healthcare initial commit

healthcare/README.md

@@ -0,0 +1,7 @@
+# 1. Preferably use python 3.8
+# 2. Install the requirmements in your venv
+# 3. upload the healthcare documents into the vector store in heatwave. Sample documents can be found in the folder healthcare_docs
+# 4. install the hw_healthcare.sql file in the DB system.
+### 4.1. Please make sure you are using the correct vector store DB and table names in the procedure hw_healthcare
+# 5. Add the DB system user and pass in the mysql_helper file.
+# 6. Run the code: `streamlit run heatwave_heathcare.py`

healthcare/heatwave_heathcare.py

@@ -0,0 +1,33 @@
+from mysql_helper import run_mysql_queries
+import streamlit as st
+
+
+if __name__ == "__main__":
+
+    gif_path= "/home/opc/hw_hc/hw.png"
+    _left, mid, _right = st.columns(3)
+    with mid:
+        st.image(gif_path,  width=250)
+
+    engine = st.radio("", ["HeatWave GenAI with vector store (containing medicine articles)", "HeatWave GenAI without vector store (base LLM, Mistral only)"])
+    question = st.text_area("Enter your question:", "", key="big-textbox", height=100, help="Ask questions about healthcare: Allergies.")
+    if st.button("Answer"):
+        if engine=="HeatWave GenAI with vector store (containing medicine articles)":
+            query=f"CALL hw_healthcare('{question}');"
+            answer= run_mysql_queries(query)
+            st.write(f'<h5 style="color:green;">Answer of the question using Heatwave with vector store and RAG:</h5>', unsafe_allow_html=True)
+            st.write(answer[0])
+            st.write("\n\n")
+            with st.container():
+                st.write(f'<h5 style="color:black; margin-bottom: 0; padding-bottom: 0;">References:</h5>', unsafe_allow_html=True)
+                st.text_area('',answer[1], height=120)
+
+        elif engine == "HeatWave GenAI without vector store (base LLM, Mistral only)":
+            query=f"CALL ml_generate('{question}');"
+            answer= run_mysql_queries(query)
+            st.write(f'<h5 style="color:brown;">Answer of the question using Heatwave without vector store:</h5>', unsafe_allow_html=True)
+            st.write(answer[0])
+
+
+
+

healthcare/hw_healthcare.sql

@@ -0,0 +1,59 @@
+DELIMITER //
+
+CREATE PROCEDURE get_unique_document_names()
+BEGIN
+    DECLARE i INT DEFAULT 0;
+    DECLARE n INT;
+    DECLARE doc_name VARCHAR(5000);
+    DECLARE unique_doc_names TEXT DEFAULT '';
+    DECLARE counter INT DEFAULT 1;
+
+    SET n = JSON_LENGTH(@output, '$.citations');
+    SET unique_doc_names = "";
+
+    WHILE i < n DO
+        SET doc_name = JSON_UNQUOTE(JSON_EXTRACT(@output, CONCAT('$.citations[', i, '].document_name')));
+        IF POSITION(doc_name IN unique_doc_names) = 0 THEN
+            SET unique_doc_names = CONCAT(unique_doc_names, counter, '. ', doc_name, '\n\n');
+            SET counter = counter + 1;
+        END IF;
+        SET i = i + 1;
+    END WHILE;
+    SELECT TRIM(BOTH '\n' FROM unique_doc_names) AS llm_output;
+END //
+
+DELIMITER ;
+
+
+DELIMITER //
+
+CREATE PROCEDURE hw_healthcare(IN prompt VARCHAR(2048))
+BEGIN
+
+CALL sys.ML_MODEL_LOAD('mistral-7b-instruct-v1', NULL);
+
+CALL sys.ML_RAG(prompt, @output,
+'{"vector_store": ["`genai_health_db`.`health_vs3`"],
+"n_citations": 5, "distance_metric": "COSINE",
+"model_options": {"temperature": 0, "repeat_penalty": 1,
+"top_p": 0.2, "max_tokens": 400 , "model_id": "mistral-7b-instruct-v1"}}');
+
+SELECT JSON_UNQUOTE(JSON_EXTRACT(@output, '$.text'));
+CALL get_unique_document_names();
+END //
+
+DELIMITER ;
+
+
+DELIMITER //
+
+CREATE PROCEDURE ml_generate(IN prompt VARCHAR(2048))
+BEGIN
+
+CALL sys.ML_MODEL_LOAD('mistral-7b-instruct-v1', NULL);
+SELECT sys.ML_GENERATE(prompt, '{"task": "generation", "temperature": 0, "repeat_penalty": 1, "top_p": 0.2, "max_tokens": 400, "model_id": "mistral-7b-instruct-v1"}')
+INTO @llm_output;
+SELECT JSON_UNQUOTE(JSON_EXTRACT(@llm_output, '$.text'));
+END //
+
+DELIMITER ;

healthcare/mysql_helper.py

@@ -0,0 +1,64 @@
+# coding: utf-8
+# Copyright (c) 2023, Oracle and/or its affiliates.  All rights reserved.
+import mysql.connector
+import os
+import logging
+import sys
+import streamlit as st
+USERNAME=""
+PASSWORD=""
+DEFAULT_HOST=""
+PORT="3306"
+DATABASE=""
+def setup_logging():
+    logging.getLogger().setLevel(logging.INFO)
+    logger = logging.getLogger(__name__)
+    if not logger.hasHandlers():
+        file = logging.FileHandler('genai_autopilot.log')
+        file.setFormatter(logging.Formatter('%(asctime)s %(levelname)-8s %(message)s', datefmt='%Y-%m-%d %H:%M:%S'))
+        file.setLevel(logging.INFO)
+        logger.addHandler(file)
+        console = logging.StreamHandler(sys.stdout)
+        console.setLevel(logging.WARNING)
+        logger.addHandler(console)
+    return logger
+
+logger = setup_logging()
+
+# Connection to the MySQL server
+def mysql_connect(username=USERNAME, password=PASSWORD, host=DEFAULT_HOST, database=DATABASE, port=PORT, connection_timeout=1, repeat=5):
+    for i in range(repeat):
+        try:
+            return mysql.connector.connect(
+                user=username,
+                password=password,
+                host=host,
+                port=port,
+                autocommit=True,
+                database=database,
+                ssl_disabled=False,
+                use_pure=True,
+                connection_timeout=connection_timeout
+            )
+        except mysql.connector.Error as err:
+            logger.warning(f"Can't connect to the backend MySQL instance ({i}): {err}")
+
+
+@st.cache_data
+def run_mysql_queries(query):
+    output = []
+    try:
+        connection = mysql_connect()
+        cursor = connection.cursor()
+        for cursor_result in cursor.execute(query, multi=True):
+            for row in cursor_result:
+                if len(row) > 0:
+                    output.append(row[0] if len(row) == 1 else row)
+        cursor.close()
+        connection.close()
+        return output
+    except Exception as e:
+        logger.info(f"Problem with query: {query}")
+        logger.warning(f"Problem with MySQL connection during execution: {str(e)}")
+
+

healthcare/requirements.txt

@@ -0,0 +1,2 @@
+mysql-connector-python
+streamlit