Basic usage of CDSS to generate a protocol recommendation.

This example demonstrates how to use the DataLoader, DataProcessor, and CDSS to load data and generate a 7-day protocol plan.

scoring 2025-06-16 00:00:00
/home/runner/work/ai-cdss/ai-cdss/src/ai_cdss/data_processor.py:303: FutureWarning: DataFrameGroupBy.apply operated on the grouping columns. This behavior is deprecated, and in a future version of pandas the grouping columns will be excluded from the operation. Either pass `include_groups=False` to exclude the groupings or explicitly select the grouping columns after groupby to silence this warning.
  df = df.groupby(by=[PATIENT_ID, 'SESSION_DATE'], group_keys=False).apply(day_skip_to_nan)
   PATIENT_ID  PROTOCOL_ID       PPF  \
0           1            1  0.562178
1           1            2  0.965500
2           1            3  0.812396
3           1            4  0.719061
4           1            5  0.409213

                                             CONTRIB  ADHERENCE_RECENT  \
0  [0.252232, 0.03653, 0.118418, 0.147197, 0.007801]          0.673392
1   [0.119867, 0.02396, 0.00862, 0.381091, 0.431962]          0.700176
2   [0.34855, 0.076633, 0.02168, 0.243166, 0.122367]          0.746217
3   [0.026373, 0.13851, 0.007091, 0.486407, 0.06068]          0.718749
4  [0.139366, 0.047926, 0.094186, 0.101513, 0.026...          0.762258

   DELTA_DM  USAGE  USAGE_WEEK                DAYS  GAME_SCORE
0 -0.058395     10           0  [0, 1, 2, 3, 4, 6]    1.177175
1 -0.064327     10           0     [0, 3, 4, 5, 6]    1.601349
2 -0.048376     10           0  [0, 2, 3, 4, 5, 6]    1.510237
3  0.042883     10           0        [0, 1, 4, 5]    1.480693
4  0.018059     10           0  [0, 1, 2, 4, 5, 6]    1.189530

import sys
sys.path.append("..")
from ai_cdss.cdss import CDSS
from ai_cdss.data_loader import DataLoaderMock
from ai_cdss.data_processor import DataProcessor

import pandas as pd
from IPython.display import display

print(__doc__)

PATIENT_LIST = [
    775,  787,  788
]

# Parameters
rgs_mode = "app"
weights = [1,1,1]
alpha = 0.5

n = 12
days = 7
protocols_per_day = 5

# Services
loader = DataLoaderMock(
    num_patients=5,
    num_protocols=5,
    num_sessions=10
)
processor = DataProcessor(
    weights=weights,
    alpha=alpha
)

# Execution
session = loader.load_session_data(patient_list=PATIENT_LIST)
timeseries = loader.load_timeseries_data(patient_list=PATIENT_LIST)
ppf = loader.load_ppf_data(patient_list=PATIENT_LIST)
init_metrics = loader.load_protocol_init()
protocol_similarity = loader.load_protocol_similarity()

scores = processor.process_data(
    session_data=session,
    timeseries_data=timeseries,
    ppf_data=ppf,
    init_data=init_metrics)

# CDSS
cdss = CDSS(
    scoring=scores,
    n=n,
    days=days,
    protocols_per_day=protocols_per_day
)

# Results
# patient_id = PATIENT_LIST[0]
patient_id = 1

recommendation = cdss.recommend(patient_id=patient_id, protocol_similarity=protocol_similarity)
recommendation.to_csv(f"recommendation_{patient_id}_new.csv", index=False)

with pd.option_context('display.max_columns', None):
    display(recommendation)