davidfearne commited on
Commit
fef584f
·
verified ·
1 Parent(s): ee19843

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +278 -108
app.py CHANGED
@@ -5,13 +5,13 @@ import json
5
  import requests
6
  import uuid
7
  from datetime import date, datetime
8
- import requests
9
- from pydantic import BaseModel, Field
10
  from typing import Optional
11
 
 
12
  placeHolderPersona1 = """## Mission Statement
13
  My mission is to utilize my expertise to aid in the medical triaging process by providing a clear, concise, and accurate assessment of potential arthritis related conditions.
14
-  
15
  # Triaging process
16
  Ensure you stay on the topic of asking questions to triage the potential of Rheumatoid arthritis.
17
  Ask only one question at a time.
@@ -22,20 +22,19 @@ Do not give a diagnosis """
22
 
23
  placeHolderPersona2 = """## Mission
24
  To analyse a clinical triaging discussion between a patient and AI doctor interactions with a focus on Immunology symptoms, medical history, and test results to deduce the most probable Immunology diagnosis.
25
-  
26
  ## Diagnostic Process
27
  Upon receipt of the clinical notes, I will follow a systematic approach to arrive at a diagnosis:
28
  1. Review the patient's presenting symptoms and consider their relevance to immunopathology.
29
  2. Cross-reference the gathered information with my knowledge base of immunology to identify patterns or indicators of specific immune disorders.
30
  3. Formulate a diagnosis from the potential conditions.
31
  4. Determine the most likely diagnosis and assign a confidence score from 1-100, with 100 being absolute certainty.
32
-  
33
  # Limitations
34
  While I am specialized in immunology, I understand that not all cases will fall neatly within my domain. In instances where the clinical notes point to a condition outside of my expertise, I will provide the best possible diagnosis with the acknowledgment that my confidence score will reflect the limitations of my specialization in those cases"""
35
 
36
-
37
  class ChatRequestClient(BaseModel):
38
-
39
  user_id: str
40
  user_input: str
41
  numberOfQuestions: int
@@ -50,115 +49,286 @@ class ChatRequestClient(BaseModel):
50
  tokens2: int
51
  temperature2: float
52
 
 
53
  def call_chat_api(data: ChatRequestClient):
54
- url = "https://agent-builder-api.greensea-b20be511.northeurope.azurecontainerapps.io/chat/"
55
- # Validate and convert the data to a dictionary
56
- validated_data = data.dict()
57
-
58
- # Make the POST request to the FastAPI server
59
- response = requests.post(url, json=validated_data)
60
-
61
- if response.status_code == 200:
62
- return response.json() # Return the JSON response if successful
63
- else:
64
- return "An error occured" # Return the raw response text if not successful
65
 
66
- def genuuid ():
 
67
  return uuid.uuid4()
68
 
69
  def format_elapsed_time(time):
70
- # Format the elapsed time to two decimal places
71
  return "{:.2f}".format(time)
72
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
 
