Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -116,71 +116,26 @@ def my_custom_tool(arg1: str, arg2: int) -> str:
|
|
| 116 |
|
| 117 |
Returns:
|
| 118 |
A list of dataset names matching the search query, or a message stating that no datasets were found.
|
| 119 |
-
If network access is restricted, simulated dataset results are returned instead.
|
| 120 |
"""
|
| 121 |
try:
|
| 122 |
keyword = arg1.strip().lower()
|
| 123 |
limit = int(arg2)
|
| 124 |
|
| 125 |
-
# Define a
|
| 126 |
medical_terms = [
|
| 127 |
-
# Anatomy / Body Parts
|
| 128 |
"skin", "brain", "lung", "chest", "abdomen", "spine", "bone", "heart", "liver", "kidney",
|
| 129 |
-
"bladder", "stomach", "colon", "rectum", "esophagus", "pancreas", "breast", "ear", "eye",
|
| 130 |
-
"retina", "tooth", "teeth", "tongue", "jaw", "neck", "wrist", "hand", "leg", "arm", "shoulder", "pelvis",
|
| 131 |
-
|
| 132 |
-
# Diseases / Conditions
|
| 133 |
-
"cancer", "tumor", "stroke", "diabetes", "pneumonia", "covid", "asthma", "eczema", "melanoma",
|
| 134 |
-
"hypertension", "alzheimer", "parkinson", "arthritis", "scoliosis", "epilepsy", "glaucoma",
|
| 135 |
-
"ulcer", "hepatitis", "leukemia", "lymphoma", "tuberculosis", "anemia", "obesity", "depression",
|
| 136 |
-
"anxiety", "bipolar", "autism", "adhd", "ptsd", "psychosis", "schizophrenia",
|
| 137 |
-
|
| 138 |
-
# Imaging Modalities
|
| 139 |
-
"mri", "ct", "xray", "x-ray", "ultrasound", "pet", "fmri", "mammo", "angiography", "radiography",
|
| 140 |
-
"echocardiogram", "spect", "dermoscopy", "colonoscopy", "endoscopy", "biopsy", "histopathology",
|
| 141 |
-
|
| 142 |
-
# Medical Specialties
|
| 143 |
"radiology", "pathology", "oncology", "cardiology", "neurology", "dermatology", "dentistry",
|
| 144 |
"ophthalmology", "urology", "orthopedics", "gastroenterology", "pulmonology", "nephrology",
|
| 145 |
"psychiatry", "pediatrics", "geriatrics", "infectious disease",
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
"lesion", "infection", "fever", "pain", "inflammation", "rash", "headache", "swelling",
|
| 149 |
-
"cough", "seizure", "dizziness", "vomiting", "diarrhea", "nausea", "fatigue", "itching",
|
| 150 |
-
|
| 151 |
-
# Common Specific Diseases
|
| 152 |
-
"breast cancer", "prostate cancer", "lung cancer", "skin cancer", "colon cancer",
|
| 153 |
-
"brain tumor", "liver cancer", "cervical cancer", "bladder cancer", "thyroid cancer",
|
| 154 |
-
|
| 155 |
-
# Procedures / Interventions
|
| 156 |
-
"surgery", "chemotherapy", "radiation", "transplant", "dialysis", "intubation", "stenting",
|
| 157 |
-
"ventilation", "vaccination", "anesthesia", "rehabilitation", "prosthetics", "orthotics",
|
| 158 |
-
|
| 159 |
-
# Lab Tests / Biomarkers
|
| 160 |
-
"blood test", "cbc", "glucose", "hemoglobin", "cholesterol", "biomarker", "urinalysis",
|
| 161 |
-
"pcr", "serology", "antibody", "antigen",
|
| 162 |
-
|
| 163 |
-
# Clinical Settings / Roles
|
| 164 |
-
"icu", "hospital", "emergency", "clinical notes", "nursing", "physician", "patient",
|
| 165 |
-
"medical record", "electronic health record", "ehr", "vitals",
|
| 166 |
-
|
| 167 |
-
# Age-based Terms
|
| 168 |
-
"pediatric", "neonatal", "infant", "child", "adolescent", "geriatrics", "elderly",
|
| 169 |
-
|
| 170 |
-
# Epidemiology / Public Health
|
| 171 |
-
"epidemiology", "prevalence", "incidence", "mortality", "public health", "health disparity",
|
| 172 |
-
"risk factor", "social determinant",
|
| 173 |
-
|
| 174 |
-
# Pharmacology / Medications
|
| 175 |
-
"drug", "medication", "pharmacology", "side effect", "adverse event", "dose", "tablet",
|
| 176 |
-
"vaccine", "clinical trial", "placebo"
|
| 177 |
]
|
| 178 |
|
| 179 |
-
# Check if keyword is in known medical terms
|
| 180 |
if not any(term in keyword for term in medical_terms):
|
| 181 |
-
return f"No medical datasets found for '{arg1}'."
