updates
Browse files- crew.py +22 -81
- dataloaderImp.py → firebaseLoader.py +0 -0
- requirements.txt +2 -1
- webscraper.py +99 -0
crew.py
CHANGED
|
@@ -5,7 +5,8 @@ import joblib
|
|
| 5 |
import numpy as np
|
| 6 |
from crewai import Agent, Task, Crew, LLM
|
| 7 |
from crewai.tools import tool
|
| 8 |
-
from
|
|
|
|
| 9 |
from typing import Union
|
| 10 |
import shap
|
| 11 |
import xgboost as xgb
|
|
@@ -36,10 +37,10 @@ placeholder_lifestyle_ID = "umWSG13rLSfHqqJlq1WluPzjLB93"
|
|
| 36 |
|
| 37 |
# Define tools for the agents
|
| 38 |
@tool("calculate_dementia_risk_audio")
|
| 39 |
-
def calculate_dementia_risk_audio(docID: str
|
| 40 |
"""
|
| 41 |
Calculates dementia risk for a specific patient using audio and transcript data from Firebase.
|
| 42 |
-
|
| 43 |
"""
|
| 44 |
try:
|
| 45 |
# Handle JSON-wrapped input (from CrewAI)
|
|
@@ -58,9 +59,7 @@ def calculate_dementia_risk_audio(docID: str, return_json: bool = False) -> Unio
|
|
| 58 |
|
| 59 |
if audio_data is None:
|
| 60 |
error_msg = f"Error: Could not retrieve or process audio data for patient '{docID}'"
|
| 61 |
-
|
| 62 |
-
return {"status": "error", "message": error_msg}
|
| 63 |
-
return error_msg
|
| 64 |
|
| 65 |
# Extract values
|
| 66 |
age = audio_data.get("age")
|
|
@@ -70,9 +69,7 @@ def calculate_dementia_risk_audio(docID: str, return_json: bool = False) -> Unio
|
|
| 70 |
|
| 71 |
if combined_features is None:
|
| 72 |
error_msg = f"Error: Feature extraction failed for patient '{docID}'"
|
| 73 |
-
|
| 74 |
-
return {"status": "error", "message": error_msg}
|
| 75 |
-
return error_msg
|
| 76 |
|
| 77 |
# Convert to DMatrix
|
| 78 |
audio_array = np.array(combined_features).reshape(1, -1)
|
|
@@ -109,38 +106,11 @@ def calculate_dementia_risk_audio(docID: str, return_json: bool = False) -> Unio
|
|
| 109 |
"docID": docID
|
| 110 |
}
|
| 111 |
|
| 112 |
-
|
| 113 |
-
return {"status": "success", **result_dict}
|
| 114 |
-
|
| 115 |
-
# Fallback: formatted string report (for CrewAI agents)
|
| 116 |
-
explanation = (
|
| 117 |
-
f"Based on speech analysis and linguistic patterns, "
|
| 118 |
-
f"the AI model detected {'concerning patterns' if prediction == 1 else 'normal patterns'} "
|
| 119 |
-
f"in speech fluency, vocabulary diversity, and cognitive response time."
|
| 120 |
-
)
|
| 121 |
-
|
| 122 |
-
mmse_note = "\n[Note: MMSE score not provided. Assessment based on speech and demographics only. Accuracy may be reduced.]"
|
| 123 |
-
|
| 124 |
-
report = f"""
|
| 125 |
-
DEMENTIA RISK ASSESSMENT (AUDIO)
|
| 126 |
-
================================
|
| 127 |
-
Risk Level: {risk_level}
|
| 128 |
-
Confidence: {confidence:.2%}
|
| 129 |
-
Prediction: {'Positive' if prediction == 1 else 'Negative'}
|
| 130 |
-
Age: {age}
|
| 131 |
-
Gender: {gender_str}
|
| 132 |
-
{mmse_note}
|
| 133 |
-
|
| 134 |
-
EXPLANATION:
|
| 135 |
-
{explanation}
|
| 136 |
-
"""
|
| 137 |
-
return report
|
| 138 |
|
| 139 |
except Exception as e:
|
| 140 |
error_msg = f"Error calculating audio-based risk: {str(e)}"
|
| 141 |
-
|
| 142 |
-
return {"status": "error", "message": error_msg}
|
| 143 |
-
return error_msg
|
| 144 |
|
| 145 |
@tool("explain_dementia_risk_audio")
|
| 146 |
def explain_dementia_risk_audio(docID: str) -> str:
|
|
@@ -178,10 +148,10 @@ def explain_dementia_risk_audio(docID: str) -> str:
|
|
| 178 |
return f"SHAP explanation failed: {str(e)}"
|
| 179 |
|
| 180 |
@tool("calculate_dementia_risk_lifestyle")
|
| 181 |
-
def calculate_dementia_risk_lifestyle(docID: str
|
| 182 |
"""
|
| 183 |
Calculates dementia risk for a specific patient using lifestyle data.
