Update app.py
Browse files
app.py
CHANGED
@@ -1,6 +1,7 @@
|
|
1 |
import sys
|
2 |
import os
|
3 |
import pandas as pd
|
|
|
4 |
import gradio as gr
|
5 |
from typing import List, Tuple, Dict, Any
|
6 |
import hashlib
|
@@ -10,15 +11,20 @@ from datetime import datetime
|
|
10 |
import time
|
11 |
from collections import defaultdict
|
12 |
|
13 |
-
# Configuration
|
14 |
-
|
15 |
-
|
16 |
-
os.makedirs(REPORT_DIR, exist_ok=True)
|
17 |
|
18 |
-
|
19 |
-
|
20 |
-
os.
|
21 |
-
os.
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
current_dir = os.path.dirname(os.path.abspath(__file__))
|
24 |
src_path = os.path.abspath(os.path.join(current_dir, "src"))
|
@@ -30,20 +36,34 @@ from txagent.txagent import TxAgent
|
|
30 |
MAX_TOKENS = 32768
|
31 |
CHUNK_SIZE = 10000
|
32 |
MAX_NEW_TOKENS = 2048
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
def clean_response(text: str) -> str:
|
35 |
-
|
|
|
|
|
|
|
|
|
36 |
text = re.sub(r"\[.*?\]|\bNone\b", "", text, flags=re.DOTALL)
|
37 |
text = re.sub(r"\n{3,}", "\n\n", text)
|
|
|
38 |
return text.strip()
|
39 |
|
|
|
|
|
|
|
40 |
def process_patient_data(df: pd.DataFrame) -> Dict[str, Any]:
|
41 |
-
"""Process patient data into structured format"""
|
42 |
data = {
|
43 |
'bookings': defaultdict(list),
|
44 |
'medications': defaultdict(list),
|
45 |
'diagnoses': defaultdict(list),
|
46 |
'tests': defaultdict(list),
|
|
|
|
|
47 |
'timeline': []
|
48 |
}
|
49 |
|
@@ -62,62 +82,100 @@ def process_patient_data(df: pd.DataFrame) -> Dict[str, Any]:
|
|
62 |
|
63 |
data['bookings'][booking].append(entry)
|
64 |
data['timeline'].append(entry)
|
|
|
65 |
|
66 |
form_lower = entry['form'].lower()
|
67 |
-
if 'medication' in form_lower:
|
68 |
data['medications'][entry['item']].append(entry)
|
69 |
-
elif 'diagnosis' in form_lower:
|
70 |
data['diagnoses'][entry['item']].append(entry)
|
71 |
-
elif 'test' in form_lower:
|
72 |
data['tests'][entry['item']].append(entry)
|
|
|
|
|
73 |
|
74 |
return data
|
75 |
|
76 |
def generate_analysis_prompt(patient_data: Dict[str, Any], bookings: List[str]) -> str:
|
77 |
-
|
78 |
-
prompt = [
|
79 |
"**Comprehensive Patient Analysis**",
|
80 |
f"Analyzing {len(bookings)} bookings",
|
81 |
"",
|
82 |
-
"**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
]
|
84 |
|
85 |
for entry in patient_data['timeline']:
|
86 |
if entry['booking'] in bookings:
|
87 |
-
|
|
|
|
|
88 |
|
89 |
-
|
90 |
"",
|
91 |
-
"**
|
92 |
-
"
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
"",
|
97 |
-
"
|
|
|
|
|
|
|
|
|
|
|
98 |
])
|
99 |
|
100 |
-
return "\n".join(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
101 |
|
102 |
def init_agent():
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
|
|
107 |
|
108 |
-
|
109 |
model_name="mims-harvard/TxAgent-T1-Llama-3.1-8B",
|
110 |
rag_model_name="mims-harvard/ToolRAG-T1-GTE-Qwen2-1.5B",
|
111 |
-
tool_files_dict={"new_tool":
|
112 |
force_finish=True,
|
113 |
