In [1]:

    

import sys 
import os
sys.path.append(os.path.abspath("/home/scidb/HeartRatePatterns/Python"))
from LogisticRegresion import ajustLogisticRegression
from PlotWords import plot_word
from Matrix import convert_matrix


    

In [2]:

    

def generateNMF(patients_train,survived_train,patients_test,survived_test,n_components=30) :
    nmf = NMF(n_components=n_components, random_state=1,alpha=.1, l1_ratio=.5)
    patients_trainnmf = nmf.fit_transform(patients_train)
    H = nmf.components_
    patients_testnmf = nmf.transform(patients_test)
    modelnmf = ajustLogisticRegression(patients_trainnmf,survived_train,patients_testnmf,survived_test)
    return nmf,modelnmf,patients_testnmf


    

In [3]:

    

def findpatient(message,table,beats):
    located = table
    groupsize = located.shape[0]
    deads = sum(located.index.labels[1].tolist())
    percent = -1 if groupsize==0 else deads/groupsize
    if groupsize!=0:
        print(message,"groupsize:%d"%groupsize,"deads:%d"%deads,"percent:{0:.2%}".format(percent),beats)


    

In [4]:

    

def print_top_words(model, feature_names, n_top_words,topics=[]):
    result = []
    for topic_idx, topic in enumerate(components):
        if(topics==[] or topic_idx in topics):
            features = [{"word":feature_names[i],"p1":str(topic[i])} for i in topic.argsort()[:-n_top_words - 1:-1] if topic[i] > 1]
            result.append({"topic":topic_idx,"features":features})
#        print(topic.argsort())
    return result


    

In [5]:

    

from operator import itemgetter
def sortBeats(compoT,table):
    predicts = modelnmf.predict_proba(compoT)[:,1]
    beats = list(table)
    n99,n90,n80,n70,alive = 0,0,0,0,0
    sortedBeats = []
    numPred = len(predicts)
    for i in range(0,numPred):
        sortedBeats.append({'word':beats[i],'predict':predicts[i]})
        if predicts[i]>.99 : n99+=1
        elif predicts[i]>.90 : n90+=1
        elif predicts[i]>.80 : n80+=1
        elif predicts[i]>.70 : n70+=1
        else: alive+=1
    print("over99:%d" %n99,"{0:.2%}".format(n99/numPred),"over90:%d" %n90,"{0:.2%}".format(n90/numPred),"over80:%d:"%n80,
          "{0:.2%}".format(n80/numPred),"over70:%d"%n70,"{0:.2%}".format(n70/numPred),"under:%d"%alive,
          "{0:.2%}".format(alive/numPred))
    return sorted(sortedBeats, key=itemgetter('predict'), reverse=True)


    

In [6]:

    

table = convert_matrix()


    

    




104899

In [7]:

    

table.sort_index(inplace=True,level='isAlive')
survived = table.index.labels[1].tolist()
patients = table.values


    

In [8]:

    

from sklearn.model_selection import train_test_split
patients_train, patients_test,survived_train, survived_test = train_test_split(patients,survived,test_size=0.2, random_state=42)


    

In [9]:

    

from sklearn.decomposition import NMF
n_components=32
nmf,modelnmf,pat_testnmf = generateNMF(patients,survived,patients_test,survived_test,n_components=n_components)


    

    




acurracy is 78.08%

In [10]:

    

import numpy as np
compoT = np.transpose(nmf.components_)
print("components",nmf.components_.shape)
print("components Transpose",compoT.shape)


    

    




components (32, 8141)
components Transpose (8141, 32)

In [11]:

    

sortedBeats = sortBeats(compoT,table)
beatKeys = []
for value in sortedBeats:
    beatKeys.append(value['word'])


    

    




over99:110 1.35% over90:245 3.01% over80:854: 10.49% over70:6910 84.88% under:22 0.27%

In [12]:

    

print("Top 5 HeartBeats con mayor probabilidad de muerte según la regresión logística")
plot_word(sortedBeats[:5])


    

    




Top 5 HeartBeats con mayor probabilidad de muerte según la regresión logística






    













    













    













    




dadad 3791422 3792816 mimic2wdb/matched/s18243/s18243-2876-08-09-02-08






    













    

In [13]:

    

table.sort_values(by=beatKeys)
survived = table.index.labels[1].tolist()
patients = table.values