|
| 184 |
-
|
| 185 |
"""
|
| 186 |
try:
|
| 187 |
# Handle JSON-wrapped input (from CrewAI)
|
|
@@ -199,9 +169,7 @@ def calculate_dementia_risk_lifestyle(docID: str, return_json: bool = False) ->
|
|
| 199 |
lifestyle_data = pull_lifestyle(docID)
|
| 200 |
if not lifestyle_data:
|
| 201 |
error = f"Error: No lifestyle data found for document ID '{docID}'"
|
| 202 |
-
|
| 203 |
-
return {"status": "error", "message": error}
|
| 204 |
-
return error
|
| 205 |
|
| 206 |
FEATURE_MAPPING = {
|
| 207 |
'Age': 'age',
|
|
@@ -227,15 +195,11 @@ def calculate_dementia_risk_lifestyle(docID: str, return_json: bool = False) ->
|
|
| 227 |
firestore_key = FEATURE_MAPPING.get(feat_name)
|
| 228 |
if firestore_key is None:
|
| 229 |
error = f"Error: No mapping defined for feature '{feat_name}'"
|
| 230 |
-
|
| 231 |
-
return {"status": "error", "message": error}
|
| 232 |
-
return error
|
| 233 |
val = lifestyle_data.get(firestore_key)
|
| 234 |
if val is None:
|
| 235 |
error = f"Error: Missing required lifestyle feature '{feat_name}' (Firestore key: '{firestore_key}')"
|
| 236 |
-
|
| 237 |
-
return {"status": "error", "message": error}
|
| 238 |
-
return error
|
| 239 |
features.append(val)
|
| 240 |
|
| 241 |
# Convert to DMatrix
|
|
@@ -272,35 +236,11 @@ def calculate_dementia_risk_lifestyle(docID: str, return_json: bool = False) ->
|
|
| 272 |
"docID": docID
|
| 273 |
}
|
| 274 |
|
| 275 |
-
|
| 276 |
-
return {"status": "success", **result_dict}
|
| 277 |
-
|
| 278 |
-
# Original string report
|
| 279 |
-
explanation = (
|
| 280 |
-
f"Based on lifestyle factors including diet, physical activity, and medical history, "
|
| 281 |
-
f"the AI model detected {'concerning patterns' if prediction == 1 else 'normal patterns'} "
|
| 282 |
-
f"associated with increased dementia risk."
|
| 283 |
-
)
|
| 284 |
-
|
| 285 |
-
report = f"""
|
| 286 |
-
DEMENTIA RISK ASSESSMENT (LIFESTYLE)
|
| 287 |
-
===================================
|
| 288 |
-
Risk Level: {risk_level}
|
| 289 |
-
Confidence: {confidence:.2%}
|
| 290 |
-
Prediction: {'Positive' if prediction == 1 else 'Negative'}
|
| 291 |
-
Age: {age}
|
| 292 |
-
Gender: {gender_str}
|
| 293 |
-
|
| 294 |
-
EXPLANATION:
|
| 295 |
-
{explanation}
|
| 296 |
-
"""
|
| 297 |
-
return report
|
| 298 |
|
| 299 |
except Exception as e:
|
| 300 |
error_msg = f"Error calculating risk: {str(e)}"
|
| 301 |
-
|
| 302 |
-
return {"status": "error", "message": error_msg}
|
| 303 |
-
return error_msg
|
| 304 |
|
| 305 |
@tool("explain_dementia_risk_lifestyle")
|
| 306 |
def explain_dementia_risk_lifestyle(docID: str) -> str:
|
|
@@ -359,17 +299,18 @@ def explain_dementia_risk_lifestyle(docID: str) -> str:
|
|
| 359 |
return f"SHAP explanation failed: {str(e)}"
|
| 360 |
|
| 361 |
@tool("gather_news_patient")
|
| 362 |
-
def gather_news_patient(
|
| 363 |
"""
|
| 364 |
Gathers dementia news articles, news articles that are related to and important to patients.
|
| 365 |
-
|
| 366 |
"""
|
| 367 |
return "patient news"
|
|
|
|
| 368 |
@tool("gather_news_doctor")
|
| 369 |
-
def gather_news_doctor(
|
| 370 |
"""
|
| 371 |
Gathers dementia news articles, news articles that are related to and important to doctors.
