Regression

In [1]:

import sys
import pandas as pd
import os

project_root = os.path.abspath(os.path.join(os.getcwd(), "../../"))

sys.path.append(project_root)
data_dir = os.path.join(project_root, 'data')

data_file_path = os.path.join(data_dir, 'RADCURE_processed_clinical.csv')
df = pd.read_csv(data_file_path, index_col=0)

df.drop(columns=["Study ID", "survival_time", "death"], inplace=True)

import sys import pandas as pd import os project_root = os.path.abspath(os.path.join(os.getcwd(), "../../")) sys.path.append(project_root) data_dir = os.path.join(project_root, 'data') data_file_path = os.path.join(data_dir, 'RADCURE_processed_clinical.csv') df = pd.read_csv(data_file_path, index_col=0) df.drop(columns=["Study ID", "survival_time", "death"], inplace=True)

In [2]:

from jarvais.analyzer import Analyzer
from rich import print

analyzer = Analyzer(
    data=df, 
    output_dir='./outputs/analyzer',
    categorical_columns= [
      "Sex",
      "T Stage",
      "N Stage",
      "Stage",
      "Smoking Status",
      "Disease Site",
      "HPV Combined",
      "Chemotherapy",
    ],
    continuous_columns = [
      "age at dx",
      "Dose"
    ],
    target_variable='Dose', 
    task='classification'
)
analyzer.encoding_module.enabled = False # AutoGluon will handle encoding

print(analyzer)

analyzer.run()

from jarvais.analyzer import Analyzer from rich import print analyzer = Analyzer( data=df, output_dir='./outputs/analyzer', categorical_columns= [ "Sex", "T Stage", "N Stage", "Stage", "Smoking Status", "Disease Site", "HPV Combined", "Chemotherapy", ], continuous_columns = [ "age at dx", "Dose" ], target_variable='Dose', task='classification' ) analyzer.encoding_module.enabled = False # AutoGluon will handle encoding print(analyzer) analyzer.run()

/home/joshua-siraj/Documents/CDI/jarvais/.pixi/envs/dev/lib/python3.12/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm
13:06:25 [warning  ] Date columns not specified. Inferring from remaining columns. [jarvais] call=analyzer.__init__:76

Analyzer(
    AnalyzerSettings(
        output_dir=PosixPath('outputs/analyzer'),
        categorical_columns=[
            'Sex',
            'T Stage',
            'N Stage',
            'Stage',
            'Smoking Status',
            'Disease Site',
            'HPV Combined',
            'Chemotherapy'
        ],
        continuous_columns=['age at dx', 'Dose'],
        date_columns=[],
        task='classification',
        target_variable='Dose',
        generate_report=True,
        settings_path=None,
        settings_schema_path=None,
        missingness=MissingnessModule(
            categorical_strategy={
                'Sex': 'unknown',
                'T Stage': 'unknown',
                'N Stage': 'unknown',
                'Stage': 'unknown',
                'Smoking Status': 'unknown',
                'Disease Site': 'unknown',
                'HPV Combined': 'unknown',
                'Chemotherapy': 'unknown'
            },
            continuous_strategy={'age at dx': 'median', 'Dose': 'median'},
            enabled=True
        ),
        outlier=OutlierModule(
            categorical_strategy={
                'Sex': 'frequency',
                'T Stage': 'frequency',
                'N Stage': 'frequency',
                'Stage': 'frequency',
                'Smoking Status': 'frequency',
                'Disease Site': 'frequency',
                'HPV Combined': 'frequency',
                'Chemotherapy': 'frequency'
            },
            continuous_strategy={'age at dx': 'none', 'Dose': 'none'},
            threshold=0.01,
            enabled=True,
            categorical_mapping={}
        ),
        encoding=OneHotEncodingModule(
            columns=[
                'Sex',
                'T Stage',
                'N Stage',
                'Stage',
                'Smoking Status',
                'Disease Site',
                'HPV Combined',
                'Chemotherapy'
            ],
            target_variable='Dose',
            prefix_sep='|',
            enabled=False
        ),
        visualization=VisualizationModule(
            plots=['corr', 'pairplot', 'umap', 'frequency_table', 'multiplot'],
            enabled=True
        )
    )
)

         [info     ] Performing missingness analysis... [jarvais] call=missingness.__call__:43
         [info     ] Performing outlier analysis... [jarvais] call=outlier.__call__:60
         [info     ] Plotting Correlation Matrix... [jarvais] call=visualization.__call__:115