enable_checker=True,
|
114 |
step_rag_num=4,
|
115 |
seed=100,
|
116 |
additional_default_tools=[]
|
117 |
)
|
|
|
|
|
118 |
|
119 |
def analyze_with_agent(agent, prompt: str) -> str:
|
120 |
-
"""Run analysis with error handling"""
|
121 |
try:
|
122 |
response = ""
|
123 |
for result in agent.run_gradio_chat(
|
@@ -129,7 +187,11 @@ def analyze_with_agent(agent, prompt: str) -> str:
|
|
129 |
call_agent=False,
|
130 |
conversation=[],
|
131 |
):
|
132 |
-
if isinstance(result,
|
|
|
|
|
|
|
|
|
133 |
response += clean_response(result) + "\n"
|
134 |
elif hasattr(result, 'content'):
|
135 |
response += clean_response(result.content) + "\n"
|
@@ -139,64 +201,76 @@ def analyze_with_agent(agent, prompt: str) -> str:
|
|
139 |
return f"Error in analysis: {str(e)}"
|
140 |
|
141 |
def create_ui(agent):
|
142 |
-
with gr.Blocks(title="Patient History Analyzer") as demo:
|
143 |
-
gr.Markdown("# 🏥 Patient History
|
144 |
|
145 |
with gr.Tabs():
|
146 |
-
with gr.TabItem("
|
147 |
with gr.Row():
|
148 |
-
with gr.Column():
|
149 |
-
|
|
|
|
|
|
|
|
|
150 |
analyze_btn = gr.Button("Analyze", variant="primary")
|
|
|
151 |
|
152 |
-
with gr.Column():
|
153 |
output = gr.Markdown()
|
154 |
-
report = gr.File(label="Download Report"
|
155 |
|
156 |
with gr.TabItem("Instructions"):
|
157 |
gr.Markdown("""
|
158 |
-
|
159 |
1. Upload patient history Excel
|
160 |
2. Click Analyze
|
161 |
-
3. View
|
162 |
|
163 |
**Required Columns:**
|
164 |
- Booking Number
|
165 |
- Interview Date
|
166 |
- Interviewer
|
167 |
- Form Name
|
168 |
-
- Form Item
|
169 |
- Item Response
|
170 |
- Description
|
171 |
""")
|
172 |
|
173 |
def analyze(file):
|
174 |
if not file:
|
175 |
-
raise gr.Error("Please upload a file
|
176 |
|
177 |
try:
|
178 |
-
# Process file
|
179 |
df = pd.read_excel(file.name)
|
180 |
patient_data = process_patient_data(df)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
181 |
|
182 |
-
#
|
183 |
-
|
184 |
-
|
|
|
|
|
185 |
|
186 |
-
|
187 |
-
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
188 |
-
report_path = os.path.join(REPORT_DIR, f"report_{timestamp}.md")
|
189 |
with open(report_path, 'w') as f:
|
190 |
-
f.write(
|
191 |
|
192 |
-
|
193 |
|
194 |
except Exception as e:
|
195 |
-
raise gr.Error(f"
|
196 |
|
197 |
analyze_btn.click(
|
198 |
analyze,
|
199 |
-
inputs=
|
200 |
outputs=[output, report]
|
201 |
)
|
202 |
|
@@ -210,7 +284,7 @@ if __name__ == "__main__":
|
|
210 |
server_name="0.0.0.0",
|
211 |
server_port=7860,
|
212 |
show_error=True,
|
213 |
-
allowed_paths=[
|
214 |
)
|
215 |
except Exception as e:
|
216 |
print(f"Error: {str(e)}")
|
|
|
1 |
import sys
|
2 |
import os
|
3 |
import pandas as pd
|
4 |
+
import json
|
5 |
import gradio as gr
|
6 |
from typing import List, Tuple, Dict, Any
|
7 |
import hashlib
|
|
|
11 |
import time
|
12 |
from collections import defaultdict
|
13 |
|
14 |
+
# Configuration and setup
|
15 |
+
persistent_dir = "/data/hf_cache"
