Create app.py

app.py

@@ -0,0 +1,80 @@
+# 📦 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()