Skip to main content

​
Overview

Predictions estimate a metric for a hypothetical case profile — a procedure (or condition / DRG) plus demographics and any secondary conditions — and present the result with confidence and prediction-interval context.

​
When to use it

  • You want a forward-looking estimate for a case that hasn’t happened yet.
  • You’re scoping the likely price, cost, time, or profit of a planned procedure.
  • You want to see how an estimate varies when grouped by client, provider, surgeon, or payor.

​
Inputs

  • Domain & main concept — a procedure, condition, or DRG (resolve it first via the Entity Resolver).
  • Metric — see supported values below.
  • Gender — see below.
  • Age — see below.
  • Secondary conditions — see below.
  • Group by / filter — client, provider, surgeon, or payor.

​
Supported metrics

​
metric
string
One of: price, cost, time, profit.
profit is derived (computed from price and cost) rather than predicted directly. Metric availability can vary by organization — if a metric isn’t offered for your procedure or organization, it isn’t supported in your context.

​
Gender

​
gender
string
M, F, or unknown/blank.
Any value other than M or F (including empty) is treated as unknown. An unknown value does not error — it lowers the prediction’s confidence rather than blocking it.

​
Age

​
age
string
A numeric age 0–120, or a legacy age band.
Either a number (clipped to the 0–120 range) or one of the legacy age-band labels: Less than 2, 2-5, 6-12, 13-19, 20-24, 25-44, 45-65, Over 65.

​
Secondary conditions

Secondary conditions are provided as standardized OMOP concept IDs, not a fixed list of choices. Resolve each one via the Entity Resolver first. The kind of concept expected for the “secondary” slot depends on the domain of your main concept.

​
Steps

1

Select the main concept

Choose the procedure, condition, or DRG.
2

Set demographics

Choose gender and age.
3

Add secondary conditions

Add any relevant secondary conditions (resolved concept IDs).
4

Choose a metric

Select price, cost, time, or profit.
5

Group and filter

Group by client, provider, surgeon, or payor, and filter within that group.
6

Read the estimate

Review the estimate together with its confidence and prediction-interval context.

​
Outputs

  • A point estimate for the chosen metric.
  • Prediction intervals / ranges so the estimate can be read as a band rather than a single guaranteed number.
  • Model context such as a confidence indicator and an overall prediction-quality signal.

​
Notes & caveats

  • Estimates, not guarantees. Always read predictions as ranges and weigh the confidence/quality signals.
  • Metric vocabulary differs from Analytics. Predictions use price for the billed amount; Analytics calls the comparable figure revenue.
  • Availability varies by organization and procedure, so not every metric is offered in every context.
  • An unknown gender lowers confidence rather than causing an error.
  • Results are scoped to your organization.

​
Support

Need help configuring or interpreting a prediction? See Support or email admin@valiancehealth.ai.