|
| 372 |
-
|
| 373 |
"""
|
| 374 |
return "doctor news"
|
| 375 |
|
|
@@ -479,7 +420,7 @@ def run_crewai_lifestyle(docID: str) -> dict:
|
|
| 479 |
if not lifestyle_data:
|
| 480 |
return {"status": "error", "message": f"No lifestyle data for {docID}"}
|
| 481 |
|
| 482 |
-
risk_result = calculate_dementia_risk_lifestyle.run(docID
|
| 483 |
if risk_result["status"] != "success":
|
| 484 |
return risk_result
|
| 485 |
shap_explanation = explain_dementia_risk_lifestyle.run(docID)
|
|
@@ -514,7 +455,7 @@ def run_crewai_audio(docID: str) -> dict:
|
|
| 514 |
raw_output = crew_audio.kickoff()
|
| 515 |
|
| 516 |
# Get structured result
|
| 517 |
-
risk_result = calculate_dementia_risk_audio.run(docID
|
| 518 |
if risk_result["status"] != "success":
|
| 519 |
return risk_result
|
| 520 |
shap_explanation = explain_dementia_risk_audio.run(docID)
|
|
|
|
| 5 |
import numpy as np
|
| 6 |
from crewai import Agent, Task, Crew, LLM
|
| 7 |
from crewai.tools import tool
|
| 8 |
+
from firebaseLoader import *
|
| 9 |
+
from webscraper import search_dementia_news
|
| 10 |
from typing import Union
|
| 11 |
import shap
|
| 12 |
import xgboost as xgb
|
|
|
|
| 37 |
|
| 38 |
# Define tools for the agents
|
| 39 |
@tool("calculate_dementia_risk_audio")
|
| 40 |
+
def calculate_dementia_risk_audio(docID: str) -> Union[str, dict]:
|
| 41 |
"""
|
| 42 |
Calculates dementia risk for a specific patient using audio and transcript data from Firebase.
|
| 43 |
+
Returns a structured dict.
|
| 44 |
"""
|
| 45 |
try:
|
| 46 |
# Handle JSON-wrapped input (from CrewAI)
|
|
|
|
| 59 |
|
| 60 |
if audio_data is None:
|
| 61 |
error_msg = f"Error: Could not retrieve or process audio data for patient '{docID}'"
|
| 62 |
+
return {"status": "error", "message": error_msg}
|
|
|
|
|
|
|
| 63 |
|
| 64 |
# Extract values
|
| 65 |
age = audio_data.get("age")
|
|
|
|
| 69 |
|
| 70 |
if combined_features is None:
|
| 71 |
error_msg = f"Error: Feature extraction failed for patient '{docID}'"
|
| 72 |
+
return {"status": "error", "message": error_msg}
|
|
|
|
|
|
|
| 73 |
|
| 74 |
# Convert to DMatrix
|
| 75 |
audio_array = np.array(combined_features).reshape(1, -1)
|
|
|
|
| 106 |
"docID": docID
|
| 107 |
}
|
| 108 |
|
| 109 |
+
return {"status": "success", **result_dict}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
|
| 111 |
except Exception as e:
|
| 112 |
error_msg = f"Error calculating audio-based risk: {str(e)}"
|
| 113 |
+
return {"status": "error", "message": error_msg}
|
|
|
|
|
|
|
| 114 |
|
| 115 |
@tool("explain_dementia_risk_audio")
|
| 116 |
def explain_dementia_risk_audio(docID: str) -> str:
|
|
|
|
| 148 |
return f"SHAP explanation failed: {str(e)}"
|
| 149 |
|
| 150 |
@tool("calculate_dementia_risk_lifestyle")
|
| 151 |
+
def calculate_dementia_risk_lifestyle(docID: str) -> Union[str, dict]:
|
| 152 |
"""
|
| 153 |
Calculates dementia risk for a specific patient using lifestyle data.