+-----------------------+-------------------+-----------+-------------+
|                       |                   | Missing   | Overall     |
+=======================+===================+===========+=============+
| n                     |                   |           | 3346        |
+-----------------------+-------------------+-----------+-------------+
| age at dx, mean (SD)  |                   | 0         | 62.3 (11.6) |
+-----------------------+-------------------+-----------+-------------+
| Dose, mean (SD)       |                   | 0         | 66.7 (5.8)  |
+-----------------------+-------------------+-----------+-------------+
| Sex, n (%)            | Female            |           | 686 (20.5)  |
+-----------------------+-------------------+-----------+-------------+
|                       | Male              |           | 2660 (79.5) |
+-----------------------+-------------------+-----------+-------------+
| T Stage, n (%)        | None              |           | 12 (0.4)    |
+-----------------------+-------------------+-----------+-------------+
|                       | T0                |           | 167 (5.0)   |
+-----------------------+-------------------+-----------+-------------+
|                       | T1                |           | 454 (13.6)  |
+-----------------------+-------------------+-----------+-------------+
|                       | T1 (2)            |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | T1a               |           | 179 (5.3)   |
+-----------------------+-------------------+-----------+-------------+
|                       | T1b               |           | 88 (2.6)    |
+-----------------------+-------------------+-----------+-------------+
|                       | T2                |           | 927 (27.7)  |
+-----------------------+-------------------+-----------+-------------+
|                       | T2 (2)            |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | T2a               |           | 4 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | T2b               |           | 5 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | T3                |           | 861 (25.7)  |
+-----------------------+-------------------+-----------+-------------+
|                       | T3 (2)            |           | 3 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | T4                |           | 116 (3.5)   |
+-----------------------+-------------------+-----------+-------------+
|                       | T4a               |           | 358 (10.7)  |
+-----------------------+-------------------+-----------+-------------+
|                       | T4b               |           | 121 (3.6)   |
+-----------------------+-------------------+-----------+-------------+
|                       | TX                |           | 4 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | Tis               |           | 44 (1.3)    |
+-----------------------+-------------------+-----------+-------------+
|                       | rT0               |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
| N Stage, n (%)        | N0                |           | 1147 (34.3) |
+-----------------------+-------------------+-----------+-------------+
|                       | N1                |           | 344 (10.3)  |
+-----------------------+-------------------+-----------+-------------+
|                       | N2                |           | 182 (5.4)   |
+-----------------------+-------------------+-----------+-------------+
|                       | N2a               |           | 125 (3.7)   |
+-----------------------+-------------------+-----------+-------------+
|                       | N2b               |           | 791 (23.6)  |
+-----------------------+-------------------+-----------+-------------+
|                       | N2c               |           | 532 (15.9)  |
+-----------------------+-------------------+-----------+-------------+
|                       | N3                |           | 170 (5.1)   |
+-----------------------+-------------------+-----------+-------------+
|                       | N3a               |           | 13 (0.4)    |
+-----------------------+-------------------+-----------+-------------+
|                       | N3b               |           | 28 (0.8)    |
+-----------------------+-------------------+-----------+-------------+
|                       | NX                |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | None              |           | 13 (0.4)    |
+-----------------------+-------------------+-----------+-------------+
| Stage, n (%)          | 0                 |           | 44 (1.3)    |
+-----------------------+-------------------+-----------+-------------+
|                       | I                 |           | 352 (10.5)  |
+-----------------------+-------------------+-----------+-------------+
|                       | IB                |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | II                |           | 400 (12.0)  |
+-----------------------+-------------------+-----------+-------------+
|                       | IIA               |           | 2 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | IIB               |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | III               |           | 605 (18.1)  |
+-----------------------+-------------------+-----------+-------------+
|                       | IIIA              |           | 2 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | IIIC              |           | 2 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | IV                |           | 12 (0.4)    |
+-----------------------+-------------------+-----------+-------------+
|                       | IVA               |           | 1581 (47.3) |
+-----------------------+-------------------+-----------+-------------+
|                       | IVB               |           | 309 (9.2)   |
+-----------------------+-------------------+-----------+-------------+
|                       | IVC               |           | 2 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | None              |           | 27 (0.8)    |
+-----------------------+-------------------+-----------+-------------+
|                       | X                 |           | 6 (0.2)     |
+-----------------------+-------------------+-----------+-------------+
| Smoking Status, n (%) | Current           |           | 1139 (34.0) |
+-----------------------+-------------------+-----------+-------------+
|                       | Ex-smoker         |           | 1290 (38.6) |
+-----------------------+-------------------+-----------+-------------+
|                       | Non-smoker        |           | 872 (26.1)  |
+-----------------------+-------------------+-----------+-------------+
|                       | unknown           |           | 45 (1.3)    |
+-----------------------+-------------------+-----------+-------------+
| Disease Site, n (%)   | benign tumor      |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | esophagus         |           | 33 (1.0)    |
+-----------------------+-------------------+-----------+-------------+
|                       | hypopharynx       |           | 162 (4.8)   |
+-----------------------+-------------------+-----------+-------------+
|                       | lacrimal gland    |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | larynx            |           | 877 (26.2)  |
+-----------------------+-------------------+-----------+-------------+
|                       | lip & oral cavity |           | 100 (3.0)   |
+-----------------------+-------------------+-----------+-------------+
|                       | nasal cavity      |           | 62 (1.9)    |
+-----------------------+-------------------+-----------+-------------+
|                       | nasopharynx       |           | 355 (10.6)  |
+-----------------------+-------------------+-----------+-------------+
|                       | orbit             |           | 1 (0.0)     |
+-----------------------+-------------------+-----------+-------------+
|                       | oropharynx        |           | 1501 (44.9) |
+-----------------------+-------------------+-----------+-------------+
|                       | other             |           | 2 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | paraganglioma     |           | 7 (0.2)     |
+-----------------------+-------------------+-----------+-------------+
|                       | paranasal sinus   |           | 28 (0.8)    |
+-----------------------+-------------------+-----------+-------------+
|                       | salivary glands   |           | 4 (0.1)     |
+-----------------------+-------------------+-----------+-------------+
|                       | sarcoma           |           | 20 (0.6)    |
+-----------------------+-------------------+-----------+-------------+
|                       | skin              |           | 24 (0.7)    |
+-----------------------+-------------------+-----------+-------------+
|                       | unknown           |           | 168 (5.0)   |
+-----------------------+-------------------+-----------+-------------+
| HPV Combined, n (%)   | 1.0               |           | 1139 (34.0) |
+-----------------------+-------------------+-----------+-------------+
|                       | None              |           | 2207 (66.0) |
+-----------------------+-------------------+-----------+-------------+
| Chemotherapy, n (%)   | 0                 |           | 1923 (57.5) |
+-----------------------+-------------------+-----------+-------------+
|                       | 1                 |           | 1423 (42.5) |
+-----------------------+-------------------+-----------+-------------+