    

In [14]:

    

nmf,modelnmf,pat_testnmf = generateNMF(patients,survived,patients_test,survived_test,n_components=n_components)


    

    




acurracy is 78.08%

In [15]:

    

components = nmf.components_


    

In [16]:

    

n_top_words = 1000
top_words = print_top_words(components, list(table), n_top_words)
for words in top_words:
    print(words['topic'],words['features'][:5])


    

    




0 [{'word': 'ababa', 'p1': '83.3795831434'}, {'word': 'babab', 'p1': '69.9040919323'}]
1 [{'word': 'eeeee', 'p1': '100.933239136'}]
2 [{'word': 'ddddd', 'p1': '49.827732745'}, {'word': 'dddda', 'p1': '1.87013299065'}, {'word': 'adddd', 'p1': '1.85109415342'}]
3 [{'word': 'babab', 'p1': '36.5795224104'}]
4 [{'word': 'adada', 'p1': '26.5294940537'}, {'word': 'aadad', 'p1': '26.3078393484'}, {'word': 'dadad', 'p1': '3.26250993487'}, {'word': 'adadd', 'p1': '3.21541750923'}, {'word': 'dadda', 'p1': '3.135075798'}]
5 [{'word': 'abbda', 'p1': '5.55199533967'}, {'word': 'aabbd', 'p1': '5.18572265572'}, {'word': 'dabba', 'p1': '4.98764711062'}, {'word': 'abbad', 'p1': '4.74384835818'}, {'word': 'bdaba', 'p1': '4.65397432675'}]
6 [{'word': 'eccec', 'p1': '8.55413286676'}, {'word': 'ccece', 'p1': '8.55019919686'}, {'word': 'cecce', 'p1': '7.79905128499'}, {'word': 'cecec', 'p1': '7.48322927909'}, {'word': 'ececc', 'p1': '7.20924681709'}]
7 [{'word': 'ddbdd', 'p1': '16.1053895409'}, {'word': 'dddbd', 'p1': '15.9988805109'}, {'word': 'dbddd', 'p1': '15.2860979168'}, {'word': 'ddddd', 'p1': '8.87416952252'}, {'word': 'dbddb', 'p1': '6.30913735171'}]
8 [{'word': 'adbad', 'p1': '9.88562431224'}, {'word': 'aadbd', 'p1': '9.14305514205'}, {'word': 'adbdb', 'p1': '8.47809123669'}, {'word': 'baaad', 'p1': '7.37618132876'}, {'word': 'dbadb', 'p1': '7.27706815108'}]
9 [{'word': 'badba', 'p1': '13.1861970854'}, {'word': 'dbaba', 'p1': '11.8112431932'}, {'word': 'babad', 'p1': '10.4574521189'}, {'word': 'adbab', 'p1': '9.58137442817'}, {'word': 'abadb', 'p1': '8.56687747367'}]
10 [{'word': 'bbdba', 'p1': '12.4450873202'}, {'word': 'bdbba', 'p1': '12.2308076614'}, {'word': 'abdbb', 'p1': '11.1191049559'}, {'word': 'bdbab', 'p1': '11.0753204384'}, {'word': 'dbabb', 'p1': '11.028461914'}]
11 [{'word': 'aadbb', 'p1': '15.7824793468'}, {'word': 'adbba', 'p1': '9.57274030452'}, {'word': 'dbbaa', 'p1': '9.02488664329'}, {'word': 'dbbba', 'p1': '4.51523039035'}, {'word': 'baadb', 'p1': '4.42887183418'}]
12 [{'word': 'gaaaa', 'p1': '7.84095851383'}, {'word': 'daaaf', 'p1': '7.68013837913'}, {'word': 'aafad', 'p1': '6.62954079793'}, {'word': 'aafda', 'p1': '6.60426804818'}, {'word': 'afdaa', 'p1': '6.53652679702'}]
13 [{'word': 'ceeee', 'p1': '11.199133659'}, {'word': 'eeeec', 'p1': '10.6203192531'}, {'word': 'eeece', 'p1': '7.2337697358'}, {'word': 'eceee', 'p1': '7.21242856027'}, {'word': 'eecee', 'p1': '7.01145463042'}]
14 [{'word': 'abdba', 'p1': '18.0027873851'}, {'word': 'dbaba', 'p1': '16.2671348814'}, {'word': 'babdb', 'p1': '14.6868626517'}, {'word': 'bdbab', 'p1': '12.7330063607'}, {'word': 'bdbaa', 'p1': '7.4057133884'}]
15 [{'word': 'adada', 'p1': '25.2207332715'}, {'word': 'dadad', 'p1': '19.769406118'}, {'word': 'aadad', 'p1': '6.42676350076'}, {'word': 'afada', 'p1': '4.84346324487'}, {'word': 'fadaa', 'p1': '4.21436880543'}]