|
| 154 |
+
Returns a structured dict
|
| 155 |
"""
|
| 156 |
try:
|
| 157 |
# Handle JSON-wrapped input (from CrewAI)
|
|
|
|
| 169 |
lifestyle_data = pull_lifestyle(docID)
|
| 170 |
if not lifestyle_data:
|
| 171 |
error = f"Error: No lifestyle data found for document ID '{docID}'"
|
| 172 |
+
return {"status": "error", "message": error}
|
|
|
|
|
|
|
| 173 |
|
| 174 |
FEATURE_MAPPING = {
|
| 175 |
'Age': 'age',
|
|
|
|
| 195 |
firestore_key = FEATURE_MAPPING.get(feat_name)
|
| 196 |
if firestore_key is None:
|
| 197 |
error = f"Error: No mapping defined for feature '{feat_name}'"
|
| 198 |
+
return {"status": "error", "message": error}
|
|
|
|
|
|
|
| 199 |
val = lifestyle_data.get(firestore_key)
|
| 200 |
if val is None:
|
| 201 |
error = f"Error: Missing required lifestyle feature '{feat_name}' (Firestore key: '{firestore_key}')"
|
| 202 |
+
return {"status": "error", "message": error}
|
|
|
|
|
|
|
| 203 |
features.append(val)
|
| 204 |
|
| 205 |
# Convert to DMatrix
|
|
|
|
| 236 |
"docID": docID
|
| 237 |
}
|
| 238 |
|
| 239 |
+
return {"status": "success", **result_dict}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 240 |
|
| 241 |
except Exception as e:
|
| 242 |
error_msg = f"Error calculating risk: {str(e)}"
|
| 243 |
+
return {"status": "error", "message": error_msg}
|
|
|
|
|
|
|
| 244 |
|
| 245 |
@tool("explain_dementia_risk_lifestyle")
|
| 246 |
def explain_dementia_risk_lifestyle(docID: str) -> str:
|
|
|
|
| 299 |
return f"SHAP explanation failed: {str(e)}"
|
| 300 |
|
| 301 |
@tool("gather_news_patient")
|
| 302 |
+
def gather_news_patient():
|
| 303 |
"""
|
| 304 |
Gathers dementia news articles, news articles that are related to and important to patients.
|
| 305 |
+
Returns a structured dict.
|
| 306 |
"""
|
| 307 |
return "patient news"
|
| 308 |
+
|
| 309 |
@tool("gather_news_doctor")
|
| 310 |
+
def gather_news_doctor(max_count = 4):
|
| 311 |
"""
|
| 312 |
Gathers dementia news articles, news articles that are related to and important to doctors.
|
| 313 |
+
Returns a structured dict.
|
| 314 |
"""
|
| 315 |
return "doctor news"
|
| 316 |
|
|
|
|
| 420 |
if not lifestyle_data:
|
| 421 |
return {"status": "error", "message": f"No lifestyle data for {docID}"}
|
| 422 |
|
| 423 |
+
risk_result = calculate_dementia_risk_lifestyle.run(docID)
|
| 424 |
if risk_result["status"] != "success":
|
| 425 |
return risk_result
|
| 426 |
shap_explanation = explain_dementia_risk_lifestyle.run(docID)
|
|
|
|
| 455 |
raw_output = crew_audio.kickoff()
|
| 456 |
|
| 457 |
# Get structured result
|
| 458 |
+
risk_result = calculate_dementia_risk_audio.run(docID)
|
| 459 |
if risk_result["status"] != "success":
|
| 460 |
return risk_result
|
| 461 |
shap_explanation = explain_dementia_risk_audio.run(docID)
|
dataloaderImp.py → firebaseLoader.py
RENAMED
|
File without changes
|
requirements.txt
CHANGED
|
@@ -13,4 +13,5 @@ openai-whisper
|
|
| 13 |
en-core-web-sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1-py3-none-any.whl
|
| 14 |
shap
|
| 15 |
xgboost==3.0.5
|
| 16 |
-
litellm
|
|
|
|
|
|
| 13 |
en-core-web-sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1-py3-none-any.whl
|
| 14 |
shap
|
| 15 |
xgboost==3.0.5
|
| 16 |
+
litellm
|
| 17 |
+
litellm[proxy]
|
webscraper.py
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import requests
|
| 2 |
+
from bs4 import BeautifulSoup
|
| 3 |
+
from duckduckgo_search import DDGS
|
| 4 |
+
import time
|
| 5 |
+
import re
|
| 6 |
+
from typing import List, Dict
|
| 7 |
+
|
| 8 |
+
def clean_text(text: str) -> str:
|
| 9 |
+
"""Clean and normalize text: remove extra whitespace, fix line breaks."""
|
| 10 |
+
if not text:
|
| 11 |
+
return ""
|
| 12 |
+
text = re.sub(r'\s+', ' ', text).strip()
|
| 13 |
+
return text[:1000]
|
| 14 |
+
|
| 15 |
+
def scrape_article_content(url: str, timeout: int = 8) -> str:
|
| 16 |
+
"""Scrape main article text using requests + BeautifulSoup."""
|
| 17 |
+
try:
|
| 18 |
+
headers = {
|
| 19 |
+
"User-Agent": "Mozilla/5.0 (compatible; DementiaNewsBot/1.0; +https://huggingface.co/spaces/plutoa386/...)"