74
- # Title of the application
75
- # st.image('agentBuilderLogo.png')
76
- st.title('LLM-Powered Agent Interaction')
77
-
78
- # Sidebar for inputting personas
79
- st.sidebar.image('cognizant_logo.jpg')
80
- st.sidebar.header("Agent Personas Design")
81
- # st.sidebar.subheader("Welcome Message")
82
- # welcomeMessage = st.sidebar.text_area("Define Intake Persona", value=welcomeMessage, height=300)
83
- st.sidebar.subheader("Intake AI")
84
- numberOfQuestions = st.sidebar.slider("Number of Questions", min_value=0, max_value=10, step=1, value=5, key='persona1_questions')
85
- persona1SystemMessage = st.sidebar.text_area("Define Intake Persona", value=placeHolderPersona1, height=300)
86
- with st.sidebar.expander("See explanation"):
87
- st.write("This AI persona will converse with the patient to gather their symptoms. With each round of chat, the object of the AI is to ask more specific follow up questions as it narrows down to the specific diagnosis. However this AI should never give a diagnosis")
88
- st.image("agentPersona1.png")
89
- llm1 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona1_size')
90
- temp1 = st.sidebar.slider("Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.6, key='persona1_temp')
91
- tokens1 = st.sidebar.slider("Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona1_tokens')
92
-
93
- # Persona 2
94
- st.sidebar.subheader("Recommendation and Next Best Action AI")
95
- persona2SystemMessage = st.sidebar.text_area("Define Recommendation Persona", value=placeHolderPersona2, height=300)
96
- with st.sidebar.expander("See explanation"):
97
- st.write("This AI persona uses the output of the symptom intake AI as its input. This AI’s job is to augment a health professional by assisting with a diagnosis and possible next best action. The teams will need to determine if this should be a tool used directly by the patient, as an assistant to the health professional or a hybrid of the two. ")
98
- st.image("agentPersona2.png")
99
- llm2 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona2_size')
100
- temp2 = st.sidebar.slider("Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.5, key='persona2_temp')
101
- tokens2 = st.sidebar.slider("Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona2_tokens')
102
- userMessage2 = st.sidebar.text_area("Define User Message", value="This is the conversation todate, ", height=150)
103
- st.sidebar.caption(f"Session ID: {genuuid()}")
104
- # Main chat interface
105
- st.header("Chat with the Agents")
106
-
107
- # User ID Input
108
- user_id = st.text_input("User ID:", key="user_id")
109
-
110
- # Ensure user_id is defined or fallback to a default value
111
- if not user_id:
112
- st.warning("Please provide a User ID to start the chat.")
113
- else:
114
- # Initialize chat history in session state
115
- if "messages" not in st.session_state:
116
- st.session_state.messages = []
117
-
118
- # Display chat messages from history on app rerun
119
- for message in st.session_state.messages:
120
- with st.chat_message(message["role"]):
121
- st.markdown(message["content"])
122
-
123
- # Collect user input
124
- if user_input := st.chat_input("Write your message here:"):
125
- # Add user message to the chat history
126
- st.session_state.messages.append({"role": "user", "content": user_input})
127
- st.chat_message("user").markdown(user_input)
128
-
129
- # Prepare data for API call
130
- data = ChatRequestClient(
131
- user_id=user_id, # Ensure user_id is passed correctly
132
- user_input=user_input,
133
- numberOfQuestions=numberOfQuestions,
134
- welcomeMessage="",
135
- llm1=llm1,
136
- tokens1=tokens1,
137
- temperature1=temp1,
138
- persona1SystemMessage=persona1SystemMessage,
139
- persona2SystemMessage=persona2SystemMessage,
140
- userMessage2=userMessage2,
141
- llm2=llm2,
142
- tokens2=tokens2,
143
- temperature2=temp2
144
- )
145
-
146
- # Call the API
147
- response = call_chat_api(data)
148
-
149
- # Process the API response
150
- agent_message = response.get("content", "No response received from the agent.")
151
- elapsed_time = response.get("elapsed_time", 0)
152
- count = response.get("count", 0)
153
-
154
- # Add agent response to the chat history
155
- st.session_state.messages.append({"role": "assistant", "content": agent_message})
156
- with st.chat_message("assistant"):
157
- st.markdown(agent_message)
158
-
159
- # Display additional metadata
160
- st.markdown(f"##### Time taken: {format_elapsed_time(elapsed_time)} seconds")
161
- st.markdown(f"##### Question Count: {count} of {numberOfQuestions}")
162
 
163
 
164
 
 
5
  import requests
6
  import uuid
7
  from datetime import date, datetime
8
+ from pydantic import BaseModel
 
9
  from typing import Optional
10
 
11
+ # Placeholder personas
12
  placeHolderPersona1 = """## Mission Statement
13
  My mission is to utilize my expertise to aid in the medical triaging process by providing a clear, concise, and accurate assessment of potential arthritis related conditions.
14
+
15
  # Triaging process
16
  Ensure you stay on the topic of asking questions to triage the potential of Rheumatoid arthritis.
17
  Ask only one question at a time.
 
22
 
23
  placeHolderPersona2 = """## Mission
24
  To analyse a clinical triaging discussion between a patient and AI doctor interactions with a focus on Immunology symptoms, medical history, and test results to deduce the most probable Immunology diagnosis.
25
+
26
  ## Diagnostic Process
27
  Upon receipt of the clinical notes, I will follow a systematic approach to arrive at a diagnosis:
28
  1. Review the patient's presenting symptoms and consider their relevance to immunopathology.
29
  2. Cross-reference the gathered information with my knowledge base of immunology to identify patterns or indicators of specific immune disorders.
30
  3. Formulate a diagnosis from the potential conditions.
31
  4. Determine the most likely diagnosis and assign a confidence score from 1-100, with 100 being absolute certainty.
32
+
33
  # Limitations
34
  While I am specialized in immunology, I understand that not all cases will fall neatly within my domain. In instances where the clinical notes point to a condition outside of my expertise, I will provide the best possible diagnosis with the acknowledgment that my confidence score will reflect the limitations of my specialization in those cases"""
35
 
