Skip to main content

​
stratify_risk

Assign risk bands (low/medium/high) to patients based on predicted probabilities. 
stratify_risk(
    probabilities: pd.Series,
    low_threshold: float = 0.35,
    high_threshold: float = 0.7
) -> pd.DataFrame
​
probabilities
pd.Series
required
Series of predicted probabilities for each patient (values between 0 and 1)
​
low_threshold
float
default:"0.35"
Threshold below which patients are classified as low risk
​
high_threshold
float
default:"0.7"
Threshold at or above which patients are classified as high risk
​
return
pd.DataFrame
DataFrame with columns:
  • risk_probability: Original probability values (float)
  • risk_band: Risk category as “low”, “medium”, or “high” (str)

​
Risk Band Classification

  • Low: probability < low_threshold (default < 0.35)
  • Medium: low_threshold ≤ probability < high_threshold (default 0.35-0.7)
  • High: probability ≥ high_threshold (default ≥ 0.7)

​
summarize_risk_bands

Calculate the prevalence of each risk band in a stratified population. 
summarize_risk_bands(risk_frame: pd.DataFrame) -> dict[str, float]
​
risk_frame
pd.DataFrame
required
DataFrame output from stratify_risk containing a ‘risk_band’ column
​
return
dict[str, float]
Dictionary with normalized prevalence for each risk band:
  • low_prevalence: Proportion of low risk patients (0.0-1.0)
  • medium_prevalence: Proportion of medium risk patients (0.0-1.0)
  • high_prevalence: Proportion of high risk patients (0.0-1.0)
All values sum to 1.0. Missing bands default to 0.0.

Build docs developers (and LLMs) love

Get started for freeTalk to us