Skip to main content

​
run_streaming_inference

def run_streaming_inference(
    X: pd.DataFrame,
    model,
    chunk_size: int
) -> pd.DataFrame
Runs inference on patient data in streaming mode, processing chunks sequentially to predict patient risk.
​
X
pd.DataFrame
required
Input features DataFrame containing patient data for risk prediction
​
model
sklearn-compatible model
required
Trained model with predict_proba method for risk prediction. Must return probability scores for binary classification
​
chunk_size
int
required
Number of rows to process per chunk in streaming mode
​
return
pd.DataFrame
DataFrame with the following columns:
  • risk_probability: Predicted probability of positive risk (class 1)
  • risk_label: Binary label (0 or 1) based on 0.5 probability threshold
The returned DataFrame maintains the original index from the input and is sorted by index.

​
Example

from real_time.inference import run_streaming_inference
import pandas as pd

# Assume model is already trained
X_test = pd.DataFrame(...)  # Patient features

results = run_streaming_inference(
    X=X_test,
    model=trained_model,
    chunk_size=100
)

print(results.head())
# Output:
#    risk_probability  risk_label
# 0          0.234567           0
# 1          0.876543           1
# 2          0.456789           0

Build docs developers (and LLMs) love

Get started for freeTalk to us