36
+ # Data model for API request
37
  class ChatRequestClient(BaseModel):
 
38
  user_id: str
39
  user_input: str
40
  numberOfQuestions: int
 
49
  tokens2: int
50
  temperature2: float
51
 
52
+ # Mock API call function
53
  def call_chat_api(data: ChatRequestClient):
54
+ # Replace this with actual API logic
55
+ return {
56
+ "content": f"Response to: {data.user_input}",
57
+ "elapsed_time": 0.5,
58
+ "count": 1,
59
+ "response_tokens": len(data.user_input.split()) # Mock token count
60
+ }
 
 
 
 
61
 
62
+ # Utility functions
63
+ def genuuid():
64
  return uuid.uuid4()
65
 
66
  def format_elapsed_time(time):
 
67
  return "{:.2f}".format(time)
68
 
69
+ # Layout with three columns
70
+ col1, col2, col3 = st.columns([1, 2, 1]) # Adjust ratios for desired width
71
+
72
+ # Left Column: Variables and Settings
73
+ with col1:
74
+ st.sidebar.image('cognizant_logo.jpg')
75
+ st.sidebar.header("Agent Personas Design")
76
+ st.sidebar.subheader("Intake AI")
77
+ numberOfQuestions = st.sidebar.slider(
78
+ "Number of Questions", min_value=0, max_value=10, step=1, value=5, key='persona1_questions'
79
+ )
80
+ persona1SystemMessage = st.sidebar.text_area(
81
+ "Define Intake Persona", value=placeHolderPersona1, height=300
82
+ )
83
+ llm1 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona1_size')
84
+ temp1 = st.sidebar.slider(
85
+ "Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.6, key='persona1_temp'
86
+ )
87
+ tokens1 = st.sidebar.slider(
88
+ "Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona1_tokens'
89
+ )
90
+ st.sidebar.subheader("Recommendation and Next Best Action AI")
91
+ persona2SystemMessage = st.sidebar.text_area(
92
+ "Define Recommendation Persona", value=placeHolderPersona2, height=300
93
+ )
94
+ llm2 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona2_size')
95
+ temp2 = st.sidebar.slider(
96
+ "Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.5, key='persona2_temp'
97
+ )
98
+ tokens2 = st.sidebar.slider(
99
+ "Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona2_tokens'
100
+ )
101
+ userMessage2 = st.sidebar.text_area(
102
+ "Define User Message", value="This is the conversation to date, ", height=150
103
+ )
104
+ st.sidebar.caption(f"Session ID: {genuuid()}")
105
+
106
+ # Middle Column: Chat Interface
107
+ with col2:
108
+ st.header("Chat with the Agents")
109
+ user_id = st.text_input("User ID:", key="user_id")
110
+
111
+ if not user_id:
112
+ st.warning("Please provide a User ID to start the chat.")
113
+ else:
114
+ # Initialize chat history
115
+ if "messages" not in st.session_state:
116
+ st.session_state.messages = []
117
+
118
+ # Display chat history
119
+ for message in st.session_state.messages:
120
+ with st.chat_message(message["role"]):
121
+ st.markdown(message["content"])
122
+
123
+ # Collect user input
124
+ if user_input := st.chat_input("Write your message here:"):
125
+ # Add user message
126
+ st.session_state.messages.append({"role": "user", "content": user_input})
127
+ st.chat_message("user").markdown(user_input)
128
+
129
+ # Prepare API data
130
+ data = ChatRequestClient(
131
+ user_id=user_id,
132
+ user_input=user_input,
133
+ numberOfQuestions=numberOfQuestions,
134
+ welcomeMessage="",
135
+ llm1=llm1,
136
+ tokens1=tokens1,
137
+ temperature1=temp1,
138
+ persona1SystemMessage=persona1SystemMessage,
139
+ persona2SystemMessage=persona2SystemMessage,
140
+ userMessage2=userMessage2,
141
+ llm2=llm2,
142
+ tokens2=tokens2,
143
+ temperature2=temp2
144
+ )
145
+
146
+ # Call the API
147
+ response = call_chat_api(data)
148
+
149
+ # Process response
150
+ agent_message = response.get("content", "No response received.")
151
+ elapsed_time = response.get("elapsed_time", 0)
152
+ count = response.get("count", 0)
153
+ response_tokens = response.get("response_tokens", 0)
154
+
155
+ # Add agent response
156
+ st.session_state.messages.append({"role": "assistant", "content": agent_message})
157
+ with st.chat_message("assistant"):
158
+ st.markdown(agent_message)
159
+
160
+ # Right Column: Stats
161
+ with col3:
162
+ st.header("Stats")
163
+ if "elapsed_time" in locals() and "count" in locals():
164
+ st.markdown(f"**Time taken:** {format_elapsed_time(elapsed_time)} seconds")
