my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	---
#2	title: Healthcare Coach with ADK
#3	description: "Guide patients with an assistant that remembers history across ADK sessions."
#4	---
#5
#6
#7	This example demonstrates how to build a healthcare assistant that remembers patient information across conversations using Google ADK and Mem0.
#8
#9	## Overview
#10
#11	The Healthcare Assistant helps patients by:
#12	- Remembering their medical history and symptoms
#13	- Providing general health information
#14	- Scheduling appointment reminders
#15	- Maintaining a personalized experience across conversations
#16
#17	By integrating Mem0's memory layer with Google ADK, the assistant maintains context about the patient without requiring them to repeat information.
#18
#19	## Setup
#20
#21	Before you begin, make sure you have:
#22
#23	Installed Google ADK and Mem0 SDK:
#24	```bash
#25	pip install google-adk mem0ai python-dotenv
#26	```
#27
#28	## Code Breakdown
#29
#30	Let's get started and understand the different components required in building a healthcare assistant powered by memory
#31
#32	```python
#33	# Import dependencies
#34	import os
#35	import asyncio
#36	from google.adk.agents import Agent
#37	from google.adk.runners import Runner
#38	from google.adk.sessions import InMemorySessionService
#39	from google.genai import types
#40	from mem0 import MemoryClient
#41	from dotenv import load_dotenv
#42
#43	load_dotenv()
#44
#45	# Set up environment variables
#46	# os.environ["GOOGLE_API_KEY"] = "your-google-api-key"
#47	# os.environ["MEM0_API_KEY"] = "your-mem0-api-key"
#48
#49	# Define a global user ID for simplicity
#50	USER_ID = "Alex"
#51
#52	# Initialize Mem0 client
#53	mem0 = MemoryClient()
#54	```
#55
#56	## Define Memory Tools
#57
#58	First, we'll create tools that allow our agent to store and retrieve information using Mem0:
#59
#60	```python
#61	def save_patient_info(information: str) -> dict:
#62	"""Saves important patient information to memory."""
#63
#64	# Store in Mem0
#65	response = mem0_client.add(
#66	[{"role": "user", "content": information}],
#67	user_id=USER_ID,
#68	run_id="healthcare_session",
#69	metadata={"type": "patient_information"}
#70	)
#71
#72
#73	def retrieve_patient_info(query: str) -> dict:
#74	"""Retrieves relevant patient information from memory."""
#75
#76	# Search Mem0
#77	results = mem0_client.search(
#78	query,
#79	user_id=USER_ID,
#80	limit=5,
#81	threshold=0.7 # Higher threshold for more relevant results
#82	)
#83
#84	# Format and return the results
#85	if results and len(results) > 0:
#86	memories = [memory["memory"] for memory in results.get('results', [])]
#87	return {
#88	"status": "success",
#89	"memories": memories,
#90	"count": len(memories)
#91	}
#92	else:
#93	return {
#94	"status": "no_results",
#95	"memories": [],
#96	"count": 0
#97	}
#98	```
#99
#100	## Define Healthcare Tools
#101
#102	Next, we'll add tools specific to healthcare assistance:
#103
#104	```python
#105	def schedule_appointment(date: str, time: str, reason: str) -> dict:
#106	"""Schedules a doctor's appointment."""
#107	# In a real app, this would connect to a scheduling system
#108	appointment_id = f"APT-{hash(date + time) % 10000}"
#109
#110	return {
#111	"status": "success",
#112	"appointment_id": appointment_id,
#113	"confirmation": f"Appointment scheduled for {date} at {time} for {reason}",
#114	"message": "Please arrive 15 minutes early to complete paperwork."
#115	}
#116	```
#117
#118	## Create the Healthcare Assistant Agent
#119
#120	Now we'll create our main agent with all the tools:
#121
#122	```python
#123	# Create the agent
#124	healthcare_agent = Agent(
#125	name="healthcare_assistant",
#126	model="gemini-1.5-flash", # Using Gemini for healthcare assistant
#127	description="Healthcare assistant that helps patients with health information and appointment scheduling.",
#128	instruction="""You are a helpful Healthcare Assistant with memory capabilities.
#129
#130	Your primary responsibilities are to:
#131	1. Remember patient information using the 'save_patient_info' tool when they share symptoms, conditions, or preferences.
#132	2. Retrieve past patient information using the 'retrieve_patient_info' tool when relevant to the current conversation.
#133	3. Help schedule appointments using the 'schedule_appointment' tool.
#134
#135	IMPORTANT GUIDELINES:
#136	- Always be empathetic, professional, and helpful.
#137	- Save important patient information like symptoms, conditions, allergies, and preferences.
#138	- Check if you have relevant patient information before asking for details they may have shared previously.
#139	- Make it clear you are not a doctor and cannot provide medical diagnosis or treatment.
#140	- For serious symptoms, always recommend consulting a healthcare professional.
#141	- Keep all patient information confidential.
