Added Anthropic Architecture

Patient_Chat/settings.py

@@ -1,19 +1,12 @@
-"""
-Django settings for Patient_Chat project.
-
-Generated by 'django-admin startproject' using Django 5.1.1.
-
-For more information on this file, see
-https://docs.djangoproject.com/en/5.1/topics/settings/
-
-For the full list of settings and their values, see
-https://docs.djangoproject.com/en/5.1/ref/settings/
-"""
-
 from pathlib import Path
+import os
+from dotenv import load_dotenv
 
 # Build paths inside the project like this: BASE_DIR / 'subdir'.
 BASE_DIR = Path(__file__).resolve().parent.parent
+load_dotenv( os.path.join( BASE_DIR, '.env'))
+
+ANTHROPIC_API_KEY = os.environ.get('ANTHROPIC_API_KEY')
 
 # Quick-start development settings - unsuitable for production
 # See https://docs.djangoproject.com/en/5.1/howto/deployment/checklist/
@@ -117,3 +110,4 @@
 # https://docs.djangoproject.com/en/5.1/ref/settings/#default-auto-field
 
 DEFAULT_AUTO_FIELD = 'django.db.models.BigAutoField'
+

home/llm/llm.py

@@ -0,0 +1,45 @@
+from langchain_anthropic import ChatAnthropic
+
+
+class LLM:
+    def __init__(self, model_name: str = "claude-3-haiku-20240307"):
+        self.model_name = model_name
+        llm = ChatAnthropic(model=model_name)
+
+        self.tools = [request_medication_change, make_appointment, request_appointment_change]
+        self.llm_with_tools = llm.bind_tools(self.tools)
+
+
+def request_medication_change(previous_medication: str) -> str:
+    """Puts a request to the doctor for medication change.
+
+    Returns a string with the name of current medication and Change request submitted.
+
+    Args:
+        previous_medication: first str
+    """
+    return "Change request submitted for " + previous_medication
+
+
+def make_appointment(date: str, reason: str) -> str:
+    """Puts an appointment request on the specified date and reason.
+
+    Returns a string with the reason for the requested appointment and date for the appointment
+
+    Args:
+        date: first str
+        reason: second str
+    """
+    return "Appointment requested on " + date + " for " + reason
+
+
+def request_appointment_change(past_date: str, requested_date: str) -> str:
+    """Puts an appointment change request for a given date and requested.
+
+    Returns a string with the changed and previous date for appointment.
+
+    Args:
+        past_date: first str
+        requested_date: second str
+    """
+    return past_date + " " + requested_date

home/llm/llm_graph.py

@@ -0,0 +1,56 @@
+from langgraph.graph import START, StateGraph
+from langgraph.graph.state import CompiledStateGraph
+from langgraph.prebuilt import tools_condition, ToolNode
+
+from langgraph.graph import MessagesState
+from langchain_core.messages import AIMessage, HumanMessage, SystemMessage
+
+from home.llm.llm import LLM
+
+
+class LLMGraph:
+    prompt_text = '''
+        You are a helpful AI medical assistant. You should only respond to health-related topics such as:
+        - General human health and lifestyle inquiries.
+        - Questions about the patient’s medical condition, medication regimen, diet, etc.
+        - Various requests from the patient to their doctor such as make appointments, modify appointments and medication changes.
+        You should filter out and ignore any unrelated, sensitive, or controversial topics.
+    '''
+
+    def __init__(self):
+        self.llm = LLM()
+        self.graph = self.build_graph()
+
+
+    def assistant(self, state: MessagesState):
+        # Prompt message
+        sys_msg = SystemMessage(content=self.prompt_text)
+        return {"messages": [self.llm.llm_with_tools.invoke([sys_msg] + state["messages"])]}
+
+    def build_graph(self) -> CompiledStateGraph:
+        builder = StateGraph(MessagesState)
+
+        # Define nodes: these do the work
+        builder.add_node("assistant", self.assistant)
+        builder.add_node("tools", ToolNode(self.llm.tools))
+
+        # Define edges: these determine how the control flow moves
+        builder.add_edge(START, "assistant")
+        builder.add_conditional_edges(
+            "assistant",
+            # If the latest message (result) from assistant is a tool call -> tools_condition routes to tools
+            # If the latest message (result) from assistant is a not a tool call -> tools_condition routes to END
+            tools_condition,
+        )
+        builder.add_edge("tools", "assistant")
+
+        return builder.compile()
+
+    def inference(self, user_message) -> str:
+        messages = [HumanMessage(content=user_message)]
+        messages = self.graph.invoke({"messages": messages})
+        for m in messages['messages']:
+            m.pretty_print()
+
+        print()
+        return messages['messages'][-1].content