Outlier Report:
  - No Outliers found in Sex
  - Outliers found in T Stage: ['nan: 12 out of 3346', 'T2b: 5 out of 3346', 'T2a: 4 out of 3346', 'TX: 4 out of 3346', 'T3 (2): 3 out of 3346', 'T2 (2): 1 out of 3346', 'T1 (2): 1 out of 3346', 'rT0: 1 out of 3346']
  - Outliers found in N Stage: ['N3b: 28 out of 3346', 'N3a: 13 out of 3346', 'nan: 13 out of 3346', 'NX: 1 out of 3346']
  - Outliers found in Stage: ['nan: 27 out of 3346', 'IV: 12 out of 3346', 'X: 6 out of 3346', 'IIA: 2 out of 3346', 'IIIA: 2 out of 3346', 'IIIC: 2 out of 3346', 'IVC: 2 out of 3346', 'IB: 1 out of 3346', 'IIB: 1 out of 3346']
  - No Outliers found in Smoking Status
  - Outliers found in Disease Site: ['paranasal sinus: 28 out of 3346', 'skin: 24 out of 3346', 'sarcoma: 20 out of 3346', 'paraganglioma: 7 out of 3346', 'salivary glands: 4 out of 3346', 'other: 2 out of 3346', 'benign tumor: 1 out of 3346', 'lacrimal gland: 1 out of 3346', 'orbit: 1 out of 3346']
  - No Outliers found in HPV Combined
  - No Outliers found in Chemotherapy

