upd

notebooks/deployment-assets/truthfinder.py

@@ -0,0 +1,67 @@
+def getrocprob(raw_scores,rocdata):
+    sample_prevalence=rocdata['positive_samples']/rocdata['total_samples']
+    prevalence = rocdata.get('prevalence',sample_prevalence)
+
+    from zedstat.zedstat import score_to_probability
+    prob=score_to_probability(raw_scores,df=rocdata['roc'],
+                         total_samples=rocdata['total_samples'],
+                         positive_samples=rocdata['positive_samples'],
+                         prevalence=prevalence)
+    return prob
+
+
+def _diss_linear(s, qnet, missing_response='',missing_diss_value=0):
+    import numpy as np
+    Ds = qnet.predict_distributions(s)
+    diss_values = []
+    for i in range(len(s)):
+        prob = float(Ds[i].get(str(s[i]),np.max(list(Ds[i].values())) ))
+        max_prob = max(Ds[i].values())
+        diss_value = 1 - prob / max_prob if max_prob != 0 else 0
+        diss_values.append(diss_value)
+    return diss_values
+
+def truthfinder(patients_responses, problem_type):
+    import pandas as pd
+    from quasinet.qnet import load_qnet
+    import pickle
+
+    patients_responses = [{patient_id: {question_id: str(int(response))
+                                         for question_id,
+                                         response in patient_responses.items()}
+                            for patient_id, patient_responses in patient.items()}
+                           for patient in patients_responses]
+
+
+    model_path = f"models/{problem_type}/random_order_full_model_0.joblib.gz"
+    classifier_path = f"classifiers/{problem_type}/runif-classifier-{346}.pkl"
+    roc_path =  f"classifiers/{problem_type}/runif-roc-{346}.pkl"
+    with open(roc_path, 'rb') as file:
+        rocdata = pickle.load(file)
+
+    model = load_qnet(model_path)
+    classifier = pd.read_pickle(classifier_path)
+    all_data_samples = []
+
+    for patient_response in patients_responses:
+        for patient_id, responses in patient_response.items():
+            resp_df = pd.DataFrame([responses], columns=model.feature_names)
+            data_samples = (resp_df
+                            .fillna('')  # Replace missing values with empty strings
+                            .astype(str)  # Convert all values to strings
+                            .values)
+            all_data_samples.append(data_samples[0])
+
+    diss_values = [_diss_linear(sample, model) for sample in all_data_samples]
+    proba = classifier.predict_proba(diss_values)[:,1]
+
+    estimated_probability_of_event = getrocprob(proba,rocdata)
+    output_dict = {}
+    for idx, patient_response in enumerate(patients_responses):
+        for patient_id in patient_response.keys():
+            output_dict[patient_id] = {'probability':estimated_probability_of_event[idx][0],
+                                       'ci': tuple(estimated_probability_of_event[idx][1:]),
+                                       'rawscore':proba[idx]} 
+    return output_dict
+
+

notebooks/deployment-assets/tscript1.py

@@ -0,0 +1,14 @@
+from truthfinder import truthfinder
+
+
+
+
+problem_type = 'global'
+
+patients_responses = [{"0": {"20": '1', "31": 2.0, "36": 1.0, "51": 2.0, "60": 1.0, "75": 1.0, "85": 1.0, "115": 1.0, "123": 2.0, "147": 2.0, "150": 3.0, "179": 2.0, "312": 2.0, "355": 2.0, "436": 4.0, "572": 3.0, "719": 2.0, "737": 2.0, "747": 2.0, "817": 2.0, "855": 2.0, "938": 1.0, "983": 2.0, "1053": 2.0, "1060": 2.0, "1069": 1.0, "1132": 2.0, "4201": 1.0, "4203": 1, "4247": 1.0, "4250": 2.0, "4251": 2.0, "4254": 2.0, "4255": 1, "4262": 2.0, "4278": 1, "4280": 1, "4287": 1, "4289": 1, "4439": 1.0, "4441": 1.0, "4442": 2.0, "4452": 1.0, "4453": 0, "4454": 0, "4455": 0, "4456": 0.0, "4525": 1, "4527": 1, "4529": 3.0, "4531": 1.0, "4532": 2.0, "4545": 3.0, "4553": 1.0, "4555": 1, "4571": 1.0, "4574": 2.0, "4587": 1.0}}, {"9": {"20": 2.0, "31": 1.0, "36": 1.0, "46": 1.0, "51": 1.0, "53": 1.0, "61": 2.0, "62": 2.0, "76": 1.0, "85": 1.0, "109": 3.0, "110": 4.0, "111": 3.0, "123": 2.0, "149": 4.0, "210": 3.0, "255": 2.0, "576": 1.0, "603": 4.0, "840": 2.0, "841": 2.0, "855": 2.0, "881": 2.0, "927": 2.0, "928": 2.0, "936": 2.0, "1009": 3.0, "1041": 3.0, "1080": 3.0, "1125": 2.0, "4203": 1, "4247": 1.0, "4248": 4.0, "4250": 2.0, "4251": 3.0, "4255": 3, "4262": 1.0, "4278": 1, "4280": 1, "4287": 1, "4289": 1, "4355": 3.0, "4439": 3.0, "4441": 3.0, "4451": 2.0, "4452": 3.0, "4453": 0, "4454": 0, "4455": 0, "4456": 0.0, "4525": 2, "4527": 1, "4554": 2.0, "4555": 3, "4586": 1.0, "4587": 3.0, "4590": 1.0, "4596": 3.0}}, {"PHX1": {"20": '1', "31": 2.0, "36": 1.0, "51": 2.0, "60": 1.0, "75": 1.0, "85": 1.0, "115": 1.0, "123": 2.0, "147": 2.0, "150": 3.0, "179": 2.0, "312": 2.0, "355": 2.0, "436": 4.0, "572": 3.0, "719": 2.0, "737": 2.0, "747": 2.0, "817": 2.0, "855": 2.0, "938": 1.0, "983": 2.0, "1053": 2.0, "1060": 2.0, "1069": 1.0, "1132": 2.0, "4201": 1.0, "4203": 1, "4247": 1.0, "4250": 2.0, "4251": 2.0, "4254": 2.0, "4255": 1, "4262": 2.0, "4278": 1, "4280": 1} }
+]             
+
+
+
+result = truthfinder(patients_responses, problem_type)
+print(result)