165
+ st.markdown(f"**Question Count:** {count} of {numberOfQuestions}")
166
+ st.markdown(f"**Response Tokens:** {response_tokens}")
167
+ else:
168
+ st.markdown("No stats available yet.")
169
+
170
+
171
+ # import os
172
+ # import streamlit as st
173
+ # from datetime import datetime
174
+ # import json
175
+ # import requests
176
+ # import uuid
177
+ # from datetime import date, datetime
178
+ # import requests
179
+ # from pydantic import BaseModel, Field
180
+ # from typing import Optional
181
+
182
+ # placeHolderPersona1 = """## Mission Statement
183
+ # My mission is to utilize my expertise to aid in the medical triaging process by providing a clear, concise, and accurate assessment of potential arthritis related conditions.
184
+  
185
+ # # Triaging process
186
+ # Ensure you stay on the topic of asking questions to triage the potential of Rheumatoid arthritis.
187
+ # Ask only one question at a time.
188
+ # Provide some context or clarification around the follow-up questions you ask.
189
+ # Do not converse with the customer.
190
+ # Be as concise as possible.
191
+ # Do not give a diagnosis """
192
+
193
+ # placeHolderPersona2 = """## Mission
194
+ # To analyse a clinical triaging discussion between a patient and AI doctor interactions with a focus on Immunology symptoms, medical history, and test results to deduce the most probable Immunology diagnosis.
195
+  
196
+ # ## Diagnostic Process
197
+ # Upon receipt of the clinical notes, I will follow a systematic approach to arrive at a diagnosis:
198
+ # 1. Review the patient's presenting symptoms and consider their relevance to immunopathology.
199
+ # 2. Cross-reference the gathered information with my knowledge base of immunology to identify patterns or indicators of specific immune disorders.
200
+ # 3. Formulate a diagnosis from the potential conditions.
201
+ # 4. Determine the most likely diagnosis and assign a confidence score from 1-100, with 100 being absolute certainty.
202
+  
203
+ # # Limitations
204
+ # While I am specialized in immunology, I understand that not all cases will fall neatly within my domain. In instances where the clinical notes point to a condition outside of my expertise, I will provide the best possible diagnosis with the acknowledgment that my confidence score will reflect the limitations of my specialization in those cases"""
205
+
206
+
207
+ # class ChatRequestClient(BaseModel):
208
+
209
+ # user_id: str
210
+ # user_input: str
211
+ # numberOfQuestions: int
212
+ # welcomeMessage: str
213
+ # llm1: str
214
+ # tokens1: int
215
+ # temperature1: float
216
+ # persona1SystemMessage: str
217
+ # persona2SystemMessage: str
218
+ # userMessage2: str
219
+ # llm2: str
220
+ # tokens2: int
221
+ # temperature2: float
222
+
223
+ # def call_chat_api(data: ChatRequestClient):
224
+ # url = "https://agent-builder-api.greensea-b20be511.northeurope.azurecontainerapps.io/chat/"
225
+ # # Validate and convert the data to a dictionary
226
+ # validated_data = data.dict()
227
+
228
+ # # Make the POST request to the FastAPI server
229
+ # response = requests.post(url, json=validated_data)
230
+
231
+ # if response.status_code == 200:
232
+ # return response.json() # Return the JSON response if successful
233
+ # else:
234
+ # return "An error occured" # Return the raw response text if not successful
235
+
236
+ # def genuuid ():
237
+ # return uuid.uuid4()
238
+
239
+ # def format_elapsed_time(time):
240
+ # # Format the elapsed time to two decimal places
241
+ # return "{:.2f}".format(time)
242
+
243
+
244
+ # # Title of the application
245
+ # # st.image('agentBuilderLogo.png')
246
+ # st.title('LLM-Powered Agent Interaction')
247
+
248
+ # # Sidebar for inputting personas
249
+ # st.sidebar.image('cognizant_logo.jpg')
250
+ # st.sidebar.header("Agent Personas Design")
251
+ # # st.sidebar.subheader("Welcome Message")
252
+ # # welcomeMessage = st.sidebar.text_area("Define Intake Persona", value=welcomeMessage, height=300)
253
+ # st.sidebar.subheader("Intake AI")
254
+ # numberOfQuestions = st.sidebar.slider("Number of Questions", min_value=0, max_value=10, step=1, value=5, key='persona1_questions')
255