|
16 |
+
os.makedirs(persistent_dir, exist_ok=True)
|
|
|
17 |
|
18 |
+
model_cache_dir = os.path.join(persistent_dir, "txagent_models")
|
19 |
+
tool_cache_dir = os.path.join(persistent_dir, "tool_cache")
|
20 |
+
file_cache_dir = os.path.join(persistent_dir, "cache")
|
21 |
+
report_dir = os.path.join(persistent_dir, "reports")
|
22 |
+
|
23 |
+
for directory in [model_cache_dir, tool_cache_dir, file_cache_dir, report_dir]:
|
24 |
+
os.makedirs(directory, exist_ok=True)
|
25 |
+
|
26 |
+
os.environ["HF_HOME"] = model_cache_dir
|
27 |
+
os.environ["TRANSFORMERS_CACHE"] = model_cache_dir
|
28 |
|
29 |
current_dir = os.path.dirname(os.path.abspath(__file__))
|
30 |
src_path = os.path.abspath(os.path.join(current_dir, "src"))
|
|
|
36 |
MAX_TOKENS = 32768
|
37 |
CHUNK_SIZE = 10000
|
38 |
MAX_NEW_TOKENS = 2048
|
39 |
+
MAX_BOOKINGS_PER_CHUNK = 5
|
40 |
+
|
41 |
+
def file_hash(path: str) -> str:
|
42 |
+
with open(path, "rb") as f:
|
43 |
+
return hashlib.md5(f.read()).hexdigest()
|
44 |
|
45 |
def clean_response(text: str) -> str:
|
46 |
+
try:
|
47 |
+
text = text.encode('utf-8', 'surrogatepass').decode('utf-8')
|
48 |
+
except UnicodeError:
|
49 |
+
text = text.encode('utf-8', 'replace').decode('utf-8')
|
50 |
+
|
51 |
text = re.sub(r"\[.*?\]|\bNone\b", "", text, flags=re.DOTALL)
|
52 |
text = re.sub(r"\n{3,}", "\n\n", text)
|
53 |
+
text = re.sub(r"[^\n#\-\*\w\s\.,:\(\)]+", "", text)
|
54 |
return text.strip()
|
55 |
|
56 |
+
def estimate_tokens(text: str) -> int:
|
57 |
+
return len(text) // 3.5
|
58 |
+
|
59 |
def process_patient_data(df: pd.DataFrame) -> Dict[str, Any]:
|
|
|
60 |
data = {
|
61 |
'bookings': defaultdict(list),
|
62 |
'medications': defaultdict(list),
|
63 |
'diagnoses': defaultdict(list),
|
64 |
'tests': defaultdict(list),
|
65 |
+
'procedures': defaultdict(list),
|
66 |
+
'doctors': set(),
|
67 |
'timeline': []
|
68 |
}
|
69 |
|
|
|
82 |
|
83 |
data['bookings'][booking].append(entry)
|
84 |
data['timeline'].append(entry)
|
85 |
+
data['doctors'].add(entry['doctor'])
|
86 |
|
87 |
form_lower = entry['form'].lower()
|
88 |
+
if 'medication' in form_lower or 'drug' in form_lower:
|
89 |
data['medications'][entry['item']].append(entry)
|
90 |
+
elif 'diagnosis' in form_lower or 'condition' in form_lower:
|
91 |
data['diagnoses'][entry['item']].append(entry)
|
92 |
+
elif 'test' in form_lower or 'lab' in form_lower or 'result' in form_lower:
|
93 |
data['tests'][entry['item']].append(entry)
|
94 |
+
elif 'procedure' in form_lower or 'surgery' in form_lower:
|
95 |
+
data['procedures'][entry['item']].append(entry)
|
96 |
|
97 |
return data
|
98 |
|
99 |
def generate_analysis_prompt(patient_data: Dict[str, Any], bookings: List[str]) -> str:
|
100 |
+
prompt_lines = [
|
|
|
101 |
"**Comprehensive Patient Analysis**",
|
102 |
f"Analyzing {len(bookings)} bookings",
|
103 |
"",
|
104 |
+
"**Key Analysis Points:**",
|
105 |
+
"- Chronological progression of symptoms",