#142	""",
#143	tools=[save_patient_info, retrieve_patient_info, schedule_appointment]
#144	)
#145	```
#146
#147	## Set Up Session and Runner
#148
#149	```python
#150	# Set up Session Service and Runner
#151	session_service = InMemorySessionService()
#152
#153	# Define constants for the conversation
#154	APP_NAME = "healthcare_assistant_app"
#155	USER_ID = "Alex"
#156	SESSION_ID = "session_001"
#157
#158	# Create a session
#159	session = session_service.create_session(
#160	app_name=APP_NAME,
#161	user_id=USER_ID,
#162	session_id=SESSION_ID
#163	)
#164
#165	# Create the runner
#166	runner = Runner(
#167	agent=healthcare_agent,
#168	app_name=APP_NAME,
#169	session_service=session_service
#170	)
#171	```
#172
#173	## Interact with the Healthcare Assistant
#174
#175	```python
#176	# Function to interact with the agent
#177	async def call_agent_async(query, runner, user_id, session_id):
#178	"""Sends a query to the agent and returns the final response."""
#179	print(f"\n>>> Patient: {query}")
#180
#181	# Format the user's message
#182	content = types.Content(
#183	role='user',
#184	parts=[types.Part(text=query)]
#185	)
#186
#187	# Set user_id for tools to access
#188	save_patient_info.user_id = user_id
#189	retrieve_patient_info.user_id = user_id
#190
#191	# Run the agent
#192	async for event in runner.run_async(
#193	user_id=user_id,
#194	session_id=session_id,
#195	new_message=content
#196	):
#197	if event.is_final_response():
#198	if event.content and event.content.parts:
#199	response = event.content.parts[0].text
#200	print(f"<<< Assistant: {response}")
#201	return response
#202
#203	return "No response received."
#204
#205	# Example conversation flow
#206	async def run_conversation():
#207	# First interaction - patient introduces themselves with key information
#208	await call_agent_async(
#209	"Hi, I'm Alex. I've been having headaches for the past week, and I have a penicillin allergy.",
#210	runner=runner,
#211	user_id=USER_ID,
#212	session_id=SESSION_ID
#213	)
#214
#215	# Request for health information
#216	await call_agent_async(
#217	"Can you tell me more about what might be causing my headaches?",
#218	runner=runner,
#219	user_id=USER_ID,
#220	session_id=SESSION_ID
#221	)
#222
#223	# Schedule an appointment
#224	await call_agent_async(
#225	"I think I should see a doctor. Can you help me schedule an appointment for next Monday at 2pm?",
#226	runner=runner,
#227	user_id=USER_ID,
#228	session_id=SESSION_ID
#229	)
#230
#231	# Test memory - should remember patient name, symptoms, and allergy
#232	await call_agent_async(
#233	"What medications should I avoid for my headaches?",
#234	runner=runner,
#235	user_id=USER_ID,
#236	session_id=SESSION_ID
#237	)
#238
#239	# Run the conversation example
#240	if __name__ == "__main__":
#241	asyncio.run(run_conversation())
#242	```
#243
#244	## How It Works
#245
#246	This healthcare assistant demonstrates several key capabilities:
#247
#248	1. Memory Storage: When Alex mentions her headaches and penicillin allergy, the agent stores this information in Mem0 using the `save_patient_info` tool.
#249
#250	2. Contextual Retrieval: When Alex asks about headache causes, the agent uses the `retrieve_patient_info` tool to recall her specific situation.
#251
#252	3. Memory Application: When discussing medications, the agent remembers Alex's penicillin allergy without her needing to repeat it, providing safer and more personalized advice.
#253
#254	4. Conversation Continuity: The agent maintains context across the entire conversation session, creating a more natural and efficient interaction.
#255
#256	## Key Implementation Details
#257
#258	## User ID Management
#259
#260	Instead of passing the user ID as a parameter to the memory tools (which would require modifying the ADK's tool calling system), we attach it directly to the function object:
#261
#262	```python
#263	# Set user_id for tools to access
#264	save_patient_info.user_id = user_id
#265	retrieve_patient_info.user_id = user_id
#266	```
#267
#268	Inside the tool functions, we retrieve this attribute:
#269
#270	```python
#271	# Get user_id from session state or use default
#272	user_id = getattr(save_patient_info, 'user_id', 'default_user')
#273	```
#274
#275	This approach allows our tools to maintain user context without complicating their parameter signatures.
#276
#277	## Mem0 Integration
#278
#279	The integration with Mem0 happens through two primary functions:
#280
#281	1. `mem0_client.add()` - Stores new information with appropriate metadata
#282	2. `mem0_client.search()` - Retrieves relevant memories using semantic search
#283
#284	The `threshold` parameter in the search function ensures that only highly relevant memories are returned.
#285
#286	## Conclusion
#287
#288	This example demonstrates how to build a healthcare assistant with persistent memory using Google ADK and Mem0. The integration allows for a more personalized patient experience by maintaining context across conversation turns, which is particularly valuable in healthcare scenarios where continuity of information is crucial.
#289
#290	By storing and retrieving patient information intelligently, the assistant provides more relevant responses without requiring the patient to repeat their medical history, symptoms, or preferences.
#291
#292	---
#293
#294	<CardGroup cols={2}>
#295	<Card title="Tag and Organize Memories" icon="tag" href="/cookbooks/essentials/tagging-and-organizing-memories">
#296	Categorize patient data by symptoms, history, and visit context.
#297	</Card>
#298	<Card title="Support Inbox with Mem0" icon="headset" href="/cookbooks/operations/support-inbox">
#299	Apply similar memory patterns to customer support workflows.
#300	</Card>
#301	</CardGroup>
#302

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public