Third-next-available has not moved in two years. The template was set by a committee that has since dissolved; half the slots are held by rules nobody can trace to a decision, and the call center is measured on abandonment while the clinic is measured on utilization, so the two of them optimize against each other every single day.
Scheduling is where AI is easiest to buy and hardest to bank. A no-show model is only worth what your overbooking policy lets you do with it. An optimizer is only worth what your template governance permits it to change. Buy the model without fixing the policy underneath, and you get a more accurate prediction feeding a schedule that cannot act on it, which is an expensive way to be right.
This book works the other order. It starts with the metrics that actually move and the ones that only look like they do, sets the template governance a model needs to be useful, and only then evaluates the tool. It says plainly where the category is thin and which claims to make a vendor demonstrate on your data instead of theirs.
What this book will not do. It will not promise you an access improvement, because the size of one depends on your template, your payer mix, and your no-show pattern, none of which a book can see. It does not rank named products. It gives you the measurement and the evaluation method, and leaves the conclusion where it belongs, which is with you and the people who run your schedule.
What you will be able to do
- Instrument access with a metric set that survives a board question, and retire the ones that measure the schedule rather than the patient.
- Separate demand you can predict from demand you cannot, and stop trying to optimize the second.
- Set the overbooking and release policy before the no-show model arrives, so the prediction has somewhere to land.
- Score a scheduling vendor across seven dimensions with weights you set in advance and can defend afterwards.
- Design a pilot with a baseline, a stop rule and an owner, so it ends in a decision rather than an extension.
- Say what the tool must show on your own historical data before you sign.
The frameworks this book builds
- The access metrics dashboard: the small number of measures that move together, with the definition written down so it cannot drift.
- The seven-dimension vendor evaluation: capability, integration, workflow fit, evidence, security, service and exit, weighted before you see a demo.
The companion instruments. Every framework here is published as a working file in the AI-Powered Scheduling Optimization Toolkit: 6 workbooks and documents in Excel and Word, licensed to a single organization, available separately from the Institute. Data validation on the cells you fill, IFERROR on every division, formula cells protected and your cells left open. The book gives you enough to build these yourself. The toolkit exists so that you do not have to.
Written from the operator's chair. Neel Chauhan, MD MBA is a physician-executive whose career spans three national healthcare systems. Quantitative claims carry a named, dated source in the same sentence, or the sentence says where the evidence stops. Composites are labeled as composites. The Healthcare AI Institute accepts no vendor sponsorship, holds no vendor equity and takes no referral fees, which is the only reason a chapter is free to say that a category is not ready.
Who this is for
Access leaders, ambulatory COOs and the operations teams who own the schedule. Written for the person who will have to make the template change, not the person who signs the purchase order.
