than first bookings and frees slots immediately. Add new bookings per visit type once the rules for that type are written down.
- 2
Write the scheduling rules before the prompts
For each visit type, record who may book it, what must happen first and what the patient must be told. The agent calls these rules as tools; it never infers eligibility from the conversation.
- 3
Make reminders two way and well timed
Test reminder timing per clinic. University Hospitals Coventry and Warwickshire found that a reminder 14 days ahead with a second one four days ahead let patients cancel early enough to rebook. Let the patient act in the same thread.
- 4
Use risk scores to offer help, not to penalise
Use the missed appointment score to send extra reminders, offer better times or transport support. Do not use it to deny or downgrade bookings, and check it for bias by deprivation, ethnicity and age.
- 5
Close the loop on freed slots
Connect cancellations to the waiting list so a freed slot is offered at once to suitable patients, and measure how many freed slots are actually used.
Guardrails
- No clinical advice or triage decisions by the agent; symptoms go to staff or urgent care guidance
- Only slots and visit types the scheduling system confirms for that patient
- Identity verification before any appointment detail is disclosed or changed
- Missed appointment scores used only to offer support, reviewed for bias
- Clear AI disclosure and an easy route to a person on every channel
KPIs to instrument
- Missed appointment rate per clinic, against a control group or the prior period
- Share of freed slots rebooked from the waiting list
- Booking, rebooking and cancellation containment per visit type
- Call wait and abandonment rates for patient access lines
- Missed appointment rates by deprivation band, ethnicity and age
Human in the loop
Patient access staff own exceptions: urgent symptoms, complex bookings across several services, safeguarding concerns and patients who ask for a person. Service managers approve the scheduling rules per visit type, and a clinical safety officer signs off the scope and reviews a sample of conversations each month.
Common failure modes
- Booking the wrong visit type
- The patient arrives without the required test or preparation. Enforce rules per visit type through tools and confirm preparation in writing.
- Reminder fatigue
- Too many messages lead patients to ignore them or cancel at the last minute. Test timing and frequency per clinic.
- A risk model that widens inequality
- Scores that correlate with deprivation lead to overbooking or deprioritising the same groups. Use scores for support only and monitor outcomes by group.
- Symptoms missed in a booking call
- A patient mentions a red flag symptom while rebooking. Detect it and route to clinical staff or urgent care guidance at once.
What are the risks and rules?
EU AI Act
Depends on design
Booking, rescheduling and reminders carry transparency duties: patients must be told they are dealing with AI (Article 50(1)). It becomes high risk if a public authority, or a provider acting on its behalf, uses it to evaluate eligibility for healthcare services (Annex III point 5(a)), or if it acts as an emergency healthcare patient triage system (Annex III point 5(d)). Clinical triage may also make it a medical device, which is high risk under Article 6(1) when the device needs a notified body assessment. Keep the agent to scheduling and use risk scores only to offer support.
Rules that apply
Guidance
- Medical devices: software and artificial intelligence (AI) (Medicines and Healthcare products Regulatory Agency, Europe). UK guidance on when software, including AI, is regulated as a medical device; relevant if the agent moves from scheduling into triage.
- Declaratory ruling on AI generated voices under the TCPA (FCC 24-17) (Federal Communications Commission, North America). AI generated voices count as artificial or prerecorded voice under the TCPA, which matters for automated reminder calls in the US.
Controls to put in place
- Clinical safety case with the agent's scope written down and signed off
- Audit trail of every booking, change and cancellation the agent made
- Data protection impact assessment covering health data, recordings and risk scores
- Regular bias review of the missed appointment model
- Regression tests for scheduling rules on every change
Frequently asked questions
- Does AI actually reduce missed hospital appointments?
- Public pilots suggest it can. NHS England reports a 30% fall in non attendance at Mid and South Essex NHS Foundation Trust with software that predicts missed appointments and books backup slots, and a drop from 10% to 4% in one patient group at University Hospitals Coventry and Warwickshire after AI analysis changed reminder timing. These are pilot results, and none of them is reported against a control group.
- How is this different from a general appointment booking chatbot?
- A general booking agent finds a location and a free time. A patient scheduling agent has to respect referral, triage and visit type rules, write into the health record, handle preparation instructions and stay out of clinical advice, which makes the integration and the safety case the main work.
- Can a voice agent handle patient calls at scale?
- WellSpan Health reports that its AI agent Ana manages more than 160,000 patient calls a month, including inbound calls and primary care scheduling. PolyAI reports 30% containment and a 72% shorter handle time for routine requests at Howard Brown Health, where the agent guides patients through scheduling in MyChart; booking, rescheduling and cancelling through Epic is the announced next phase.
- Is a missed appointment prediction model high risk?
- Used to send extra reminders or offer help, it is usually not. It needs more care when scores decide who is overbooked or deprioritised, because that can limit access to care for the groups who already miss most appointments.
How to cite this page
Blits.ai AI Use Case Library, "AI agent for patient appointment scheduling, reminders and no show reduction", last verified 27 September 2026, https://www.blits.ai/ai-use-cases/patient-appointment-scheduling-and-reminders-agent. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 27 September 2026: First published