|
| 182 |
|
| 183 |
-
# Try
|
| 184 |
try:
|
| 185 |
response = requests.get(
|
| 186 |
f"https://huggingface.co/api/datasets?search={keyword}&limit={limit}",
|
|
@@ -188,15 +143,14 @@ def my_custom_tool(arg1: str, arg2: int) -> str:
|
|
| 188 |
)
|
| 189 |
response.raise_for_status()
|
| 190 |
datasets = response.json()
|
| 191 |
-
except
|
| 192 |
-
#
|
| 193 |
datasets = [{"id": f"example/{keyword}-dataset-{i+1}"} for i in range(limit)]
|
| 194 |
|
| 195 |
-
# Return
|
| 196 |
if not datasets:
|
| 197 |
-
return f"No
|
| 198 |
|
| 199 |
-
# Collect and return dataset names
|
| 200 |
results = [f"- {ds.get('id', 'Unknown')}" for ds in datasets[:limit]]
|
| 201 |
return f"Medical datasets related to '{arg1}':\n" + "\n".join(results)
|
| 202 |
|
|
@@ -206,6 +160,7 @@ def my_custom_tool(arg1: str, arg2: int) -> str:
|
|
| 206 |
|
| 207 |
|
| 208 |
|
|
|
|
| 209 |
@tool
|
| 210 |
def get_current_time_in_timezone(timezone: str) -> str:
|
| 211 |
"""
|
|
@@ -250,15 +205,30 @@ model = InferenceClientModel(
|
|
| 250 |
with open("prompts.yaml", 'r') as stream:
|
| 251 |
prompt_templates = yaml.safe_load(stream)
|
| 252 |
|
| 253 |
-
# Create the agent
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 254 |
agent = CodeAgent(
|
| 255 |
model=model,
|
| 256 |
tools=[final_answer, get_current_time_in_timezone, my_custom_tool],
|
| 257 |
max_steps=6,
|
| 258 |
-
verbosity_level=
|
| 259 |
planning_interval=None,
|
| 260 |
-
name=
|
| 261 |
-
description=
|
|
|
|
|
|
|
|
|
|
|
|
|
| 262 |
prompt_templates=prompt_templates
|
| 263 |
)
|
| 264 |
|
|
|
|
| 116 |
|
| 117 |
Returns:
|
| 118 |
A list of dataset names matching the search query, or a message stating that no datasets were found.
|
|
|
|
| 119 |
"""
|
| 120 |
try:
|
| 121 |
keyword = arg1.strip().lower()
|
| 122 |
limit = int(arg2)
|
| 123 |
|
| 124 |
+
# Define a list of medical terms
|
| 125 |
medical_terms = [
|
|
|
|
| 126 |
"skin", "brain", "lung", "chest", "abdomen", "spine", "bone", "heart", "liver", "kidney",
|
| 127 |
+
"bladder", "stomach", "colon", "rectum", "esophagus", "pancreas", "breast", "ear", "eye",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 128 |
"radiology", "pathology", "oncology", "cardiology", "neurology", "dermatology", "dentistry",
|
| 129 |
"ophthalmology", "urology", "orthopedics", "gastroenterology", "pulmonology", "nephrology",
|
| 130 |
"psychiatry", "pediatrics", "geriatrics", "infectious disease",
|
| 131 |
+
"mri", "ct", "xray", "x-ray", "ultrasound", "pet", "fmri", "mammo", "angiography", "radiography",
|
| 132 |
+
"cancer", "tumor", "stroke", "diabetes", "melanoma", "eczema", "asthma", "thyroid"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
]
|
| 134 |
|
|
|
|
| 135 |
if not any(term in keyword for term in medical_terms):
|
| 136 |
+
return f"No medical datasets found for '{arg1}'. Please try another medical term."
|
| 137 |
|
| 138 |
+
# Try online query to Hugging Face
|
| 139 |
try:
|
| 140 |
response = requests.get(
|
| 141 |
f"https://huggingface.co/api/datasets?search={keyword}&limit={limit}",
|
|
|
|
| 143 |
)
|
| 144 |
response.raise_for_status()
|
| 145 |
datasets = response.json()
|
| 146 |
+
except Exception:
|
| 147 |
+
# Network-restricted fallback
|
| 148 |
datasets = [{"id": f"example/{keyword}-dataset-{i+1}"} for i in range(limit)]
|
| 149 |
|
| 150 |
+
# Return formatted list
|
| 151 |
if not datasets:
|
| 152 |
+
return f"No datasets found for '{arg1}'."
|
| 153 |
|
|
|
|
| 154 |
results = [f"- {ds.get('id', 'Unknown')}" for ds in datasets[:limit]]
|
| 155 |
return f"Medical datasets related to '{arg1}':\n" + "\n".join(results)
|
| 156 |
|
|
|
|
| 160 |
|
| 161 |
|
| 162 |
|
| 163 |
+
|
| 164 |
@tool
|
| 165 |
def get_current_time_in_timezone(timezone: str) -> str:
|
| 166 |
"""
|
|
|
|
| 205 |
with open("prompts.yaml", 'r') as stream:
|
| 206 |
prompt_templates = yaml.safe_load(stream)
|
| 207 |
|
| 208 |
+
# # Create the agent
|
| 209 |
+
# agent = CodeAgent(
|
| 210 |
+
# model=model,
|
| 211 |
+
# tools=[final_answer, get_current_time_in_timezone, my_custom_tool],
|
| 212 |
+
# max_steps=6,
|
| 213 |
+
# verbosity_level=2,
|
| 214 |
+
# planning_interval=None,
|
| 215 |
+
# name=None,
|
| 216 |
+
# description=None,
|
| 217 |
+
# prompt_templates=prompt_templates
|
| 218 |
+
# )
|
| 219 |
+
|
| 220 |
agent = CodeAgent(
|
| 221 |
model=model,
|
| 222 |
tools=[final_answer, get_current_time_in_timezone, my_custom_tool],
|
| 223 |
max_steps=6,
|
| 224 |
+
verbosity_level=1,
|
| 225 |
planning_interval=None,
|
| 226 |
+
name="MedDataSearchAgent",
|
| 227 |
+
description=(
|
| 228 |
+
"An intelligent agent that searches Hugging Face datasets related to "
|
| 229 |
+
"medical conditions, body parts, and imaging modalities. "
|
| 230 |
+
"Use 'my_custom_tool' whenever the user requests medical data or datasets."
|
| 231 |
+
),
|
| 232 |
prompt_templates=prompt_templates
|
| 233 |
)
|
| 234 |
|