|
106 |
+
"- Medication changes and interactions",
|
107 |
+
"- Diagnostic consistency across providers",
|
108 |
+
"- Missed diagnostic opportunities",
|
109 |
+
"- Gaps in follow-up",
|
110 |
+
"",
|
111 |
+
"**Patient Timeline:**"
|
112 |
]
|
113 |
|
114 |
for entry in patient_data['timeline']:
|
115 |
if entry['booking'] in bookings:
|
116 |
+
prompt_lines.append(
|
117 |
+
f"- {entry['date']}: {entry['form']} - {entry['item']} = {entry['response']} (by {entry['doctor']})"
|
118 |
+
)
|
119 |
|
120 |
+
prompt_lines.extend([
|
121 |
"",
|
122 |
+
"**Medication History:**",
|
123 |
+
*[f"- {med}: " + " → ".join(
|
124 |
+
f"{e['date']}: {e['response']}"
|
125 |
+
for e in entries if e['booking'] in bookings
|
126 |
+
) for med, entries in patient_data['medications'].items()],
|
127 |
"",
|
128 |
+
"**Required Analysis Format:**",
|
129 |
+
"### Diagnostic Patterns",
|
130 |
+
"### Medication Analysis",
|
131 |
+
"### Provider Consistency",
|
132 |
+
"### Missed Opportunities",
|
133 |
+
"### Recommendations"
|
134 |
])
|
135 |
|
136 |
+
return "\n".join(prompt_lines)
|
137 |
+
|
138 |
+
def chunk_bookings(patient_data: Dict[str, Any]) -> List[List[str]]:
|
139 |
+
all_bookings = list(patient_data['bookings'].keys())
|
140 |
+
booking_sizes = []
|
141 |
+
|
142 |
+
for booking in all_bookings:
|
143 |
+
entries = patient_data['bookings'][booking]
|
144 |
+
size = sum(estimate_tokens(str(e)) for e in entries)
|
145 |
+
booking_sizes.append((booking, size))
|
146 |
+
|
147 |
+
booking_sizes.sort(key=lambda x: x[1], reverse=True)
|
148 |
+
chunks = [[] for _ in range(3)]
|
149 |
+
chunk_sizes = [0, 0, 0]
|
150 |
+
|
151 |
+
for booking, size in booking_sizes:
|
152 |
+
min_chunk = chunk_sizes.index(min(chunk_sizes))
|
153 |
+
chunks[min_chunk].append(booking)
|
154 |
+
chunk_sizes[min_chunk] += size
|
155 |
+
|
156 |
+
return chunks
|
157 |
|
158 |
def init_agent():
|
159 |
+
default_tool_path = os.path.abspath("data/new_tool.json")
|
160 |
+
target_tool_path = os.path.join(tool_cache_dir, "new_tool.json")
|
161 |
+
|
162 |
+
if not os.path.exists(target_tool_path):
|
163 |
+
shutil.copy(default_tool_path, target_tool_path)
|
164 |
|
165 |
+
agent = TxAgent(
|
166 |
model_name="mims-harvard/TxAgent-T1-Llama-3.1-8B",
|
167 |
rag_model_name="mims-harvard/ToolRAG-T1-GTE-Qwen2-1.5B",
|
168 |
+
tool_files_dict={"new_tool": target_tool_path},
|
169 |
force_finish=True,
|
170 |
enable_checker=True,
|
171 |
step_rag_num=4,
|
172 |
seed=100,
|
173 |
additional_default_tools=[]
|
174 |
)
|
175 |
+
agent.init_model()
|
176 |
+
return agent
|
177 |
|
178 |
def analyze_with_agent(agent, prompt: str) -> str:
|
|
|
179 |
try:
|
180 |
response = ""
|
181 |
for result in agent.run_gradio_chat(
|
|
|
187 |
call_agent=False,
|
188 |
conversation=[],
|
189 |
):
|
190 |
+
if isinstance(result, list):
|
191 |
+
for r in result:
|
192 |
+
if hasattr(r, 'content') and r.content:
|
193 |