         [info     ] Plotting Pairplot...           [jarvais] call=visualization.__call__:118
         [info     ] Plotting UMAP...               [jarvais] call=visualization.__call__:124
13:06:35 [info     ] Plotting Frequency Table...    [jarvais] call=visualization.__call__:121
13:06:41 [info     ] Plotting Multiplot...          [jarvais] call=visualization.__call__:136
13:06:45 [warning  ] One-hot encoding is disabled.  [jarvais] call=encoding.__call__:40
Font MPDFAA+Inter28ptBold is missing the following glyphs: '
' (\n)

In [3]:

from jarvais.trainer import TrainerSupervised

trainer = TrainerSupervised(
    output_dir="./outputs/trainer", 
    target_variable="Dose", 
    task="regression",
    k_folds=2
)
    
print(trainer)

trainer.run(analyzer.data)

from jarvais.trainer import TrainerSupervised trainer = TrainerSupervised( output_dir="./outputs/trainer", target_variable="Dose", task="regression", k_folds=2 ) print(trainer) trainer.run(analyzer.data)

TrainerSupervised(
    TrainerSettings(
        output_dir=PosixPath('outputs/trainer'),
        target_variable='Dose',
        task='regression',
        stratify_on=None,
        test_size=0.2,
        random_state=42,
        explain=False,
        reduction_module=FeatureReductionModule(method=None, task='regression', keep_k=2, enabled=True),
        trainer_module=AutogluonTabularWrapper(
            output_dir=PosixPath('outputs/trainer'),
            target_variable='Dose',
            task='regression',
            eval_metric='r2',
            k_folds=2,
            extra_metrics=['root_mean_squared_error'],
            kwargs={}
        )
    )
)

13:06:50 [info     ] Skipping feature reduction.    [jarvais] call=feature_reduction.__call__:39
         [info     ] Training fold 1/2...           [jarvais] call=autogluon_trainer._train_autogluon_with_cv:192
13:07:37 [info     ] Fold 1/2 score: 0.5817623676725117 (r2) [jarvais] call=autogluon_trainer._train_autogluon_with_cv:209
         [info     ] Training fold 2/2...           [jarvais] call=autogluon_trainer._train_autogluon_with_cv:192
13:08:14 [info     ] Fold 2/2 score: 0.5929594381641264 (r2) [jarvais] call=autogluon_trainer._train_autogluon_with_cv:209

