Spaces:
Sleeping
Sleeping
Add application file
Browse files
app.py
ADDED
@@ -0,0 +1,114 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from Bio import Entrez
|
3 |
+
from transformers import pipeline
|
4 |
+
import spacy
|
5 |
+
|
6 |
+
# ---------------------------- Configuration ----------------------------
|
7 |
+
ENTREZ_EMAIL = "[email protected]" # Replace with your email!
|
8 |
+
HUGGINGFACE_API_TOKEN = "HUGGINGFACE_API_TOKEN" # Replace with your Hugging Face API token!
|
9 |
+
SUMMARIZATION_MODEL = "facebook/bart-large-cnn" # Or try "google/pegasus-large"
|
10 |
+
SPACY_MODEL = "en_core_web_sm" # Small model for faster processing
|
11 |
+
# ---------------------------- Global Variables ----------------------------
|
12 |
+
|
13 |
+
summarizer = None # Initialized in the setup function
|
14 |
+
nlp = None # Initialized in the setup function
|
15 |
+
|
16 |
+
# ---------------------------- Tool Functions ----------------------------
|
17 |
+
|
18 |
+
def search_pubmed(query: str) -> list:
|
19 |
+
"""Searches PubMed and returns a list of article IDs."""
|
20 |
+
Entrez.email = ENTREZ_EMAIL
|
21 |
+
try:
|
22 |
+
handle = Entrez.esearch(db="pubmed", term=query, retmax="5") # Limit to 5 for demonstration
|
23 |
+
record = Entrez.read(handle)
|
24 |
+
handle.close()
|
25 |
+
return record["IdList"]
|
26 |
+
except Exception as e:
|
27 |
+
return [f"Error during PubMed search: {e}"]
|
28 |
+
|
29 |
+
def fetch_abstract(article_id: str) -> str:
|
30 |
+
"""Fetches the abstract for a given PubMed article ID."""
|
31 |
+
Entrez.email = ENTREZ_EMAIL
|
32 |
+
try:
|
33 |
+
handle = Entrez.efetch(db="pubmed", id=article_id, rettype="abstract", retmode="text")
|
34 |
+
abstract = handle.read()
|
35 |
+
handle.close()
|
36 |
+
return abstract
|
37 |
+
except Exception as e:
|
38 |
+
return f"Error fetching abstract for {article_id}: {e}"
|
39 |
+
|
40 |
+
def summarize_abstract(abstract: str) -> str:
|
41 |
+
"""Summarizes an abstract using a transformer model."""
|
42 |
+
global summarizer # Access the global summarizer
|
43 |
+
if summarizer is None:
|
44 |
+
return "Summarizer not initialized. Please reload the interface."
|
45 |
+
|
46 |
+
try:
|
47 |
+
summary = summarizer(abstract, max_length=130, min_length=30, do_sample=False)[0]['summary_text']
|
48 |
+
return summary
|
49 |
+
except Exception as e:
|
50 |
+
return f"Error during summarization: {e}"
|
51 |
+
|
52 |
+
def extract_entities(text: str) -> list:
|
53 |
+
"""Extracts entities (simplified) using SpaCy."""
|
54 |
+
global nlp
|
55 |
+
try:
|
56 |
+
doc = nlp(text)
|
57 |
+
entities = [(ent.text, ent.label_) for ent in doc.ents]
|
58 |
+
return entities
|
59 |
+
except Exception as e:
|
60 |
+
return [f"Error during entity extraction: {e}"]
|
61 |
+
|
62 |
+
|
63 |
+
# ---------------------------- Agent Function (Simplified) ----------------------------
|
64 |
+
|
65 |
+
def medai_agent(query: str) -> str:
|
66 |
+
"""Orchestrates the medical literature review and summarization."""
|
67 |
+
article_ids = search_pubmed(query)
|
68 |
+
|
69 |
+
if isinstance(article_ids, list) and article_ids:
|
70 |
+
results = []
|
71 |
+
for article_id in article_ids:
|
72 |
+
abstract = fetch_abstract(article_id)
|
73 |
+
if "Error" not in abstract:
|
74 |
+
summary = summarize_abstract(abstract)
|
75 |
+
entities = extract_entities(abstract) #extract_entities(abstract)
|
76 |
+
results.append(f"**Article ID:** {article_id}\n\n**Summary:** {summary}\n\n**Entities:** {entities}\n\n---\n")
|
77 |
+
else:
|
78 |
+
results.append(f"Error processing article {article_id}: {abstract}\n\n---\n")
|
79 |
+
return "\n".join(results)
|
80 |
+
else:
|
81 |
+
return f"No articles found or error occurred: {article_ids}"
|
82 |
+
|
83 |
+
# ---------------------------- Gradio Interface ----------------------------
|
84 |
+
|
85 |
+
def setup():
|
86 |
+
"""Initializes the summarization model and Spacy model."""
|
87 |
+
global summarizer, nlp
|
88 |
+
try:
|
89 |
+
summarizer = pipeline("summarization", model=SUMMARIZATION_MODEL, token=HUGGINGFACE_API_TOKEN)
|
90 |
+
nlp = spacy.load(SPACY_MODEL)
|
91 |
+
return "MedAI Agent initialized successfully!"
|
92 |
+
except Exception as e:
|
93 |
+
return f"Initialization error: {e}"
|
94 |
+
|
95 |
+
def launch_gradio():
|
96 |
+
"""Launches the Gradio interface."""
|
97 |
+
with gr.Blocks() as iface:
|
98 |
+
gr.Markdown("# MedAI: Medical Literature Review and Summarization")
|
99 |
+
status = gr.Textbox(value="Initializing...", interactive=False)
|
100 |
+
query_input = gr.Textbox(lines=3, placeholder="Enter your medical query (e.g., 'new treatments for diabetes')...")
|
101 |
+
submit_button = gr.Button("Submit")
|
102 |
+
output_results = gr.Markdown()
|
103 |
+
|
104 |
+
submit_button.click(medai_agent, inputs=query_input, outputs=output_results)
|
105 |
+
|
106 |
+
# Initialization setup
|
107 |
+
status.value = setup()
|
108 |
+
|
109 |
+
iface.launch()
|
110 |
+
|
111 |
+
# ---------------------------- Main Execution ----------------------------
|
112 |
+
|
113 |
+
if __name__ == "__main__":
|
114 |
+
launch_gradio()
|