|
| 20 |
+
}
|
| 21 |
+
response = requests.get(url, headers=headers, timeout=timeout)
|
| 22 |
+
response.raise_for_status()
|
| 23 |
+
|
| 24 |
+
soup = BeautifulSoup(response.content, "lxml")
|
| 25 |
+
|
| 26 |
+
# Remove noisy elements
|
| 27 |
+
for elem in soup(["script", "style", "nav", "footer", "header", "aside", "form"]):
|
| 28 |
+
elem.decompose()
|
| 29 |
+
|
| 30 |
+
# Try to find main content
|
| 31 |
+
article = (
|
| 32 |
+
soup.find("article") or
|
| 33 |
+
soup.find("div", class_=re.compile(r"article|post|content|story", re.I)) or
|
| 34 |
+
soup.find("main") or
|
| 35 |
+
soup.find("div", id=re.compile(r"content|article", re.I)) or
|
| 36 |
+
soup
|
| 37 |
+
)
|
| 38 |
+
|
| 39 |
+
# Extract paragraphs
|
| 40 |
+
paragraphs = article.find_all("p")
|
| 41 |
+
text = " ".join([
|
| 42 |
+
clean_text(p.get_text())
|
| 43 |
+
for p in paragraphs
|
| 44 |
+
if len(p.get_text().strip()) > 30 # skip tiny lines
|
| 45 |
+
])
|
| 46 |
+
return text[:1500] # keep it concise for LLM
|
| 47 |
+
|
| 48 |
+
except Exception as e:
|
| 49 |
+
return f"[Content extraction failed: {type(e).__name__}]"
|
| 50 |
+
|
| 51 |
+
def search_dementia_news(audience: str = "patient", max_results: int = 3) -> List[Dict]:
|
| 52 |
+
"""
|
| 53 |
+
Returns list of news items: {title, url, snippet}
|
| 54 |
+
Uses DuckDuckGo for ethical, free, no-API-key search.
|
| 55 |
+
"""
|
| 56 |
+
try:
|
| 57 |
+
# Tailored queries
|
| 58 |
+
if audience == "patient":
|
| 59 |
+
query = (
|
| 60 |
+
'"dementia" OR "Alzheimer\'s disease" '
|
| 61 |
+
'(treatment OR care OR support OR lifestyle OR symptoms OR prevention OR early warning) '
|
| 62 |
+
'site:alz.org OR site:mayoclinic.org OR site:nia.nih.gov OR site:webmd.com OR site:healthline.com'
|
| 63 |
+
)
|
| 64 |
+
else: # doctor
|
| 65 |
+
query = (
|
| 66 |
+
'"dementia" OR "Alzheimer\'s" '
|
| 67 |
+
'(clinical trial OR biomarker OR diagnosis OR guideline OR research OR drug approval) '
|
| 68 |
+
'site:nejm.org OR site:the-lancet.com OR site:fda.gov OR site:medscape.com OR site:alzheimersresearchuk.org'
|
| 69 |
+
)
|
| 70 |
+
|
| 71 |
+
results = []
|
| 72 |
+
with DDGS() as ddgs:
|
| 73 |
+
raw_results = list(ddgs.text(
|
| 74 |
+
query,
|
| 75 |
+
region="wt-wt",
|
| 76 |
+
safesearch="off",
|
| 77 |
+
timelimit="m", # past month
|
| 78 |
+
max_results=max_results + 2 # fetch a few extra for filtering
|
| 79 |
+
))
|
| 80 |
+
|
| 81 |
+
for r in raw_results:
|
| 82 |
+
url = r.get("href", "").strip()
|
| 83 |
+
# Skip non-http or social media
|
| 84 |
+
if not url.startswith("http") or any(domain in url for domain in ["facebook.com", "twitter.com", "youtube.com", "linkedin.com"]):
|
| 85 |
+
continue
|
| 86 |
+
title = clean_text(r.get("title", ""))
|
| 87 |
+
snippet = clean_text(r.get("body", ""))
|
| 88 |
+
if len(title) < 10 or len(snippet) < 20:
|
| 89 |
+
continue
|
| 90 |
+
results.append({"title": title, "url": url, "snippet": snippet})
|
| 91 |
+
if len(results) >= max_results:
|
| 92 |
+
break
|
| 93 |
+
|
| 94 |
+
time.sleep(0.4) # be respectful
|
| 95 |
+
|
| 96 |
+
return results
|
| 97 |
+
|
| 98 |
+
except Exception as e:
|
| 99 |
+
return [{"error": f"Search failed: {str(e)}"}]
|