Model Leaderboard
----------------
+-----------------------+----------------------------+----------------------------+----------------------------+
| model                 | score_train                | score_val                  | score_test                 |
+=======================+============================+============================+============================+
| CatBoost              | R2: 0.62 [0.62, 0.62]      | R2: 0.63 [0.6, 0.67]       | R2: 0.6 [0.6, 0.6]         |
|                       | RMSE: -3.54 [-3.54, -3.53] | RMSE: -3.55 [-3.72, -3.37] | RMSE: -3.66 [-3.66, -3.66] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| WeightedEnsemble_L2   | R2: 0.63 [0.63, 0.64]      | R2: 0.63 [0.59, 0.67]      | R2: 0.6 [0.59, 0.6]        |
|                       | RMSE: -3.49 [-3.5, -3.47]  | RMSE: -3.54 [-3.73, -3.34] | RMSE: -3.66 [-3.68, -3.63] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| LightGBMXT            | R2: 0.62 [0.61, 0.62]      | R2: 0.63 [0.6, 0.67]       | R2: 0.59 [0.57, 0.6]       |
|                       | RMSE: -3.56 [-3.58, -3.55] | RMSE: -3.54 [-3.71, -3.37] | RMSE: -3.71 [-3.77, -3.66] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| NeuralNetFastAI       | R2: 0.61 [0.6, 0.61]       | R2: 0.62 [0.58, 0.66]      | R2: 0.58 [0.58, 0.58]      |
|                       | RMSE: -3.61 [-3.65, -3.57] | RMSE: -3.6 [-3.8, -3.39]   | RMSE: -3.74 [-3.75, -3.73] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| LightGBM              | R2: 0.64 [0.62, 0.66]      | R2: 0.62 [0.56, 0.68]      | R2: 0.57 [0.56, 0.57]      |
|                       | RMSE: -3.45 [-3.55, -3.35] | RMSE: -3.59 [-3.89, -3.29] | RMSE: -3.79 [-3.81, -3.78] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| ExtraTreesMSE         | R2: 0.68 [0.68, 0.69]      | R2: 0.69 [0.53, 0.85]      | R2: 0.56 [0.55, 0.57]      |
|                       | RMSE: -3.25 [-3.28, -3.23] | RMSE: -3.12 [-3.99, -2.25] | RMSE: -3.82 [-3.87, -3.78] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| RandomForestMSE       | R2: 0.68 [0.68, 0.68]      | R2: 0.69 [0.54, 0.85]      | R2: 0.55 [0.53, 0.57]      |
|                       | RMSE: -3.26 [-3.27, -3.24] | RMSE: -3.13 [-3.99, -2.28] | RMSE: -3.87 [-3.96, -3.77] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| XGBoost               | R2: 0.64 [0.63, 0.65]      | R2: 0.63 [0.52, 0.74]      | R2: 0.54 [0.52, 0.55]      |
|                       | RMSE: -3.43 [-3.48, -3.38] | RMSE: -3.5 [-4.04, -2.97]  | RMSE: -3.93 [-3.99, -3.87] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| NeuralNetTorch        | R2: 0.52 [0.49, 0.55]      | R2: 0.54 [0.52, 0.56]      | R2: 0.51 [0.49, 0.53]      |
|                       | RMSE: -4.0 [-4.13, -3.87]  | RMSE: -3.96 [-4.03, -3.88] | RMSE: -4.04 [-4.14, -3.94] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| LightGBMLarge         | R2: 0.62 [0.61, 0.63]      | R2: 0.62 [0.47, 0.76]      | R2: 0.5 [0.5, 0.5]         |
|                       | RMSE: -3.54 [-3.59, -3.48] | RMSE: -3.55 [-4.25, -2.85] | RMSE: -4.08 [-4.09, -4.08] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| SimpleRegressionModel | R2: 0.4 [0.4, 0.4]         | R2: 0.41 [0.4, 0.41]       | R2: 0.4 [0.4, 0.41]        |
|                       | RMSE: -4.46 [-4.46, -4.46] | RMSE: -4.51 [-4.52, -4.5]  | RMSE: -4.46 [-4.47, -4.45] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| KNeighborsDist        | R2: -0.03 [-0.05, -0.02]   | R2: -0.04 [-0.31, 0.24]    | R2: -0.28 [-0.36, -0.21]   |
|                       | RMSE: -5.84 [-5.89, -5.8]  | RMSE: -5.9 [-6.7, -5.11]   | RMSE: -6.53 [-6.72, -6.35] |
+-----------------------+----------------------------+----------------------------+----------------------------+
| KNeighborsUnif        | R2: 0.04 [0.03, 0.06]      | R2: 0.05 [-0.06, 0.16]     | R2: -0.08 [-0.09, -0.08]   |
|                       | RMSE: -5.63 [-5.68, -5.58] | RMSE: -5.68 [-6.02, -5.35] | RMSE: -6.0 [-6.02, -5.99]  |
+-----------------------+----------------------------+----------------------------+----------------------------+

In [4]:

from jarvais.explainer import Explainer

sensitive_features = {k: trainer.X_test[k] for k in ['N Stage', 'Disease Site', 'Sex']}

exp = Explainer.from_trainer(trainer, sensitive_features=sensitive_features)
exp.run()

from jarvais.explainer import Explainer sensitive_features = {k: trainer.X_test[k] for k in ['N Stage', 'Disease Site', 'Sex']} exp = Explainer.from_trainer(trainer, sensitive_features=sensitive_features) exp.run()

⚠️  **Possible Bias Detected in N Stage** ⚠️

=== Subgroup Analysis for 'N Stage' Using OLS Regression ===

Model Statistics:
    R-squared:                  0.185
    F-statistic:                21.426
    F-statistic p-value:        0.0000
    AIC:                        3136.36
    Log-Likelihood:             -1560.18
Model Coefficients:
    +---------------+---------------+------------------+
    | Feature       |   Coefficient |   Standard Error |
    +===============+===============+==================+
    | const         |         1.973 |            0.135 |
    +---------------+---------------+------------------+
    | N Stage_N0    |         1.967 |            0.200 |
    +---------------+---------------+------------------+
    | N Stage_N1    |         0.792 |            0.319 |
    +---------------+---------------+------------------+
    | N Stage_N2    |        -1.354 |            0.370 |
    +---------------+---------------+------------------+
    | N Stage_N2a   |         0.345 |            0.438 |
    +---------------+---------------+------------------+
    | N Stage_N2b   |        -0.496 |            0.215 |
    +---------------+---------------+------------------+
    | N Stage_N2c   |        -0.080 |            0.257 |
    +---------------+---------------+------------------+
    | N Stage_N3    |        -1.069 |            0.391 |
    +---------------+---------------+------------------+
    | N Stage_Other |         1.867 |            0.747 |
    +---------------+---------------+------------------+