+
response += clean_response(r.content) + "\n"
|
194 |
+
elif isinstance(result, str):
|
195 |
response += clean_response(result) + "\n"
|
196 |
elif hasattr(result, 'content'):
|
197 |
response += clean_response(result.content) + "\n"
|
|
|
201 |
return f"Error in analysis: {str(e)}"
|
202 |
|
203 |
def create_ui(agent):
|
204 |
+
with gr.Blocks(theme=gr.themes.Soft(), title="Patient History Analyzer") as demo:
|
205 |
+
gr.Markdown("# 🏥 Patient History Analyzer")
|
206 |
|
207 |
with gr.Tabs():
|
208 |
+
with gr.TabItem("Analysis"):
|
209 |
with gr.Row():
|
210 |
+
with gr.Column(scale=1):
|
211 |
+
file_upload = gr.File(
|
212 |
+
label="Upload Excel File",
|
213 |
+
file_types=[".xlsx"],
|
214 |
+
file_count="single"
|
215 |
+
)
|
216 |
analyze_btn = gr.Button("Analyze", variant="primary")
|
217 |
+
status = gr.Markdown("Ready")
|
218 |
|
219 |
+
with gr.Column(scale=2):
|
220 |
output = gr.Markdown()
|
221 |
+
report = gr.File(label="Download Report")
|
222 |
|
223 |
with gr.TabItem("Instructions"):
|
224 |
gr.Markdown("""
|
225 |
+
## How to Use
|
226 |
1. Upload patient history Excel
|
227 |
2. Click Analyze
|
228 |
+
3. View/download report
|
229 |
|
230 |
**Required Columns:**
|
231 |
- Booking Number
|
232 |
- Interview Date
|
233 |
- Interviewer
|
234 |
- Form Name
|
235 |
+
- Form Item
|
236 |
- Item Response
|
237 |
- Description
|
238 |
""")
|
239 |
|
240 |
def analyze(file):
|
241 |
if not file:
|
242 |
+
raise gr.Error("Please upload a file")
|
243 |
|
244 |
try:
|
|
|
245 |
df = pd.read_excel(file.name)
|
246 |
patient_data = process_patient_data(df)
|
247 |
+
chunks = chunk_bookings(patient_data)
|
248 |
+
full_report = []
|
249 |
+
|
250 |
+
for i, bookings in enumerate(chunks, 1):
|
251 |
+
prompt = generate_analysis_prompt(patient_data, bookings)
|
252 |
+
response = analyze_with_agent(agent, prompt)
|
253 |
+
full_report.append(f"## Chunk {i}\n{response}\n")
|
254 |
+
yield "\n".join(full_report), None
|
255 |
|
256 |
+
# Final summary
|
257 |
+
if len(chunks) > 1:
|
258 |
+
summary_prompt = "Create final summary combining all chunks"
|
259 |
+
summary = analyze_with_agent(agent, summary_prompt)
|
260 |
+
full_report.append(f"## Final Summary\n{summary}\n")
|
261 |
|
262 |
+
report_path = os.path.join(report_dir, f"report_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md")
|
|
|
|
|
263 |
with open(report_path, 'w') as f:
|
264 |
+
f.write("\n".join(full_report))
|
265 |
|
266 |
+
yield "\n".join(full_report), report_path
|
267 |
|
268 |
except Exception as e:
|
269 |
+
raise gr.Error(f"Error: {str(e)}")
|
270 |
|
271 |
analyze_btn.click(
|
272 |
analyze,
|
273 |
+
inputs=file_upload,
|
274 |
outputs=[output, report]
|
275 |
)
|
276 |
|
|
|
284 |
server_name="0.0.0.0",
|
285 |
server_port=7860,
|
286 |
show_error=True,
|
287 |
+
allowed_paths=["/data/hf_cache/reports"]
|
288 |
)
|
289 |
except Exception as e:
|
290 |
print(f"Error: {str(e)}")
|