Update app.py

app.py

@@ -105,33 +105,105 @@ from Gradio_UI import GradioUI
 #     except Exception as e:
 #         return f"Error searching medical datasets for '{arg1}': {str(e)}"
 
-
 @tool
 def my_custom_tool(arg1: str, arg2: int) -> str:
-    keyword = arg1.strip().lower()
-    limit = int(arg2)
-
-    medical_terms = ["skin", "brain", "lung", "breast", "cancer", "tumor", "xray", "ct", "mri", "ultrasound", "radiology"]
+    """
+    Search and retrieve publicly available medical datasets from Hugging Face based on any medical-related keyword.
 
-    if not any(term in keyword for term in medical_terms):
-        return f"No medical datasets found for '{arg1}'."
+    Args:
+        arg1: A keyword related to medical data (e.g., 'cancer', 'diabetes', 'CT scan', 'radiology', 'dermoscopy').
+        arg2: The maximum number of datasets to retrieve.
 
-    # ✅ safe API call with fallback
+    Returns:
+        A list of dataset names matching the search query, or a message stating that no datasets were found.
+        If network access is restricted, simulated dataset results are returned instead.
+    """
     try:
-        response = requests.get(
-            f"https://huggingface.co/api/datasets?search={keyword}&limit={limit}",
-            timeout=10
-        )
-        response.raise_for_status()
-        datasets = response.json()
-    except requests.exceptions.RequestException:
-        datasets = [{"id": f"example/{keyword}-dataset-{i+1}"} for i in range(limit)]
-
-    if not datasets:
-        return f"No datasets found for '{arg1}'."
-
-    results = [f"- {ds.get('id', 'Unknown')}" for ds in datasets[:limit]]
-    return f"Medical datasets related to '{arg1}':\n" + "\n".join(results)
+        keyword = arg1.strip().lower()
+        limit = int(arg2)
+
+        # Define a basic list of medically relevant terms
+        medical_terms = [
+            # Anatomy / Body Parts
+            "skin", "brain", "lung", "chest", "abdomen", "spine", "bone", "heart", "liver", "kidney",
+            "bladder", "stomach", "colon", "rectum", "esophagus", "pancreas", "breast", "ear", "eye", 
+            "retina", "tooth", "teeth", "tongue", "jaw", "neck", "wrist", "hand", "leg", "arm", "shoulder", "pelvis",
+        
+            # Diseases / Conditions
+            "cancer", "tumor", "stroke", "diabetes", "pneumonia", "covid", "asthma", "eczema", "melanoma",
+            "hypertension", "alzheimer", "parkinson", "arthritis", "scoliosis", "epilepsy", "glaucoma",
+            "ulcer", "hepatitis", "leukemia", "lymphoma", "tuberculosis", "anemia", "obesity", "depression",
+            "anxiety", "bipolar", "autism", "adhd", "ptsd", "psychosis", "schizophrenia",
+        
+            # Imaging Modalities
+            "mri", "ct", "xray", "x-ray", "ultrasound", "pet", "fmri", "mammo", "angiography", "radiography",
+            "echocardiogram", "spect", "dermoscopy", "colonoscopy", "endoscopy", "biopsy", "histopathology",
+        
+            # Medical Specialties
+            "radiology", "pathology", "oncology", "cardiology", "neurology", "dermatology", "dentistry",
+            "ophthalmology", "urology", "orthopedics", "gastroenterology", "pulmonology", "nephrology",
+            "psychiatry", "pediatrics", "geriatrics", "infectious disease",
+        
+            # Symptoms / Signs
+            "lesion", "infection", "fever", "pain", "inflammation", "rash", "headache", "swelling", 
+            "cough", "seizure", "dizziness", "vomiting", "diarrhea", "nausea", "fatigue", "itching",
+        
+            # Common Specific Diseases
+            "breast cancer", "prostate cancer", "lung cancer", "skin cancer", "colon cancer", 
+            "brain tumor", "liver cancer", "cervical cancer", "bladder cancer", "thyroid cancer",
+        
+            # Procedures / Interventions
+            "surgery", "chemotherapy", "radiation", "transplant", "dialysis", "intubation", "stenting",
+            "ventilation", "vaccination", "anesthesia", "rehabilitation", "prosthetics", "orthotics",
+        
+            # Lab Tests / Biomarkers
+            "blood test", "cbc", "glucose", "hemoglobin", "cholesterol", "biomarker", "urinalysis",
+            "pcr", "serology", "antibody", "antigen",
+        
+            # Clinical Settings / Roles
+            "icu", "hospital", "emergency", "clinical notes", "nursing", "physician", "patient",
+            "medical record", "electronic health record", "ehr", "vitals",
+        
+            # Age-based Terms
+            "pediatric", "neonatal", "infant", "child", "adolescent", "geriatrics", "elderly",
+        
+            # Epidemiology / Public Health
+            "epidemiology", "prevalence", "incidence", "mortality", "public health", "health disparity",
+            "risk factor", "social determinant",
+        
+            # Pharmacology / Medications
+            "drug", "medication", "pharmacology", "side effect", "adverse event", "dose", "tablet",
+            "vaccine", "clinical trial", "placebo"
+        ]
+
+        # Check if keyword is in known medical terms
+        if not any(term in keyword for term in medical_terms):
+            return f"No medical datasets found for '{arg1}'."
+
+        # Try fetching datasets from Hugging Face, fallback to mock data if blocked
+        try:
+            response = requests.get(
+                f"https://huggingface.co/api/datasets?search={keyword}&limit={limit}",
+                timeout=10
+            )
+            response.raise_for_status()
+            datasets = response.json()
+        except requests.exceptions.RequestException:
+            # Simulate results if network access is blocked
+            datasets = [{"id": f"example/{keyword}-dataset-{i+1}"} for i in range(limit)]
+
+        # Return message if no datasets found
+        if not datasets:
+            return f"No medical datasets found for '{arg1}'."
+
+        # Collect and return dataset names
+        results = [f"- {ds.get('id', 'Unknown')}" for ds in datasets[:limit]]
+        return f"Medical datasets related to '{arg1}':\n" + "\n".join(results)
+
+    except Exception as e:
+        return f"Error searching medical datasets for '{arg1}': {str(e)}"
+
+
 
 
 @tool