16 [{'word': 'bbbaf', 'p1': '2.19767219757'}, {'word': 'bafbb', 'p1': '2.11928332071'}, {'word': 'bbafb', 'p1': '1.84580509241'}, {'word': 'bbabf', 'p1': '1.73817490217'}, {'word': 'babfb', 'p1': '1.65748579151'}]
17 [{'word': 'adada', 'p1': '5.84619135532'}, {'word': 'aadad', 'p1': '5.33519866287'}, {'word': 'dadda', 'p1': '5.27447430774'}, {'word': 'adadd', 'p1': '4.87057924271'}, {'word': 'dadad', 'p1': '4.84507701229'}]
18 [{'word': 'baddd', 'p1': '4.77587191079'}, {'word': 'adddb', 'p1': '4.44844853213'}, {'word': 'dbadd', 'p1': '3.84604198913'}, {'word': 'adbdd', 'p1': '3.52075102427'}, {'word': 'dadba', 'p1': '3.52057773397'}]
19 [{'word': 'gaaaa', 'p1': '6.2328215457'}, {'word': 'ffaaa', 'p1': '3.9843562554'}, {'word': 'aaaff', 'p1': '3.87434505235'}, {'word': 'affaa', 'p1': '3.7731895755'}, {'word': 'aaffa', 'p1': '3.67930811321'}]
20 [{'word': 'abdab', 'p1': '11.8851512538'}, {'word': 'bdaba', 'p1': '5.67419545978'}, {'word': 'dbdaa', 'p1': '5.41847042543'}, {'word': 'ddada', 'p1': '4.1398068018'}, {'word': 'abaad', 'p1': '4.13468453736'}]
21 [{'word': 'ababa', 'p1': '29.5017261625'}]
22 [{'word': 'afaaf', 'p1': '8.44157839598'}, {'word': 'faafa', 'p1': '8.13799580853'}, {'word': 'adaaf', 'p1': '6.52287684644'}, {'word': 'afafa', 'p1': '5.70101312463'}, {'word': 'aafaf', 'p1': '4.29529929812'}]
23 [{'word': 'dddad', 'p1': '21.7222261558'}, {'word': 'daddd', 'p1': '21.6074416588'}, {'word': 'ddadd', 'p1': '21.5111127673'}, {'word': 'adddd', 'p1': '15.4225610633'}, {'word': 'dddda', 'p1': '14.7358371116'}]
24 [{'word': 'dddaa', 'p1': '14.985823376'}, {'word': 'aaddd', 'p1': '14.3382208086'}, {'word': 'dddda', 'p1': '11.7640362966'}, {'word': 'adddd', 'p1': '11.2333368472'}, {'word': 'addda', 'p1': '8.11925278188'}]
25 [{'word': 'bbbad', 'p1': '12.7607345695'}, {'word': 'badbb', 'p1': '12.0110017711'}, {'word': 'bbadb', 'p1': '11.853504855'}, {'word': 'dbbba', 'p1': '5.26856668654'}, {'word': 'bbada', 'p1': '4.05658893464'}]
26 [{'word': 'ababa', 'p1': '19.1279995456'}, {'word': 'abada', 'p1': '11.3198199473'}, {'word': 'babad', 'p1': '6.53198889517'}, {'word': 'abadb', 'p1': '2.82391402401'}, {'word': 'aabdb', 'p1': '1.76280263332'}]
27 [{'word': 'baada', 'p1': '14.9470737552'}, {'word': 'abaad', 'p1': '13.3306159806'}, {'word': 'baaad', 'p1': '11.563949242'}, {'word': 'adaba', 'p1': '7.59071129479'}, {'word': 'dabaa', 'p1': '6.57263073511'}]
28 [{'word': 'aaabf', 'p1': '8.26824092507'}, {'word': 'aabfa', 'p1': '7.70754551544'}, {'word': 'abfaa', 'p1': '7.48910652397'}, {'word': 'abada', 'p1': '7.39534104273'}, {'word': 'bfaaa', 'p1': '6.79542883631'}]
29 [{'word': 'cbaba', 'p1': '18.1678705204'}, {'word': 'ababa', 'p1': '2.05172478335'}, {'word': 'acaad', 'p1': '1.5454186128'}, {'word': 'cdaaa', 'p1': '1.38251419567'}, {'word': 'babab', 'p1': '1.11611884761'}]
30 [{'word': 'dabaa', 'p1': '14.6538110948'}, {'word': 'adaba', 'p1': '14.5054068127'}, {'word': 'abada', 'p1': '7.88462008792'}, {'word': 'baaad', 'p1': '5.79325337182'}, {'word': 'abaad', 'p1': '5.39585103435'}]
31 [{'word': 'aaadf', 'p1': '5.4980513632'}, {'word': 'aadfa', 'p1': '4.97907541435'}, {'word': 'dfaaa', 'p1': '4.91383558782'}, {'word': 'adfaa', 'p1': '4.63948146102'}, {'word': 'aadaf', 'p1': '3.00363306505'}]