=== Subgroup Analysis for 'N Stage' using FairLearn ===

    +--------------------------+--------------------+-------------------+-------------------+-------------------+
    |                          | N0                 | N1                | N2                | N2a               |
    +==========================+====================+===================+===================+===================+
    | mean_prediction          | 63.368020066269885 | 67.93983979060732 | 69.96154375774104 | 68.39650208609444 |
    +--------------------------+--------------------+-------------------+-------------------+-------------------+
    | Relative mean_prediction | 1.000 ✅           | 1.072 ✅          | 1.104 ✅          | 1.079 ✅          |
    +--------------------------+--------------------+-------------------+-------------------+-------------------+ 

⚠️  **Possible Bias Detected in Disease Site** ⚠️

=== Subgroup Analysis for 'Disease Site' Using OLS Regression ===

Model Statistics:
    R-squared:                  0.199
    F-statistic:                20.589
    F-statistic p-value:        0.0000
    AIC:                        3126.11
    Log-Likelihood:             -1554.06
Model Coefficients:
    +--------------------------------+---------------+------------------+
    | Feature                        |   Coefficient |   Standard Error |
    +================================+===============+==================+
    | const                          |         2.945 |            0.154 |
    +--------------------------------+---------------+------------------+
    | Disease Site_Other             |         2.440 |            0.593 |
    +--------------------------------+---------------+------------------+
    | Disease Site_esophagus         |         3.561 |            0.799 |
    +--------------------------------+---------------+------------------+
    | Disease Site_hypopharynx       |        -0.843 |            0.400 |
    +--------------------------------+---------------+------------------+
    | Disease Site_larynx            |         0.716 |            0.231 |
    +--------------------------------+---------------+------------------+
    | Disease Site_lip & oral cavity |         2.104 |            0.508 |
    +--------------------------------+---------------+------------------+
    | Disease Site_nasal cavity      |        -1.011 |            0.612 |
    +--------------------------------+---------------+------------------+
    | Disease Site_nasopharynx       |        -2.398 |            0.306 |
    +--------------------------------+---------------+------------------+
    | Disease Site_oropharynx        |        -1.083 |            0.200 |
    +--------------------------------+---------------+------------------+
    | Disease Site_unknown           |        -0.543 |            0.379 |
    +--------------------------------+---------------+------------------+

=== Subgroup Analysis for 'Disease Site' using FairLearn ===

    +--------------------------+-------------------+-------------------+-------------------+-------------------+
    |                          | Other             | esophagus         | hypopharynx       | larynx            |
    +==========================+===================+===================+===================+===================+
    | mean_prediction          | 65.38273620605469 | 66.37863826751709 | 68.00474707285564 | 62.70861864664469 |
    +--------------------------+-------------------+-------------------+-------------------+-------------------+
    | Relative mean_prediction | 0.953 ✅          | 0.967 ✅          | 0.991 ✅          | 0.914 ✅          |
    +--------------------------+-------------------+-------------------+-------------------+-------------------+ 

⚠️  **Possible Bias Detected in Sex** ⚠️

=== Subgroup Analysis for 'Sex' Using OLS Regression ===

Model Statistics:
    R-squared:                  0.012
    F-statistic:                4.219
    F-statistic p-value:        0.0151
    AIC:                        3254.76
    Log-Likelihood:             -1624.38
Model Coefficients:
    +------------+---------------------+--------------------+
    | Feature    |         Coefficient |     Standard Error |
    +============+=====================+====================+
    | const      |  90656299303385.656 | 40069012658174.047 |
    +------------+---------------------+--------------------+
    | Sex_Female | -90656299303382.750 | 40069012658174.148 |
    +------------+---------------------+--------------------+
    | Sex_Male   | -90656299303383.500 | 40069012658174.156 |
    +------------+---------------------+--------------------+

=== Subgroup Analysis for 'Sex' using FairLearn ===

    +--------------------------+-------------------+------------------+
    |                          | Female            | Male             |
    +==========================+===================+==================+
    | mean_prediction          | 67.11062446876808 | 66.9930581921729 |
    +--------------------------+-------------------+------------------+
    | Relative mean_prediction | 1.002 ✅          | 1.000 ✅         |
    +--------------------------+-------------------+------------------+