Spaces:
Runtime error
Runtime error
| # 📦 Installations needed locally before deploying: | |
| # Linux: sudo apt install tesseract-ocr poppler-utils | |
| # Windows: Install Tesseract from https://github.com/tesseract-ocr/tesseract | |
| import gradio as gr | |
| from pdf2image import convert_from_path | |
| from PIL import Image | |
| import pytesseract | |
| import torch | |
| from transformers import AutoTokenizer, AutoModelForCausalLM | |
| # Load MedAlpaca Model from Hugging Face | |
| model_name = "medalpaca/medalpaca-7b" | |
| tokenizer = AutoTokenizer.from_pretrained(model_name) | |
| model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32) | |
| device = "cuda" if torch.cuda.is_available() else "cpu" | |
| model.to(device) | |
| # ========== OCR FUNCTIONS ========== | |
| def extract_text_from_image(image): | |
| return pytesseract.image_to_string(image) | |
| def extract_text_from_pdf(pdf_file): | |
| try: | |
| images = convert_from_path(pdf_file.name) | |
| text = "" | |
| for page in images: | |
| text += pytesseract.image_to_string(page) + "\n" | |
| return text | |
| except Exception as e: | |
| return f"Error reading PDF: {e}" | |
| # ========== MEDALPACA RESPONSE ========== | |
| def generate_medical_explanation(text): | |
| prompt = ( | |
| "You are a helpful medical assistant. Analyze the following patient's lab report text " | |
| "and explain the abnormalities in plain, non-technical language:\n\n" + text + | |
| "\n\nAlso, highlight abnormal values with flags." | |
| ) | |
| inputs = tokenizer(prompt, return_tensors="pt").to(device) | |
| outputs = model.generate(**inputs, max_new_tokens=512, do_sample=True, temperature=0.7) | |
| result = tokenizer.decode(outputs[0], skip_special_tokens=True) | |
| return result.split(prompt)[-1].strip() | |
| # ========== MAIN APP FUNCTION ========== | |
| def analyze_file(file): | |
| if not file: | |
| return "⚠️ No file uploaded.", "" | |
| filename = file.name.lower() | |
| if filename.endswith(".pdf"): | |
| extracted_text = extract_text_from_pdf(file) | |
| else: | |
| try: | |
| img = Image.open(file.name) | |
| extracted_text = extract_text_from_image(img) | |
| except Exception as e: | |
| return f"❌ Error loading image: {e}", "" | |
| if not extracted_text.strip(): | |
| return "❌ No text found. Try uploading a clearer image or PDF.", "" | |
| ai_response = generate_medical_explanation(extracted_text) | |
| return extracted_text, ai_response | |
| # ========== GRADIO INTERFACE ========== | |
| gr.Interface( | |
| fn=analyze_file, | |
| inputs=gr.File(label="📄 Upload Lab Report (Image or PDF)"), | |
| outputs=[ | |
| gr.Textbox(label="📜 Extracted Text", lines=20), | |
| gr.Textbox(label="🧠 MedAlpaca Interpretation", lines=20) | |
| ], | |
| title="🔬 AI Lab Report Analyzer with MedAlpaca", | |
| description="Upload your medical report (image or PDF). This app extracts text using OCR and explains lab values using the MedAlpaca model." | |
| ).launch() | |