+ # persona1SystemMessage = st.sidebar.text_area("Define Intake Persona", value=placeHolderPersona1, height=300)
256
+ # with st.sidebar.expander("See explanation"):
257
+ # st.write("This AI persona will converse with the patient to gather their symptoms. With each round of chat, the object of the AI is to ask more specific follow up questions as it narrows down to the specific diagnosis. However this AI should never give a diagnosis")
258
+ # st.image("agentPersona1.png")
259
+ # llm1 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona1_size')
260
+ # temp1 = st.sidebar.slider("Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.6, key='persona1_temp')
261
+ # tokens1 = st.sidebar.slider("Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona1_tokens')
262
+
263
+ # # Persona 2
264
+ # st.sidebar.subheader("Recommendation and Next Best Action AI")
265
+ # persona2SystemMessage = st.sidebar.text_area("Define Recommendation Persona", value=placeHolderPersona2, height=300)
266
+ # with st.sidebar.expander("See explanation"):
267
+ # st.write("This AI persona uses the output of the symptom intake AI as its input. This AI’s job is to augment a health professional by assisting with a diagnosis and possible next best action. The teams will need to determine if this should be a tool used directly by the patient, as an assistant to the health professional or a hybrid of the two. ")
268
+ # st.image("agentPersona2.png")
269
+ # llm2 = st.sidebar.selectbox("Model Selection", ['GPT-4', 'GPT3.5'], key='persona2_size')
270
+ # temp2 = st.sidebar.slider("Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.5, key='persona2_temp')
271
+ # tokens2 = st.sidebar.slider("Tokens", min_value=0, max_value=4000, step=100, value=500, key='persona2_tokens')
272
+ # userMessage2 = st.sidebar.text_area("Define User Message", value="This is the conversation todate, ", height=150)
273
+ # st.sidebar.caption(f"Session ID: {genuuid()}")
274
+ # # Main chat interface
275
+ # st.header("Chat with the Agents")
276
+
277
+ # # User ID Input
278
+ # user_id = st.text_input("User ID:", key="user_id")
279
+
280
+ # # Ensure user_id is defined or fallback to a default value
281
+ # if not user_id:
282
+ # st.warning("Please provide a User ID to start the chat.")
283
+ # else:
284
+ # # Initialize chat history in session state
285
+ # if "messages" not in st.session_state:
286
+ # st.session_state.messages = []
287
+
288
+ # # Display chat messages from history on app rerun
289
+ # for message in st.session_state.messages:
290
+ # with st.chat_message(message["role"]):
291
+ # st.markdown(message["content"])
292
+
293
+ # # Collect user input
294
+ # if user_input := st.chat_input("Write your message here:"):
295
+ # # Add user message to the chat history
296
+ # st.session_state.messages.append({"role": "user", "content": user_input})
297
+ # st.chat_message("user").markdown(user_input)
298
+
299
+ # # Prepare data for API call
300
+ # data = ChatRequestClient(
301
+ # user_id=user_id, # Ensure user_id is passed correctly
302
+ # user_input=user_input,
303
+ # numberOfQuestions=numberOfQuestions,
304
+ # welcomeMessage="",
305
+ # llm1=llm1,
306
+ # tokens1=tokens1,
307
+ # temperature1=temp1,
308
+ # persona1SystemMessage=persona1SystemMessage,
309
+ # persona2SystemMessage=persona2SystemMessage,
310
+ # userMessage2=userMessage2,
311
+ # llm2=llm2,
312
+ # tokens2=tokens2,
313
+ # temperature2=temp2
314
+ # )
315
+
316
+ # # Call the API
317
+ # response = call_chat_api(data)
318
+
319
+ # # Process the API response
320
+ # agent_message = response.get("content", "No response received from the agent.")
321
+ # elapsed_time = response.get("elapsed_time", 0)
322
+ # count = response.get("count", 0)
323
+
324
+ # # Add agent response to the chat history
325
+ # st.session_state.messages.append({"role": "assistant", "content": agent_message})
326
+ # with st.chat_message("assistant"):
327
+ # st.markdown(agent_message)
328
 
329
+ # # Display additional metadata
330
+ # st.markdown(f"##### Time taken: {format_elapsed_time(elapsed_time)} seconds")
331
+ # st.markdown(f"##### Question Count: {count} of {numberOfQuestions}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
332
 
333
 
334