hackathon_chat_symptoms.py

import nltk
from nltk.corpus import stopwords
from nltk.stem import PorterStemmer
from textblob import Word
import pandas as pd
import re
import numpy as np


from autocorrect import Speller
from textblob import TextBlob
from spellchecker import SpellChecker



from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity



spec = pd.read_excel('specialist list.xlsx')
symp = pd.read_csv('symptoms.csv')


spec['PROVIDER'] = spec['PROVIDER'].str.lower()
spec['SPECIALTY'] = spec['SPECIALTY'].str.lower()
spec['COUNTY'] = spec['COUNTY'].str.lower()

def make_diagnosis(inputText):


    sms_1 = inputText


    sms_2 = 'nairobi'


    #remove punctuations using regex
    sms_1 = re.sub('[\,\.]', '', sms_1)



    #Tokenizing using regex to convert the sentense into individual words

    sms_1 = re.split('\s+', sms_1)





    #Convert it to a dataframe using pandas

    sms_1_df =  pd.DataFrame({'sms_1':sms_1})



    #Removing stop words(words that carry no meaning)
    #Use the dataframe made using pandas

    #Removing stop words
    stop = stopwords.words('english')

    sms_1_df['sms_1_without_stop'] = sms_1_df['sms_1'].apply(lambda x: " ".join(x for x in x.split() if x not in stop))



    sms1_final = sms_1_df['sms_1_without_stop'].apply(lambda x: str(TextBlob(x).correct()))


    #Drop the empty rows from the dataframe
    pd.DataFrame(sms1_final)

    nan_value = float("NaN")

    sms1_final.replace("", nan_value, inplace=True)


    #Drop the empty rows from the dataframe with dropna

    sms1_final = sms1_final.dropna(axis='index')


    #print(sms1_final) - List of symptoms


    sms1_final = list(sms1_final)




    #Match the intents to the Kaggle data set

    diagnosis = symp['Disease'].where((symp['Symptom_1'] == sms1_final[1]) & (symp['Symptom_2'] == sms1_final[2]))
        



    diagnosis = diagnosis.dropna(axis='index')
    diagnosis = diagnosis.drop_duplicates( keep = 'first')



    diagnosis = list(diagnosis)
    return diagnosis


def get_specialists(diagnosis, county):


    #import dataset that matches specialists and ICD 10 diagnoses

    diag_spec = pd.read_csv("diag.spec.icd10.csv")



    #Match the diagnosis
    #Using using the top 3 options from diagnosis

    specialist_1 = diag_spec['dgns_cd'].where((diag_spec['longdesc'] == diagnosis[0]))
    specialist_2 = diag_spec['dgns_cd'].where((diag_spec['longdesc'] == diagnosis[1]))
    specialist_3 = diag_spec['dgns_cd'].where((diag_spec['longdesc'] == diagnosis[2]))


    specialist_1 = specialist_1.dropna(axis='index')
    specialist_1 = specialist_1.drop_duplicates( keep = 'first')




    specialist_2 = specialist_2.dropna(axis='index')
    specialist_2 = specialist_2.drop_duplicates( keep = 'first')
    #specialist_2 = str(specialist_2)[6:]




    specialist_3 = specialist_3.dropna(axis='index')
    specialist_3 = specialist_3.drop_duplicates( keep = 'first')
    #specialist_3 = str(specialist_3)[6:]

    name_spec='dermatologist'
    county = 'nairobi'
    #print(str.lower(str.strip(str(specialist_1)[6:])))
    #print(name_spec)




    #we will use loc from the pandas library to match the speciality with the diagnosis

    reply_1 = spec.loc[(spec['SPECIALTY'] == str(name_spec)) & (spec['COUNTY'] == str(county))]



    #print(f'\nHello, we have found {len(reply_1)} healthcare provividers/facilities that match your search for a dermatologist, reply Yes to get their location and contact details')

    return('Hello, we have found '+str(len(reply_1)) +' healthcare provividers/facilities that match your search for a dermatologist, reply Yes to get their location and contact details')

    #reply_2 = spec.loc[(spec['SPECIALTY'] == 'Physician') & (spec['COUNTY'] == 'nairobi')]



    #print(f'\nHello, we have found {len(reply_1)} facilities with  {specialist_1}s')

#print(get_specialists(make_diagnosis('I have been having some itchinnng and a rash')))
def otherFunc():
    #The first sms is to elicit the specialty or service
    #Use ambulance as a test example

    sms = input("Hello, welcome to the mdoc health facility and provider finder, please state the type of specialist you wish to see: ")


    #sms = 'dentistry'



    #This is an if statement that loops around the different professions which are broken into regex
    #This is a quick method of stemming and lemmatization of the word
    #the Intent is matched to choices available in our data set

    if re.match(r'air.', sms):
        intent = 'air ambulance'
    elif re.match(r'amb.', sms):
        intent = 'ambulance'
    elif re.match(r'an(a|e).', sms):
        intent = 'anaesthesia'
    elif re.match(r'ca.', sms):
        intent = 'cardiology'
    elif re.match(r'coun.', sms):
        intent = 'counseling'
    elif re.match(r'den.', sms):
        intent = 'dentistry'
    elif re.match(r'der.', sms):
        intent = 'dermatologist'
    elif re.match(r'endo.', sms):
        intent = 'endocrinology'
    elif re.match(r'ent.', sms):
        intent = 'ENT Surgery'
    elif re.match(r'fam.', sms):
        intent = 'family physician'
    elif re.match(r'gas.', sms):
        intent = 'gastroenterology'
    elif re.match(r'gen.', sms):
        intent = 'gentorologist'
    elif re.match(r'gy.', sms):
        intent = 'gynaecology'
    elif re.match(r'hem.', sms):
        intent = 'hemato-oncologist'
    elif re.match(r'hom.', sms):
        intent = 'home care services'
    elif re.match(r'hos.', sms):
        intent = 'hospital'
    elif re.match(r'ima.', sms):
        intent = 'imaging services'
    elif re.match(r'lab.', sms):
        intent = 'laboratory services'
    elif re.match(r'neo.', sms):
        intent = 'neonatology'
    elif re.match(r'nep.', sms):
        intent = 'nephrology'
    elif re.match(r'n(e|u).', sms):
        intent = 'neurologist'
    elif re.match(r'nutri.', sms):
        intent = 'nutritional services'
    elif re.match(r'ob.', sms):
        intent = 'obstestrian'
    elif re.match(r'oc.', sms):
        intent = 'occupational health specialist'
    elif re.match(r'o(p|f).', sms):
        intent = 'ophthalmology'
    elif re.match(r'max.', sms):
        intent = 'oral and maxillofacial surgeon'
    elif re.match(r'ort.', sms):
        intent = 'orthopaedics'
    elif re.match(r'onc.', sms):
        intent = 'oncology'
    elif re.match(r'opt.', sms):
        intent = 'optical services'
    elif re.match(r'path.', sms):
        intent = 'pathology'
    elif re.match(r'p(a|e).', sms):
        intent = 'paediatrician'
    elif re.match(r'physic.', sms):
        intent = 'physician'
    elif re.match(r'physio.', sms):
        intent = 'physiotherapy services'
    elif re.match(r'pl.', sms):
        intent = 'plastic surgeon'
    elif re.match(r'ps.', sms):
        intent = 'psychiatry'
    elif re.match(r'pul.', sms):
        intent = 'pulmonology'
    elif re.match(r'rad.', sms):
        intent = 'radiologist'
    elif re.match(r'rhe.', sms):
        intent = 'rheumatology'
    elif re.match(r'sur.', sms):
        intent = 'general surgery'
    elif re.match(r'ur.', sms):
        intent = 'urology'
    else:
        print("Search inconclusive, could you describe your symptoms?.")	



    #Check the intent in the list of healthcare facilities and specialities
    #we will use loc from the pandas library to search for rows with both conditions

    rslt_1 = spec.loc[(spec['SPECIALTY'] == intent) & (spec['COUNTY'] == 'nairobi')]

    rslt_1

    print(f'\nHello, we have found {len(rslt_1)} possible {intent} related service providers')



    print(rslt_1)