Predicted Results from Brain Tumor Survival Model are Binary – How to Interpret?

I’m using a ‘Brain Tumor Overall Survival Prediction’ using BraTS20 Dataset.
The output from the model is a binary array, but how to predict the survival time of a patient?

Why are the predicted results binary?
How can I interpret these binary results in the context of brain tumor survival time?
Are there any specific adjustments I need to make to my model or post-processing steps?

I want to predict the survival time duration when I give an MRI input to the model, I have to design a web application for that, where the user gives the MRI to get the Overall Survival Time.

Here is where the data is converted into arrays

def getListAgeDays(id_list):   # create only age: category data

    x_val = []  # Initialize an empty list to store feature data
    y_val = []  # Initialize an empty list to store category labels
    for i in id_list:
        if (i not in age_dict):  # Check if 'i' exists in the 'age_dict' dictionary
            continue
        masks = getMaskSizesForVolume(nib.load(TRAIN_DATASET_PATH + f'\\BraTS20_Training_{i[-3:]}/BraTS20_Training_{i[-3:]}_seg.nii').get_fdata())
        # Load a segmentation mask and extract its size information
        brain_vol = getBrainSizeForVolume(nib.load(TRAIN_DATASET_PATH + f'\\BraTS20_Training_{i[-3:]}/BraTS20_Training_{i[-3:]}_t1.nii').get_fdata())
        # Load a brain volume image and extract its size information
        masks[1] = masks[1]/brain_vol  # Normalize the size of the mask
        masks[2] = masks[2]/brain_vol
        masks[3] = masks[3]/brain_vol
        merged = [age_dict[i], masks[1], masks[2], masks[3]]
        # Create a feature vector by combining 'age_dict[i]' with mask size values
        radiomics_values = df_radiomics.loc[df_radiomics['target'] == str(i)]
        # Query a DataFrame 'df_radiomics' for radiomics values associated with 'i'
        if radiomics_values.empty:
            continue  # Skip this iteration if there are no radiomics values
        merged.extend(radiomics_values.values.tolist()[0][:-1])
        # Add radiomics values to the feature vector 'merged' excluding the last value
        x_val.append(merged)  # Append the feature vector to 'x_val'
        if (days_dict[i] < 250):
            y_val.append([1, 0, 0])  # Append category labels based on 'days_dict' conditions
        elif (days_dict[i] >= 250 and days_dict[i] < 450):
            y_val.append([0, 1, 0])
        else:
            y_val.append([0, 0, 1])
    return np.array(x_val), np.array(y_val)  # Return feature data and category labels as NumPy arrays

X_all, y_all = getListAgeDays(brats_ids)  # Call the function with 'train_and_test_ids'
print(f'X_test: {X_all.shape}')  # Print the shape of the feature data

Here’s the X_test going into the model:

array([[0.424, 0.385, 0.725, 0.262, 0.304, 0.352, 0.555, 0.368, 0.276,
        0.531, 0.054, 0.286, 0.337, 0.447, 0.464, 0.334, 0.235, 0.607,
        0.789, 0.569, 0.849],
       [0.691, 0.066, 0.026, 0.165, 0.019, 0.063, 0.209, 0.81 , 0.   ,
        0.419, 0.   , 1.   , 0.383, 0.335, 0.36 , 0.392, 0.335, 0.429,
        0.381, 0.255, 0.257]....

Saved the model in Joblib

joblib.dump(model, 'model_joblib')
mj = joblib.load('model_joblib')

Here are the predictions

mj.predict(X_test)

output is:

array([0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0])