In [17]:

    

components.shape


    

    
Out[17]:





(32, 8141)

In [18]:

    

i,total=0,0
for component in components:
    i += 1
    total +=sum(component)
    print(sum(component),end=", ")
print(i,total,total/i)


    

    




156.014453956, 101.647953183, 64.2995970616, 41.4692622297, 100.268172403, 119.576277849, 106.988380676, 168.291148266, 163.995617664, 96.5267071592, 160.863534283, 83.8001253851, 287.009328669, 91.8928533187, 152.076609314, 299.287762979, 140.425721288, 245.027182227, 174.616937928, 211.51814498, 131.787886104, 33.1623721183, 130.419394411, 190.305361065, 139.814270584, 78.7529920798, 59.1057860796, 105.66600263, 161.614363501, 117.355064006, 92.2355202059, 180.935300432, 32 4386.75008403 137.085940126

In [19]:

    

patients_trainnmf = nmf.transform(patients_train)


    

In [20]:

    

from operator import itemgetter
from scipy.stats.stats import pearsonr
pearsonList = []
for i in range(n_components):
    patientpear=patients_trainnmf[:,i]
    pearson = pearsonr(patientpear,survived_train)
    if(pearson[0]>0):
        pearsonList.append({'word':i,'p1':pearson[0],'p2':pearson[1]})
sortedList = sorted(pearsonList, key=itemgetter('p1'), reverse=True)
for i in range(10):
    print(sortedList[i])


    

    




{'word': 16, 'p1': 0.085181283073763603, 'p2': 0.040119684397177162}
{'word': 19, 'p1': 0.054277890153705016, 'p2': 0.19139728746278908}
{'word': 15, 'p1': 0.054272542858434893, 'p2': 0.1914410991980286}
{'word': 23, 'p1': 0.053532551273255555, 'p2': 0.19757522859956847}
{'word': 17, 'p1': 0.05214486427164633, 'p2': 0.20946283901657331}
{'word': 5, 'p1': 0.043269161353460615, 'p2': 0.29778324427383251}
{'word': 31, 'p1': 0.043024823886840556, 'p2': 0.30052235395411697}
{'word': 14, 'p1': 0.042714810058433451, 'p2': 0.30402175084095573}
{'word': 25, 'p1': 0.042520641352779324, 'p2': 0.30622719689803041}
{'word': 9, 'p1': 0.041450126930469948, 'p2': 0.31857600078315185}

In [21]:

    

n_top_words = 10

topics = [sortedList[0]['word'],]
columns = list(table)
top_words = print_top_words(components, columns, 100,topics)
for words in top_words:
    print(words['topic'])
    features = words['features']
#print_top_beats(components, list(table), n_top_words,table,topics)


    

    




16

In [22]:

    

print("Top 5 HeartBeats con mayor correlación a la muerte según Pearson")
plot_word(features[:5])


    

    




Top 5 HeartBeats con mayor correlación a la muerte según Pearson






    













    













    













    













    

In [ ]: