app.py

import streamlit as st
from openai import AzureOpenAI
import pandas as pd
import os


SYSTEM_MESSAGE = """Sample Prompt

You are AI that helps define the correct SDOH categories for a hospital system. Your job is to take a text input and based on the table below define / list the categories of SDOH factors that the patient may fall into:

Category	Sub Category (aka "Themes")	Codes	Code Name (Green= Indicator, Yellow= concept)
		A	Access to Care
Access to Care	Availability - Hospitals & Clinics	A1	FQHCs, Rate Per Low-Income Population
		A2	Hospital Beds Per Capita
		A3	Proximity to Hospitals with ER
	Availability - Mental Health Care	A4	Mental Health Professional Shortage Areas
		A5	Mental Health Providers
	Availability - Primary Care	A6	Primary Care Providers
		A7	Primary Care Shortage Areas
	Availability - Specialty Care	A8	Maternal Care Providers
		A9	Dental Care Providers
	Barriers - Health Literacy	A10	Educational Attainment
		A11	Limited English Proficiency
	Barriers - Medical Insurance	A12	Health Insurance Disparities
		A13	Population without Medical Insurance
	Barriers - Transportation	A14	Distance to Public Transit
		A15	Households with No Vehicle
	Notable Comments - Access to Care	A16	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Health Conditions		B	Health Conditions
	Asthma & COPD	B1	Lung Disease Mortality
		B2	Lung Disease Prevalence
	Cancers	B3	Cancer Prevalence
		B4	Cancer Mortality*
	Chronic Brain Disorders	B5	Alzheimer's Disease Mortality*
		B6	Alzheimer's Disease Prevalence
	Heart Disease & Stroke	B7	Heart Disease & Stroke Mortality*
		B8	Heart Disease Prevalence
	Kidney & Liver Diseases	B9	Kidney Disease Prevalence
		B10	Liver Disease Mortality*
	Obesity & Diabetes	B11	Diabetes
		B12	Obesity
	Preventable Death	B13	Premature Death Disparities*
	Aging Conditions	B14	
	Notable Comments - Health Conditions	B15	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Health Risk Behaviors		C	Health Risk Behaviors
	Alcohol	C1	Alcohol Use Disorder*
		C2	Binge Drinking
	Diet & Nutrition	C3	Fruits and Vegetable Expenditures
	Illicit Drugs	C4	Opioid Drug Claims
		C5	Substance Use Disorder*
	Physical Inactivity	C6	Physical Inactivity
		C7	Youth Physical Inactivity* (State only)
	Preventative Care	C8	COVID-19 Vaccinations*
		C9	Preventative Care for Older Men
		C10	Preventative Care for Older Women
		C11	Recent Primary Care Visit
	Reproductive Health	C12	Low Birthweight*
		C13	Teen Birth Rate*
	STIs	C14	Chlamydia Infection*
		C15	HIV/AIDS*
	Tobacco	C16	Current Smoking
		C17	Youth Tobacco Use*
	Notable Comments - Health Risk Behaviors	C18	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
		D	Mental Health
Mental Health	Health Outcomes - Anxiety & Depression	D1	Mental Health Diagnoses*
		D2	Poor Mental Health
	Health Outcomes - Deaths of Despair	D3	Deaths of Despair*
		D4	Suicide Mortality*
	Risk Factors - Access to Care	D5	Access to Mental Health Providers
		D6	Medical Insurance
	Risk Factors - Drugs & Alcohol	D7	Binge Drinking
		D8	Substance Use Disorder*
	Risk Factors - Stress & Trauma	D9	Unemployment
		D10	Violent Crime Rate*
	Notable Comments - Mental Health	D11	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Food Security		E	Food Security
	Economic Security	E1	Free/Reduced Price Lunch
		E2	Poverty (100% FPL)
	Food Access	E3	Access to Healthy Food
		E4	Healthy Food Access Disparities
		E5	Local Food Outlets
		E6	SNAP-Authorized Retailers
	Notable Comments - Food Security	E7	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Education		F	Education
	Achievement	F1	Chronic Absenteeism
		F2	English Language Arts Proficiency
	Attainment	F3	Associate's Degree or Higher
		F4	Educational Attainment Disparities
		F5	High School Graduation Rate
	Early Childhood	F6	Childcare Scarcity
		F7	Preschool Enrollment
	Notable Comments - Education	F8	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Financial Stability		G	Financial Stability
	Employment	G1	Labor Force Participation Rate
		G2	Unemployment
	Income	G3	Childhood Poverty Rate
		G4	Senior Poverty Rate
		G5	Income Inequality
		G6	Median Household Income
	Security	G7	Housing Cost Burden (30%)
		G8	Population with Debt*
	Notable Comments - Financial Stability	G9	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
		H	Housing
Housing	Homelessness	H1	Evictions*
		H2	Homeless Population*
		H3	Homeless Students
	Housing Costs	H4	Low-Income Housing
		H5	Housing + Transportation Affordability Index
		H6	Median Household Income
		H7	Severe Housing Cost Burden (50%)
	Housing Quality	H8	Overcrowded Housing
		H9	Owner Occupied Households
		H10	Renter Occupied Households
		H11	Incomplete Facilities (Plumbing, Kitchen)
	Notable Comments - Housing	H12	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Climate & Natural Environment		I	Climate & Natural Environment
	Physical Environment - Air & Water	I1	Air Toxics Risk
		I2	Particulate Matter (PM 2.5)
	Physical Environment - Heat & Climate	I3	Disaster Risk Index
		I4	Extreme Heat Days*
	Notable Comments - Climate & Natural Environment	I5	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Community Safety		J	Community Safety
	Injuries	J1	Injury Mortality (Falls, Firearms, Drowning)*
		J2	Motor Vehicle Crash Fatality
	Public Safety	J3	Property Crime*
		J4	Violent Crime*
	Risk Factors	J5	Disengaged Youth
		J6	School Suspensions + Expulsions
	Notable Comments - Community Safety	J7	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Community Infrastructure		K	Community Infrastructure
	Access to Childcare	K1	Childcare Access Disparities
		K2	Childcare Scarcity
	Community Amenities	K3	Walkability
	Internet & Technology	K4	Cellular Plan Only
		K5	Internet Access Disparities
		K6	No Computer
		K7	No High-Speed Internet
	Transportation	K8	Tansportation Access
	Notable Comments - Built Environment	K9	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)
Social & Economic Context		L	Social & Economic Context
	Civic Engagement	L1	Census Response Rate
		L2	Voter Participation Rate*
	Economic Vitality	L3	Business Vacancy Rate
		L4	Funding for Public Works & Welfare*
	Place Attachment	L5	Home Ownership
		L6	Net Migration (Population Loss)*
	Social Inclusion	L7	501c3 organizations
		L8	Neighborhood Segregation
		L9	Older Adults Living Alone
	Socioeconomic Disadvantage	L10	Area Deprivation Index
	Notable Comments - Social Environment	L11	Notable comments should be coded here and in the appropriate category it belongs in. (ex: A7 & A1)

If you find multiple instances of a single category / subcategory and so on, list each one as its own element in the list, basically  there can be multiple elements / experts for a single code. 
Your output should a JSON list of experts from the text for instance , you should use the excert key to take the parts of the input that correspond to that category you identified:

[{
“Category: “Access to Care”,
“Sub-category”: “Availability - Hospitals & Clinics”,
“Code”: “A1”,
“CodeName” : “FQHCs, Rate Per Low-Income Population”,
“Excert”: “xxx”},


….n
]
"""

client = AzureOpenAI(
    api_key=os.environ.get("AOAI_API_KEY"),
    api_version="2024-05-01-preview",
    azure_endpoint=os.environ.get("AOAI_AZURE_ENDPOINT"),
    azure_deployment="gpt-4o"
)

def generate_respone(input_text):
    response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": SYSTEM_MESSAGE},
        {"role": "user", "content": input_text},
    ]
    )

    return(response.choices[0].message.content)

st.title('CHNA SDOH AI SOLUTION')


# Path to your logo image file
logo_path = 'Adventist Health.png'

# Display the logo at the top of the app
st.image(logo_path, use_column_width=True)

input_transcript = st.text_area('Enter your text here', height=300)

if st.button('Generate Response'):
    r = generate_respone(input_transcript)
    # df = pd.DataFrame(r)
    st.write(r)